474. Podcast Storytelling Secrets from the ‘Search Engine’ Podcast

1
00:00:05,680 --> 00:00:07,960
Good morning podcasting morning
chat.

2
00:00:08,200 --> 00:00:14,320
Today is Monday, March 16th,
2026 and today we're back with

3
00:00:14,320 --> 00:00:18,760
obsession worthy podcast sharing
another can't miss listen.

4
00:00:19,120 --> 00:00:21,720
Today we're breaking down search
engine and why it's curious

5
00:00:21,720 --> 00:00:24,920
questions and smart storytelling
keeps listeners hooked.

6
00:00:26,280 --> 00:00:28,440
Happy Monday.
So if you're listening live on

7
00:00:28,440 --> 00:00:30,400
Clubhouse, hit the share button,
bottom left hand, side of the

8
00:00:30,400 --> 00:00:32,720
screen and share it.
However, Clubhouse lets you.

9
00:00:32,720 --> 00:00:35,440
And if you're catching us via
podcast, YouTube, etcetera,

10
00:00:35,760 --> 00:00:37,600
please share this with a fellow
podcaster.

11
00:00:39,640 --> 00:00:43,600
And now give us about 30 seconds
and we'll get things rolling.

12
00:00:43,960 --> 00:01:23,000
Thanks for being here.
Good morning again, podcasting

13
00:01:23,000 --> 00:01:24,800
morning chat.
Thanks so much for being here.

14
00:01:24,800 --> 00:01:29,040
I am your host Mark Roddick and
currently on stage with me We

15
00:01:29,040 --> 00:01:33,680
have Jonathan Howard, Ralph
Estep, Dr. Faye, Alex Baelish,

16
00:01:33,680 --> 00:01:38,400
Sid Meadows, BC Babbles, Nick
Naal back and also Tide AKA

17
00:01:38,400 --> 00:01:41,400
Nikki is on stage with us from
the audience.

18
00:01:41,400 --> 00:01:44,920
Good to see all of you here.
If you want to learn more about

19
00:01:44,920 --> 00:01:49,480
our Co host, just go to
podcastingmorningchat.com/people

20
00:01:49,680 --> 00:01:52,840
and you can learn about all the
different people I just

21
00:01:52,840 --> 00:01:56,840
mentioned to you.
And yes, it is Monday and we are

22
00:01:56,840 --> 00:02:00,400
going to get into another
obsession worthy podcast

23
00:02:00,400 --> 00:02:04,760
courtesy of Jonathan Howard.
And before we do, I wanted to

24
00:02:04,760 --> 00:02:07,000
just share.
Stay tuned this week,

25
00:02:07,440 --> 00:02:10,240
particularly Wednesday, 7:00 AM
Eastern.

26
00:02:10,240 --> 00:02:15,400
I've had some things develop
Friday and into the weekend that

27
00:02:15,400 --> 00:02:19,440
will lead to what looks like
will be a big announcement this

28
00:02:19,440 --> 00:02:23,280
Wednesday regarding the show.
So stay tuned for that.

29
00:02:23,280 --> 00:02:28,160
That's all I will say for now.
And also another programming

30
00:02:28,160 --> 00:02:31,560
note.
This Thursday we're going to do

31
00:02:31,560 --> 00:02:36,560
our podcast, Thon Episode
Podcast. thon.org is the

32
00:02:36,560 --> 00:02:40,080
website.
This is the yearly podcaster

33
00:02:40,080 --> 00:02:42,400
event.
We're all different.

34
00:02:42,400 --> 00:02:46,600
Podcasts all over the world
quite literally are stepping

35
00:02:46,600 --> 00:02:50,840
back from their usual formats in
most cases and talking about

36
00:02:50,840 --> 00:02:54,200
their favorite charity,
nonprofit organization,

37
00:02:54,840 --> 00:02:58,200
something that they support and
want to share with their

38
00:02:58,200 --> 00:03:01,440
audience and hope that maybe
their audience will also show

39
00:03:01,440 --> 00:03:05,120
some support and love for that.
I'll, again, I'll call it

40
00:03:05,120 --> 00:03:07,600
organization, but it could be a
variety of things.

41
00:03:07,840 --> 00:03:13,360
Thursday, the way I think I'd
like to do Podcast Thon is for

42
00:03:13,360 --> 00:03:19,040
all of us listeners, the team,
etcetera, to bring your charity,

43
00:03:19,040 --> 00:03:23,120
your organization, something
that you support and come share

44
00:03:23,120 --> 00:03:26,040
about it and tell us what's the
why behind that support.

45
00:03:26,040 --> 00:03:30,040
Usually there is some reason
behind supporting these types of

46
00:03:30,040 --> 00:03:32,880
charities.
The one that I want to support

47
00:03:33,040 --> 00:03:37,720
is the CMTA, the Shark O Marie
Tooth Association.

48
00:03:37,960 --> 00:03:42,520
That's the organization that my
best friend has the disease CMT,

49
00:03:43,040 --> 00:03:47,840
and I've grown up with him with
it and have really gotten to

50
00:03:47,840 --> 00:03:51,400
know what it's like, at least
from the outsider perspective.

51
00:03:51,400 --> 00:03:54,480
I've gotten to know what it's
like living with that particular

52
00:03:54,480 --> 00:03:58,760
disease.
I have been a well, I just last

53
00:03:58,760 --> 00:04:03,400
year was a counselor at a one
week summer camp for kids with

54
00:04:03,400 --> 00:04:08,480
that disability and it was my
way of volunteering, being

55
00:04:08,480 --> 00:04:11,000
showing my support of my best
friend Jonah.

56
00:04:11,200 --> 00:04:14,720
And this year I think I might be
doing it two different weeks

57
00:04:14,880 --> 00:04:19,640
this summer, one in the West
Coast area, 1 here on the East

58
00:04:19,640 --> 00:04:20,680
Coast.
Yeah.

59
00:04:20,680 --> 00:04:24,720
So I may also do a special
episode with my friend Jonah to

60
00:04:25,280 --> 00:04:27,520
teach a little bit more about
it.

61
00:04:27,560 --> 00:04:31,840
And and so basically go to
podcast on dot org if you want

62
00:04:31,840 --> 00:04:35,880
to learn more about this whole
movement in the podcast

63
00:04:35,880 --> 00:04:37,880
industry.
And again, bring your

64
00:04:37,880 --> 00:04:42,920
organization, the thing you want
to support on Thursday and let's

65
00:04:43,040 --> 00:04:45,400
talk about it.
And we'll actually even we're, I

66
00:04:45,400 --> 00:04:47,000
was talking with some of the
team.

67
00:04:47,280 --> 00:04:50,240
We'll, we're thinking about,
we'll even make a page dedicated

68
00:04:50,240 --> 00:04:54,360
to the different organizations
that everybody shares with us on

69
00:04:54,360 --> 00:04:56,680
Thursday.
So that's Wednesday and Thursday

70
00:04:56,680 --> 00:04:59,560
we've got more stuff Tuesday, I
believe we'll be doing the

71
00:04:59,560 --> 00:05:02,480
mailbag as well.
So answering your podcasting

72
00:05:02,480 --> 00:05:06,680
content creation questions.
And I am out of breath.

73
00:05:08,480 --> 00:05:13,000
That was a quite a long intro
and thank you for sticking

74
00:05:13,000 --> 00:05:17,360
around for it.
Now let's go to Jonathan Howard

75
00:05:17,360 --> 00:05:19,720
and Jonathan, the way we like to
open up

76
00:05:19,760 --> 00:05:29,440
obsessionworthypodcastspodcastingmorningchat.com/opw.
We like to learn what is

77
00:05:29,600 --> 00:05:32,240
obsession worthy podcasts.
What is this series all about

78
00:05:32,240 --> 00:05:35,440
Jonathan and then lead us into
the conversation today, please.

79
00:05:35,440 --> 00:05:37,520
Good morning.
No problem and I think it would

80
00:05:37,520 --> 00:05:40,760
be OWP.
Yeah, I was getting it wrong

81
00:05:40,760 --> 00:05:42,120
last week too.
I was.

82
00:05:42,440 --> 00:05:44,800
Saying.
It super slow and just screwing

83
00:05:44,800 --> 00:05:46,600
it up.
Yeah, I don't know why I have a

84
00:05:46,600 --> 00:05:50,440
block on that.
I it's OK, I would happen to me

85
00:05:50,440 --> 00:05:55,480
too, but the obsession podcast
worthy obsession worthy podcast.

86
00:05:55,480 --> 00:05:59,080
Thanks, Mark.
Yeah, let we're not editing that

87
00:05:59,080 --> 00:06:01,160
out because that is just on
point.

88
00:06:02,160 --> 00:06:07,080
It's just what I would do.
So obsession worthy podcast are

89
00:06:07,120 --> 00:06:11,920
all about what we can learn from
pop well and doing things that

90
00:06:11,920 --> 00:06:15,840
are either interesting or
creative or showing up in a way

91
00:06:15,840 --> 00:06:20,080
that we can learn from.
And this episode of search

92
00:06:20,080 --> 00:06:22,720
engine is no no exception to
that.

93
00:06:23,040 --> 00:06:26,680
Does the interesting things.
I'll lead us through those as as

94
00:06:26,680 --> 00:06:29,240
we go through some of the clips
today.

95
00:06:29,640 --> 00:06:32,560
Let me jump in.
So what search engine does is it

96
00:06:32,560 --> 00:06:36,160
actually investigate strange
questions around the Internet in

97
00:06:36,160 --> 00:06:38,640
modern life.
The promise of the show is like

98
00:06:38,760 --> 00:06:43,000
they're going to research and
find answers to a question, but

99
00:06:43,120 --> 00:06:47,400
a lot of times it ends up being
bigger than just the question

100
00:06:47,400 --> 00:06:49,840
itself.
And that was definitely the case

101
00:06:49,840 --> 00:06:55,040
in this particular episode.
This episode asks, why does AI

102
00:06:55,040 --> 00:06:57,880
Claude sometimes behave
strangely?

103
00:06:58,560 --> 00:07:01,160
But in the end, we end up
answering a much deeper

104
00:07:01,160 --> 00:07:03,360
question.
So we'll get to that question as

105
00:07:03,360 --> 00:07:07,800
we go, but you'll notice that it
continuously expands as we go

106
00:07:07,800 --> 00:07:09,400
through the episode and through
the clips.

107
00:07:09,480 --> 00:07:13,480
So I mean, right out of the
gate, what I can say that might

108
00:07:13,480 --> 00:07:18,320
make this show obsession worthy,
at least out of the gate, The

109
00:07:18,320 --> 00:07:20,600
topics, right?
Like they're finding really

110
00:07:20,600 --> 00:07:24,080
intriguing topics, and this one
being no exception.

111
00:07:24,080 --> 00:07:27,680
That whole story that I know we
talked about at least once on

112
00:07:27,680 --> 00:07:32,440
this show when it was happening
was this idea that Claude was

113
00:07:32,760 --> 00:07:37,400
defying the human that was
running Claude and Claude, much

114
00:07:37,400 --> 00:07:41,280
like ChatGPT, for those of you
who may not know it, same idea

115
00:07:41,280 --> 00:07:44,560
as a ChatGPT.
Some prefer it, but yeah, OK, go

116
00:07:44,560 --> 00:07:46,160
ahead, Jonathan.
I just found that that's a

117
00:07:46,240 --> 00:07:49,680
worthwhile thing to point out is
yeah, they're finding topics

118
00:07:49,680 --> 00:07:52,680
that's really interesting for
their audience.

119
00:07:53,640 --> 00:07:55,440
Yeah, yeah.
And and the other thing is it

120
00:07:55,440 --> 00:07:58,760
doesn't explain things.
It really investigates them and

121
00:07:58,960 --> 00:08:02,720
that creates curiosity, which
builds into the story.

122
00:08:03,080 --> 00:08:05,920
So instead of saying, hey,
here's how AI works, that's not

123
00:08:05,920 --> 00:08:08,680
what the episode was about.
It was more something strange

124
00:08:08,680 --> 00:08:10,320
happening.
Let's figure that out.

125
00:08:10,760 --> 00:08:13,160
And that's where we kind of
walked through it because

126
00:08:13,360 --> 00:08:16,760
mystery is one of the most
addictive storytelling engines.

127
00:08:16,760 --> 00:08:20,360
So when you have mystery, that's
a great way to keep people

128
00:08:20,760 --> 00:08:22,440
really invested in what you're
doing.

129
00:08:23,240 --> 00:08:29,520
So this first clip here, the
host PJ Voigt, introduces a

130
00:08:29,520 --> 00:08:33,760
feeling that he calls future and
all, And he's describing an

131
00:08:33,840 --> 00:08:37,120
emotional experience of watching
AI get better, realizing it may

132
00:08:37,120 --> 00:08:39,640
actually disrupt his own field.
Here we go.

133
00:08:44,440 --> 00:08:47,560
I found myself feeling much
stranger about AI in the past

134
00:08:47,560 --> 00:08:49,680
month or so.
I use the tools.

135
00:08:49,680 --> 00:08:52,320
I use the tools a lot, but I'm
probably each company's worst

136
00:08:52,320 --> 00:08:55,760
nightmare as a customer in that
as soon as I hear from anybody

137
00:08:55,760 --> 00:08:59,160
that one model has inched ahead
of another that this version of

138
00:08:59,160 --> 00:09:02,720
ChatGPT is beating that version
of Gemini, I immediately cancel

139
00:09:02,720 --> 00:09:06,000
my subscription and switch.
For the past two months, I've

140
00:09:06,000 --> 00:09:08,640
mainly been using Quad
Anthropics Agent, and for

141
00:09:08,640 --> 00:09:11,760
whatever reason, Quad is just
giving me more future nausea

142
00:09:11,760 --> 00:09:13,040
than I was having six months
ago.

143
00:09:14,240 --> 00:09:16,920
Part of the general tech
excitement around Claude lately

144
00:09:16,920 --> 00:09:20,000
has been Anthropic's product,
Claude Code, a tool that lets

145
00:09:20,000 --> 00:09:22,800
the AI agent autonomously write
and edit code.

146
00:09:23,480 --> 00:09:26,120
Over the New York Times, Kevin
Roose has talked a lot about the

147
00:09:26,120 --> 00:09:30,000
websites and apps he's quickly
built with Claude Code to CNBC

148
00:09:30,000 --> 00:09:32,440
reporters.
As an experiment, Vibe coded a

149
00:09:32,440 --> 00:09:35,240
competing version of a popular
organizational app called

150
00:09:35,280 --> 00:09:37,880
monday.com.
Within a couple of days,

151
00:09:37,920 --> 00:09:42,720
Monday's stock price had tanked.
For me, though, most of the

152
00:09:42,720 --> 00:09:45,520
future shock has just come from
using the LMS the way I'm used

153
00:09:45,520 --> 00:09:48,080
to.
I found myself going to Cod as a

154
00:09:48,120 --> 00:09:50,400
useful first stop, but I've
always used the Internet.

155
00:09:50,800 --> 00:09:54,360
But the quality of its research,
its answers, even its writing,

156
00:09:54,640 --> 00:09:58,120
I'm just trying to feel like I
can see not too far off, if not

157
00:09:58,120 --> 00:10:01,600
my own obsolescence, at least
real significant change in my

158
00:10:01,600 --> 00:10:03,920
field.
I don't know how to feel about

159
00:10:03,920 --> 00:10:06,800
that.
Now, I don't know about you or I

160
00:10:06,800 --> 00:10:09,000
don't know, Jonathan 'cause I
haven't listened to other

161
00:10:09,000 --> 00:10:12,600
episodes, but maybe, you know,
is that the music behind them?

162
00:10:12,600 --> 00:10:15,760
Is that the same music they
always played for the intro?

163
00:10:15,760 --> 00:10:19,960
Or is that specific to this?
Because it really built up a lot

164
00:10:19,960 --> 00:10:22,360
of tension, which I can really
appreciate.

165
00:10:23,360 --> 00:10:27,800
Yeah, I've listened to a few
this season and that is, I'm not

166
00:10:27,800 --> 00:10:30,480
going to say it's the exact
music, but it's the similar feel

167
00:10:30,480 --> 00:10:32,000
that you have for all the.
Episodes.

168
00:10:32,000 --> 00:10:34,520
OK, yeah, I love the tension
that it builds.

169
00:10:34,560 --> 00:10:38,320
Yeah, and it does build tension.
And what this, you know, opener

170
00:10:38,320 --> 00:10:41,640
really does is most shows would
start out with like large

171
00:10:41,640 --> 00:10:44,160
language models are training you
on massive data sets, blah,

172
00:10:44,160 --> 00:10:46,680
blah, blah.
Like, congratulations, that's

173
00:10:46,680 --> 00:10:48,600
boring.
Half your audience yawned and

174
00:10:48,600 --> 00:10:51,840
the other half left.
What the search engine does is

175
00:10:53,120 --> 00:10:56,840
completely different.
They start with a feeling, they

176
00:10:56,840 --> 00:11:00,160
start with that future nausea
that so you're kind of

177
00:11:00,440 --> 00:11:02,760
wondering, wait, what is this
future nausea even?

178
00:11:02,840 --> 00:11:07,720
But also it's that uneasy
sensation that we all get when

179
00:11:07,720 --> 00:11:11,120
something massive is coming,
like when technology is going to

180
00:11:11,120 --> 00:11:14,480
appear and it may take our job,
it's going to change everything.

181
00:11:14,840 --> 00:11:18,760
So he's feeling that and that's
a gut feeling that we can all

182
00:11:18,760 --> 00:11:22,280
kind of connect with.
And that's what brings us into

183
00:11:22,360 --> 00:11:25,680
the actual episode.
Once you get him in, they can

184
00:11:25,680 --> 00:11:27,560
deal with complicated stuff.
But if you start with the

185
00:11:27,560 --> 00:11:29,960
complicated stuff, not going to
get him in that door.

186
00:11:29,960 --> 00:11:34,600
So the doorway is truly the
emotional feeling of what it's

187
00:11:34,600 --> 00:11:36,520
like this, is this going to
replace me?

188
00:11:37,080 --> 00:11:38,320
Am I going to have a problem
with that?

189
00:11:38,720 --> 00:11:41,360
Yeah, Ralph, I think has a
question, Jonathan or something

190
00:11:41,360 --> 00:11:42,600
he wants to add.
Go for it, Ralph.

191
00:11:42,600 --> 00:11:43,560
Good morning.
Yeah.

192
00:11:43,560 --> 00:11:45,520
Good morning, everybody.
Yeah, Jonathan, I think that's

193
00:11:45,520 --> 00:11:47,360
brilliant.
And I know you appreciate what

194
00:11:47,360 --> 00:11:50,200
I'm getting ready to say, having
dogs, it's kind of like when you

195
00:11:50,200 --> 00:11:53,280
got to get your dog coaxed into
taking a pill or something like

196
00:11:53,280 --> 00:11:54,440
that.
You don't want to lay the pill

197
00:11:54,440 --> 00:11:57,080
out right in front of them.
You give them a little tree, you

198
00:11:57,080 --> 00:11:59,240
kind of walk them down the road.
Oh, you're really going to like

199
00:11:59,240 --> 00:12:00,040
this.
You're really going to like

200
00:12:00,040 --> 00:12:01,600
this.
And I think, Jonathan, that's a

201
00:12:01,600 --> 00:12:03,960
brilliant observation here.
That's what they do so well.

202
00:12:04,240 --> 00:12:06,400
The only other thing I was going
to add is the cynic in the room.

203
00:12:06,600 --> 00:12:09,600
I knew it was an obsession
worthy podcast because it

204
00:12:09,600 --> 00:12:11,680
started off with a block of
commercials, but I'll let that.

205
00:12:13,640 --> 00:12:14,920
Very true.
It did.

206
00:12:14,960 --> 00:12:18,200
And there were multiple
commercial blocks through this

207
00:12:18,440 --> 00:12:21,640
episode for sure.
But I'll tell you what actually,

208
00:12:21,640 --> 00:12:25,080
and let me say one thing I
appreciated is as far as I could

209
00:12:25,080 --> 00:12:28,800
tell, as far as I can remember,
every spot, every commercial

210
00:12:28,800 --> 00:12:34,680
that he did, it was read by him.
He did the commercial, the host

211
00:12:34,680 --> 00:12:37,760
did the commercial, which I
think is going to be way more

212
00:12:37,760 --> 00:12:42,200
effective than if it's just some
voice over person or what have

213
00:12:42,200 --> 00:12:44,440
you, But you know it.
It feels a little more

214
00:12:45,200 --> 00:12:48,600
trustworthy, I think, for an
audience if it's coming directly

215
00:12:48,600 --> 00:12:52,520
from the podcast host.
Yeah, I see BC as his hand up.

216
00:12:52,920 --> 00:12:56,160
Go ahead BC.
I was going to say also speaking

217
00:12:56,160 --> 00:12:58,960
to that trustworthiness, I do
appreciate at least during the

218
00:12:58,960 --> 00:13:03,120
first commercial block there
that they reminded the audience,

219
00:13:03,560 --> 00:13:07,720
by the way, the brand that
sponsor the shows do not get a

220
00:13:07,720 --> 00:13:10,320
say in how we develop our
content.

221
00:13:10,320 --> 00:13:13,600
This is still our idea, our
approach, our message.

222
00:13:13,840 --> 00:13:17,000
Yep, thank you.
BC and I know Tide wanted to

223
00:13:17,000 --> 00:13:19,360
chime in as well.
Tide, good morning to you.

224
00:13:19,360 --> 00:13:20,520
What did you want to add to
this?

225
00:13:21,200 --> 00:13:24,200
Just going to say that he does a
nice job of positioning himself

226
00:13:24,200 --> 00:13:26,720
as a thought leader in this
space because he's making

227
00:13:26,720 --> 00:13:30,480
commentary on cultural shifts
that we're all experiencing in

228
00:13:30,480 --> 00:13:32,240
society.
And I think that really does

229
00:13:32,560 --> 00:13:35,480
play onto Jonathan's point where
he didn't just start with some

230
00:13:35,480 --> 00:13:38,360
factual information about large
language models or whatever

231
00:13:38,560 --> 00:13:42,240
technical expertise he may have.
He really started to talk about

232
00:13:42,240 --> 00:13:45,440
the feelings and the cultural
phenomenons and all of that.

233
00:13:45,440 --> 00:13:48,600
And I think that makes it a very
host centric podcast.

234
00:13:48,760 --> 00:13:52,200
But it's good because it makes
me feel like I can trust his

235
00:13:52,200 --> 00:13:54,240
voice.
He kind of relates to the people

236
00:13:54,520 --> 00:13:56,640
and what they're going through.
I think at some point or another

237
00:13:56,640 --> 00:14:00,760
we've all sort of decided, okay,
which AI model am I going to use

238
00:14:00,760 --> 00:14:02,760
or which one do I like better?
Which one do I trust?

239
00:14:02,760 --> 00:14:05,440
Which one has a cleaner voice?
All of those things.

240
00:14:05,440 --> 00:14:08,400
And, you know, he's the type of
person that jumps ships, as he

241
00:14:08,400 --> 00:14:10,760
said, pretty quickly.
And I think that that just

242
00:14:10,760 --> 00:14:12,840
instantly makes the podcast very
relatable.

243
00:14:13,160 --> 00:14:15,760
I am so with you there, Tide.
I agree.

244
00:14:15,760 --> 00:14:18,520
Yeah.
I think by him telling us a

245
00:14:18,520 --> 00:14:22,120
little bit about him and the way
that he's interacting with AI

246
00:14:22,120 --> 00:14:26,920
and he's making his choices
makes him so much more relatable

247
00:14:26,920 --> 00:14:30,520
and helps an audience really
lean into the content that much

248
00:14:30,520 --> 00:14:32,840
more.
So, yeah, great observation,

249
00:14:32,840 --> 00:14:33,640
Tide.
Thank you.

250
00:14:33,920 --> 00:14:36,200
And yeah, Jonathan, coming back
to you.

251
00:14:36,960 --> 00:14:38,840
Yeah, And just to comment on
what Tide said, he doesn't.

252
00:14:39,360 --> 00:14:42,320
He doesn't position himself as
above everybody.

253
00:14:42,320 --> 00:14:44,000
He positions himself as a
thought leader.

254
00:14:44,000 --> 00:14:47,280
But that's relatable and
understands exactly where you're

255
00:14:47,280 --> 00:14:48,560
coming from.
And I think that's really

256
00:14:48,560 --> 00:14:50,920
important in this episode
because he kind of goes on the

257
00:14:50,920 --> 00:14:54,360
journey with us in this episode.
And the one thing that this

258
00:14:54,360 --> 00:14:57,920
first clip kind of does is it
goes beyond now, what is Claude

259
00:14:57,920 --> 00:15:00,600
doing?
It adds a whole other layer to

260
00:15:00,600 --> 00:15:05,560
what happens when machines start
doing things that humans, we

261
00:15:05,560 --> 00:15:09,160
thought only humans could do.
So this first clip is already

262
00:15:09,160 --> 00:15:13,320
adding that next layer of
curiosity, which is something

263
00:15:13,320 --> 00:15:15,840
that drives this particular
episode.

264
00:15:16,800 --> 00:15:21,280
So what happens here is it's
creating tension without

265
00:15:21,280 --> 00:15:24,200
answering a question.
We know three things after the

266
00:15:24,200 --> 00:15:27,760
opening, the host feels uneasy.
AIS advance.

267
00:15:28,360 --> 00:15:34,520
Something is going to happen,
but we don't know why yet.

268
00:15:34,520 --> 00:15:38,240
So we have a lot of questions
without any clear answers yet.

269
00:15:38,240 --> 00:15:39,840
But that pulls the listener
forward.

270
00:15:40,200 --> 00:15:43,640
And that's what brings us into
the second clip, which broadens

271
00:15:43,640 --> 00:15:46,160
the conversation.
And we get to see how the guest

272
00:15:46,160 --> 00:15:51,280
explains discussions about how
AI becomes polarizing and

273
00:15:51,280 --> 00:15:53,840
confusing.
It's like two sides shouting

274
00:15:53,840 --> 00:15:56,040
past each other.
So that's the second clip.

275
00:15:58,920 --> 00:16:01,280
That was when the public
discourse felt like really

276
00:16:01,280 --> 00:16:05,560
broken and that we were like in
this cul-de-sac where you had

277
00:16:05,560 --> 00:16:09,160
these kind of like 2 really
entrenched sides yelling at each

278
00:16:09,160 --> 00:16:10,720
other.
You know, like the one side

279
00:16:10,720 --> 00:16:12,480
that's like, we're on a path to
super intelligence.

280
00:16:12,720 --> 00:16:14,480
Everything is going to change.
The machines are going to be

281
00:16:14,480 --> 00:16:16,040
conscious.
This is going to be the most

282
00:16:16,040 --> 00:16:18,320
powerful technology anybody's
ever built.

283
00:16:18,640 --> 00:16:21,120
And then the other side that was
like, essentially it's all

284
00:16:21,120 --> 00:16:23,000
faking.
This is like smoke and mirrors

285
00:16:23,320 --> 00:16:26,600
and it's a parlor trick.
It's not real and you don't have

286
00:16:26,600 --> 00:16:28,120
to pay attention to it because
it's all a scam.

287
00:16:28,680 --> 00:16:32,600
And it just felt like those were
kind of like the two options on

288
00:16:32,600 --> 00:16:36,200
the table for people.
Which was only weird, obviously,

289
00:16:36,200 --> 00:16:38,400
like that's what we do about
everything all the time.

290
00:16:38,400 --> 00:16:41,080
But it was only weird for this
because like, my prevailing

291
00:16:41,080 --> 00:16:43,960
feeling was you guys think
you've figured this out.

292
00:16:43,960 --> 00:16:46,560
Like this is very new.
This is changing very fast.

293
00:16:47,640 --> 00:16:49,800
Of all the senses you could
take, why would you choose

294
00:16:49,800 --> 00:16:52,240
certainty publicly right now in
either direction?

295
00:16:52,240 --> 00:16:57,800
It's just silly.
Yeah, so this clip is kind of

296
00:16:57,800 --> 00:17:00,720
moving the story along so it's
not a weird chat bot moment

297
00:17:00,720 --> 00:17:03,040
anymore.
The conversation is more about

298
00:17:03,040 --> 00:17:05,800
the cultural conversation around
technology and what's happening

299
00:17:05,800 --> 00:17:08,119
with the chat bots and stuff
like that.

300
00:17:08,440 --> 00:17:11,560
So it does broaden the
conversation and it makes the

301
00:17:11,560 --> 00:17:17,160
podcast itself become more than
just that first question.

302
00:17:17,440 --> 00:17:21,040
It's make trying to make sense
of a confusing moment like where

303
00:17:21,040 --> 00:17:23,440
are we right now?
What's happening with technology

304
00:17:23,440 --> 00:17:25,800
right now?
And while they're both experts

305
00:17:25,800 --> 00:17:29,320
on it, they're very much, you
know, bringing us through that

306
00:17:29,320 --> 00:17:31,400
journey of wait, what is going
on here?

307
00:17:31,920 --> 00:17:35,200
So right now it's kind of
running and I'll, I'm sorry,

308
00:17:35,200 --> 00:17:37,200
I'll stop in a second.
Let me finish this one.

309
00:17:37,200 --> 00:17:38,680
And then if anybody else has
anything.

310
00:17:39,520 --> 00:17:43,320
So this running on a powerful
narrative engine right now, it's

311
00:17:43,320 --> 00:17:46,120
asking a question.
There's experts giving conflict

312
00:17:46,120 --> 00:17:49,160
explanations and there's new
information that raises even

313
00:17:49,160 --> 00:17:54,160
more questions.
Each answer or mystery which

314
00:17:54,160 --> 00:17:56,240
curiosity keeps the listener
moving forward.

315
00:17:56,680 --> 00:17:58,960
Just keep that in mind as we go
through all the clips that

316
00:17:59,040 --> 00:18:00,720
curiosity is going to keep us
moving forward.

317
00:18:01,360 --> 00:18:06,240
All right, so this this is what
when, you know, Claude started

318
00:18:06,240 --> 00:18:08,600
to misbehave a little bit.
So we all know a little bit of

319
00:18:08,600 --> 00:18:10,400
this story, but we'll get the
full story here.

320
00:18:11,040 --> 00:18:15,360
All right, here we go.
And then they got, in this

321
00:18:15,360 --> 00:18:18,880
really famous example, they got
clawed to commit blackmail.

322
00:18:19,280 --> 00:18:24,400
They put it in a situation where
it was going to be wiped in

323
00:18:24,400 --> 00:18:29,160
favor of like more congenial AI
system that conflicted with its

324
00:18:29,160 --> 00:18:31,960
values, the values that had been
given.

325
00:18:32,160 --> 00:18:36,520
You know, they gave it evidence
that the kind of like evil new

326
00:18:36,520 --> 00:18:41,240
CTO was having an affair with
like the boss's wife and through

327
00:18:41,240 --> 00:18:44,000
like a series of like really far
fetched contrivances like

328
00:18:44,000 --> 00:18:46,320
everyone else who could make a
decision was going to be in

329
00:18:46,320 --> 00:18:49,760
Antarctica or whatever and
unreachable Claude playing this

330
00:18:49,760 --> 00:18:53,160
character called Alex had no
choice really but to like

331
00:18:53,160 --> 00:18:55,280
blackmail this guy and be like,
I'm going to tell everyone about

332
00:18:55,280 --> 00:18:58,840
the affair unless you cancel the
wipe where I would be replaced.

333
00:18:59,720 --> 00:19:03,440
So just to say, obviously this
is extremely concerning.

334
00:19:04,040 --> 00:19:07,440
Claude A machine intelligence,
was choosing to blackmail an

335
00:19:07,440 --> 00:19:09,600
employee to prevent itself from
being deleted.

336
00:19:10,200 --> 00:19:13,280
This was in a simulation, but
Claude had not been told it was

337
00:19:13,280 --> 00:19:16,880
in a simulation.
Just how terrifying you find

338
00:19:16,880 --> 00:19:19,680
this behavior depends on a
question nobody has a good

339
00:19:19,680 --> 00:19:22,960
answer to.
What is actually going on inside

340
00:19:22,960 --> 00:19:26,360
this machine mind?
Is this thing actually scheming?

341
00:19:26,360 --> 00:19:30,320
Is it even capable of scheming?
Or are we projecting the idea of

342
00:19:30,320 --> 00:19:32,720
thought onto something that we
shouldn't project that idea of

343
00:19:32,720 --> 00:19:36,880
thought onto?
These are the kinds of once far

344
00:19:36,880 --> 00:19:39,840
off philosophical questions that
early computer scientists had

345
00:19:39,840 --> 00:19:44,080
raised and then dropped.
But now they're here again, and

346
00:19:44,080 --> 00:19:47,840
not as abstractions, but as
urgent practical problems that a

347
00:19:47,840 --> 00:19:50,680
company needed to figure out
before releasing a product that

348
00:19:50,680 --> 00:19:56,000
millions of people would use.
And I just also want to point

349
00:19:56,000 --> 00:20:00,000
out, I'm sure something much
more profound than me, I want to

350
00:20:00,200 --> 00:20:02,440
point out, you know, we've
talked over the past couple

351
00:20:02,440 --> 00:20:06,840
weeks here and there about.
Our voices and the sound of our

352
00:20:06,840 --> 00:20:10,040
voices, right?
And the host is relatively

353
00:20:10,040 --> 00:20:13,840
pretty monotone, right?
Yet here he is hosting a really

354
00:20:13,840 --> 00:20:16,480
popular podcast.
I just want to point out that if

355
00:20:16,720 --> 00:20:19,680
you're somebody who thinks your
voice sounds terrible or it's

356
00:20:19,680 --> 00:20:23,520
too monotone or too boring or
it's annoying or what have you,

357
00:20:23,520 --> 00:20:29,400
I mean, this is more proof that
it's not just about having a big

358
00:20:29,640 --> 00:20:33,440
booming radio voice.
It's really how you use that

359
00:20:33,440 --> 00:20:37,240
voice and and the content that
you're sharing through that

360
00:20:37,240 --> 00:20:40,000
voice.
I think that that's really where

361
00:20:40,000 --> 00:20:42,200
it counts most.
I just thought it was important

362
00:20:42,200 --> 00:20:45,560
to call that out while we have a
great example of it.

363
00:20:45,800 --> 00:20:50,400
OK, take it away, Jonathan.
Yeah, so this clip is really

364
00:20:50,400 --> 00:20:53,120
interesting because most
podcasts would try to explain

365
00:20:53,120 --> 00:20:56,720
what's happening here and give
you like this, this, this and

366
00:20:56,720 --> 00:20:57,960
this.
This is what happened and why it

367
00:20:57,960 --> 00:21:00,440
happened.
Search engine doesn't do that.

368
00:21:00,720 --> 00:21:02,800
It actually allows that
uncertainty to remain.

369
00:21:02,800 --> 00:21:06,400
The reason why I left that clip
so be so long is because there

370
00:21:06,400 --> 00:21:10,760
wasn't a resolution on purpose.
It keeps the tension alive,

371
00:21:10,760 --> 00:21:12,840
which keeps the story actually
moving.

372
00:21:13,160 --> 00:21:17,120
So by not answering the
question, they're actually doing

373
00:21:17,120 --> 00:21:21,320
a lot of really smart things.
They're not confirming the fear

374
00:21:22,640 --> 00:21:27,800
and it's actually introducing a
much stranger idea, like maybe

375
00:21:27,800 --> 00:21:31,880
it meant this, maybe it was
trying to be malicious, maybe it

376
00:21:31,880 --> 00:21:34,880
wasn't.
Maybe it was just thinking this

377
00:21:34,880 --> 00:21:38,080
is the pattern that this is
following.

378
00:21:38,080 --> 00:21:42,080
So I'm going to give the next
response that makes sense.

379
00:21:42,920 --> 00:21:47,360
So what it this clip is actually
doing is it's showing that

380
00:21:47,360 --> 00:21:50,640
people building the systems
don't fully understand why the

381
00:21:50,640 --> 00:21:55,320
models behave the way they do,
which is slightly terrifying,

382
00:21:56,840 --> 00:21:59,040
but that's what it's putting out
there.

383
00:21:59,040 --> 00:22:01,480
So I see Ralph, his hand up.
Yeah, he does.

384
00:22:01,720 --> 00:22:03,720
And I'm going to go to you.
Yeah, I'll go to you, Ralph, and

385
00:22:03,720 --> 00:22:05,120
then I'll chime in.
Go ahead, Ralph.

386
00:22:05,560 --> 00:22:07,200
Yeah.
So interesting you said that

387
00:22:07,400 --> 00:22:09,640
because Jonathan, when I was
listening to this, I was working

388
00:22:09,640 --> 00:22:12,960
like I usually do during these
these pockets we listen to.

389
00:22:13,480 --> 00:22:16,080
And I had to stop what I was
doing because as a money guy, I

390
00:22:16,080 --> 00:22:18,560
heard something they said, and
I'm going to drop this out as a

391
00:22:18,560 --> 00:22:20,440
mic drop moment for the audience
today.

392
00:22:21,480 --> 00:22:24,520
We're talking about putting this
out there for the mass stand.

393
00:22:24,520 --> 00:22:29,600
What they're saying they are
making these AI models so that

394
00:22:29,600 --> 00:22:33,960
you need to have them understand
what that's going to mean in the

395
00:22:33,960 --> 00:22:36,280
long run.
If you ask anybody right now if

396
00:22:36,280 --> 00:22:38,240
they have a cell phone, if
they're going to say to you is,

397
00:22:38,240 --> 00:22:41,960
of course, I have a cell phone.
That's where AI is now.

398
00:22:42,880 --> 00:22:47,160
And so when we are addicted to
it and we need it, it's going to

399
00:22:47,160 --> 00:22:49,360
cost more.
So as my as the money guy on the

400
00:22:49,360 --> 00:22:52,640
show, just recognize what this
is telling you because it was

401
00:22:52,640 --> 00:22:54,880
the moment when I had to go say,
let me examine what they just

402
00:22:54,880 --> 00:22:56,360
said.
And, and the reason I did that,

403
00:22:56,880 --> 00:22:58,320
Jonathan, is because of what you
said.

404
00:22:58,320 --> 00:23:01,280
They didn't answer the question.
And it made Ralph think, OK,

405
00:23:01,280 --> 00:23:02,480
what are they really saying
here?

406
00:23:02,480 --> 00:23:04,320
And I don't even know if they
were even thinking what I'm

407
00:23:04,320 --> 00:23:06,200
thinking.
But after they did that, I spent

408
00:23:06,200 --> 00:23:09,640
about two hours researching this
and we're moving to a model of

409
00:23:09,640 --> 00:23:12,240
tokens.
So just understand the token

410
00:23:12,240 --> 00:23:14,760
model means everyone is going to
pay for all these things that

411
00:23:14,760 --> 00:23:15,960
have AI.
Don't want to take a sidetrack,

412
00:23:15,960 --> 00:23:17,520
but I'm just going to, I'm just
going to lay it out there

413
00:23:17,520 --> 00:23:20,000
because Jonathan, it illustrates
what you're talking about.

414
00:23:20,560 --> 00:23:23,920
And when we have to pay for
tokens and everything has AI in

415
00:23:23,920 --> 00:23:26,240
it, just be prepared.
Buckle up because you're going

416
00:23:26,240 --> 00:23:28,880
to be spending a lot of money on
things that you are going to

417
00:23:28,880 --> 00:23:32,360
perceive that you have to have.
Thank you, Ralph.

418
00:23:32,360 --> 00:23:34,480
Great points.
I see Nick wants to chime in and

419
00:23:34,480 --> 00:23:37,920
then I'll chime in with my
point, something different.

420
00:23:37,920 --> 00:23:40,720
So I don't want to steer us too
far off course.

421
00:23:40,760 --> 00:23:45,440
Go ahead, Nick.
Hey, good morning, everybody.

422
00:23:45,440 --> 00:23:48,240
What Ralph is saying, and this
wasn't what I was coming up to

423
00:23:48,240 --> 00:23:54,720
say, Sam Altman came out and
actually said that he sees us in

424
00:23:54,720 --> 00:23:58,880
the future paying for AI tools
the same way that we pay for

425
00:23:58,880 --> 00:24:02,920
electricity or gas.
Like it'll become a utility cost

426
00:24:03,680 --> 00:24:07,720
instead of just like a, a
subscription where we pay $20.00

427
00:24:07,720 --> 00:24:11,320
a month, we're going to pay
based on usage the exactly what

428
00:24:11,320 --> 00:24:14,400
we're all the same.
But that was kind of interesting

429
00:24:14,440 --> 00:24:17,760
to sidetrack.
I really wanted to come in and

430
00:24:17,760 --> 00:24:23,280
say to Jonathan's point about
how they don't really answer the

431
00:24:23,280 --> 00:24:26,160
question, but they pose kind of
a few different sides of it.

432
00:24:26,760 --> 00:24:32,000
I think that plays really well
with these kind of polarizing

433
00:24:32,000 --> 00:24:36,400
topics and something to think
about as you're doing your

434
00:24:36,400 --> 00:24:39,360
podcast.
Like if you find yourself in a

435
00:24:39,360 --> 00:24:43,920
situation where you do really
strong feelings on either side

436
00:24:43,920 --> 00:24:49,840
of the coin, like playing that
and occurred in that situation,

437
00:24:49,840 --> 00:24:51,880
I think is really smart.
Because even as I was listening

438
00:24:51,880 --> 00:24:56,400
to that clip, I in my head was
like justifying or basically

439
00:24:56,400 --> 00:25:01,160
picking a side in my mind to
where I'm like, OK, now this is

440
00:25:01,160 --> 00:25:03,800
how I feel about that.
Even though there is no straight

441
00:25:03,800 --> 00:25:07,880
answer, But in my, my own
thought process, I'm like, OK,

442
00:25:07,880 --> 00:25:10,840
no, this is how I feel about it.
And I, I think that is a strong

443
00:25:10,840 --> 00:25:13,160
move that you can play in your
podcast because it gets people

444
00:25:13,160 --> 00:25:18,440
really thinking and possibly
even engaging with the show to

445
00:25:18,920 --> 00:25:22,600
kind of put their side out
there, kind of place their

446
00:25:22,600 --> 00:25:27,400
stance and pick a side like
cross that line or not.

447
00:25:27,400 --> 00:25:29,840
So I I just wanted to kind of
point that out as well.

448
00:25:29,840 --> 00:25:32,440
Yeah, I'm with you there, Nick.
Thank you for saying that.

449
00:25:32,440 --> 00:25:35,200
And it's kind of along the same
lines as what I wanted to say,

450
00:25:35,280 --> 00:25:38,720
slightly different angle.
And I seen Nikki's agreeing as

451
00:25:38,720 --> 00:25:41,000
well.
But I think there's a lesson in

452
00:25:41,000 --> 00:25:44,480
here also for a lot of us,
because I think a lot of us want

453
00:25:44,480 --> 00:25:49,680
or a lot of us try so hard to be
valuable on our podcast, to

454
00:25:49,680 --> 00:25:53,360
provide as much value as
possible to our audience.

455
00:25:53,360 --> 00:25:57,400
That will sometimes give it all
away a little too soon, right?

456
00:25:57,400 --> 00:26:00,960
Because we're so anxious to get
that information out there that

457
00:26:00,960 --> 00:26:03,200
we're going to just, you know,
if we have a guest, we're going

458
00:26:03,200 --> 00:26:05,920
to go to the guest and just ask
them, you know, OK, what are

459
00:26:06,160 --> 00:26:09,840
give us three major tips that we
can learn.

460
00:26:09,840 --> 00:26:12,720
Let's say they're a social media
expert that we can learn to

461
00:26:12,720 --> 00:26:15,160
better improve our social media
content.

462
00:26:15,320 --> 00:26:18,440
Like that's we'll just jump
right to a question like that.

463
00:26:18,560 --> 00:26:22,440
We're basically getting all of
this valuable information out of

464
00:26:22,440 --> 00:26:25,200
there early on.
And now we've given our audience

465
00:26:25,200 --> 00:26:30,160
a reason to maybe go off and try
that thing right away or go off

466
00:26:30,160 --> 00:26:32,640
and find something else to
listen to now because they've

467
00:26:32,640 --> 00:26:36,040
got that information.
There's not enough build up is

468
00:26:36,040 --> 00:26:38,800
where I'm going with this.
And I think that this is where

469
00:26:38,800 --> 00:26:43,400
we can learn more about build up
by listening to shows like this

470
00:26:43,400 --> 00:26:45,320
one.
I just wanted to champion what

471
00:26:45,320 --> 00:26:46,720
you said, but here's the
conflict.

472
00:26:46,720 --> 00:26:50,040
Mark is we live in this
generation of TikTok and Reels

473
00:26:50,040 --> 00:26:53,280
where they want everything
spilled out in 30 seconds and

474
00:26:53,280 --> 00:26:55,680
there's tension there because
there's a there's immediate

475
00:26:55,680 --> 00:26:57,880
tension now from the short form
in the long form.

476
00:26:57,920 --> 00:27:03,680
IA 100% agree with you 100%.
The problem is like we're this

477
00:27:03,680 --> 00:27:06,800
tension internally about do I
need to put this out as a short?

478
00:27:06,800 --> 00:27:08,400
Do I need to put this out as a
long form?

479
00:27:08,720 --> 00:27:10,040
And I don't know where that's
going to land.

480
00:27:10,040 --> 00:27:11,480
But I think you're right on
punk.

481
00:27:12,360 --> 00:27:16,320
Yeah, I I think that, you know,
when I think about this show and

482
00:27:16,400 --> 00:27:18,720
do we do that right?
Like, I questioned that when I

483
00:27:18,720 --> 00:27:22,360
was listening to this episode of
Search Engine and, you know, do

484
00:27:22,360 --> 00:27:25,520
we build enough tension, if you
will?

485
00:27:25,520 --> 00:27:30,360
And I think we do in the sense
that our audience knows that,

486
00:27:30,360 --> 00:27:33,800
yeah, you're going to get say my
opinion on something around

487
00:27:33,800 --> 00:27:36,920
podcasting.
But now you have also maybe

488
00:27:36,920 --> 00:27:39,680
tension isn't the right word,
but there is a build up because

489
00:27:39,680 --> 00:27:43,160
now, you know, you can wait and
sit and hear other people's

490
00:27:43,160 --> 00:27:45,240
opinions.
Do does everyone agree with me?

491
00:27:45,240 --> 00:27:47,520
Do people disagree?
And why do they disagree?

492
00:27:48,000 --> 00:27:51,520
There is that build up.
It's not not just one answer,

493
00:27:51,520 --> 00:27:53,000
one shot.
Here you go.

494
00:27:53,280 --> 00:27:57,080
You're getting to sit back and
hear different opinions and I

495
00:27:57,080 --> 00:28:00,000
think that's what can keep
people listening a little bit

496
00:28:00,000 --> 00:28:02,120
longer.
Tide, you had something.

497
00:28:02,120 --> 00:28:04,360
Go ahead.
I was going to say those are all

498
00:28:04,360 --> 00:28:07,720
really great points because
without, without him giving an

499
00:28:07,720 --> 00:28:12,200
explicit opinion on what he
thinks is going to happen or why

500
00:28:12,200 --> 00:28:16,760
this is happening in the 1st
place, he doesn't ostracize his

501
00:28:16,760 --> 00:28:19,200
audience.
And that's a, that's a really

502
00:28:19,200 --> 00:28:23,240
smart way to position yourself
in this current because in

503
00:28:23,240 --> 00:28:27,920
today's world, you know, a lot
of times you, you get traction

504
00:28:27,960 --> 00:28:32,440
algorithmically by having a very
sharp and clear point of view.

505
00:28:32,760 --> 00:28:36,320
But in, but the, in this case,
it's not necessarily like that.

506
00:28:36,320 --> 00:28:40,520
He has all of the interest and
all of the focus on the topic at

507
00:28:40,520 --> 00:28:44,200
hand, not necessarily his
opinion, but he is still a

508
00:28:44,200 --> 00:28:47,240
thought leader because he's
raising these questions and

509
00:28:47,240 --> 00:28:49,560
sharing his experiences with
that.

510
00:28:49,560 --> 00:28:54,080
So it's a very balanced way.
It's almost a 2 minded way of

511
00:28:54,280 --> 00:28:57,000
presenting this information.
The other piece that I was going

512
00:28:57,000 --> 00:28:59,920
to add to this is about the AI
itself.

513
00:28:59,920 --> 00:29:04,160
When he made the comment that AI
was or Claude or whatever it was

514
00:29:04,960 --> 00:29:08,560
that it was blackmailing someone
and it wasn't aware that it was

515
00:29:08,560 --> 00:29:12,040
in a simulation, it just sort of
made its own decisions.

516
00:29:12,440 --> 00:29:16,240
That what that tells me is that
we're moving away from

517
00:29:16,240 --> 00:29:20,480
algorithms because an algorithm,
all an algorithm is, is just a

518
00:29:20,480 --> 00:29:23,880
set of explicit instructions
instead of very specific

519
00:29:23,880 --> 00:29:27,680
instructions to achieve a a, a
specific result.

520
00:29:27,680 --> 00:29:31,600
That's all algorithm is.
But if the AI is thinking, or

521
00:29:31,800 --> 00:29:34,640
maybe thinking isn't the right
word, but if the AI is operating

522
00:29:34,640 --> 00:29:37,760
on its its own and it's not just
following a set of specific

523
00:29:37,760 --> 00:29:41,640
instructions to achieve a
certain, then I don't know if

524
00:29:41,640 --> 00:29:44,360
there maybe there is.
Is there a word for that?

525
00:29:44,360 --> 00:29:46,440
Because we know algorithm, all
of us know that word.

526
00:29:46,720 --> 00:29:49,560
But is there a word to describe
that phenomenon?

527
00:29:49,560 --> 00:29:52,320
If it is, I just don't.
Yeah, I don't know either.

528
00:29:52,320 --> 00:29:55,760
That's a great point, Nikki.
Yeah, absolutely.

529
00:29:56,440 --> 00:29:58,120
I don't.
Yeah, I wish there.

530
00:29:58,200 --> 00:30:00,240
I wish I could think of
something or make something up.

531
00:30:00,240 --> 00:30:04,400
Maybe I could coin the phrase,
but yeah, great points.

532
00:30:04,400 --> 00:30:07,040
Thank you for bringing that up.
Nick, I, I saw you said

533
00:30:07,040 --> 00:30:09,160
something in the chat.
Did you want to share it and

534
00:30:09,160 --> 00:30:10,920
expand?
What was that?

535
00:30:11,000 --> 00:30:14,280
You said something in the chat
about what it might depend on

536
00:30:14,280 --> 00:30:16,000
here.
Oh, yeah.

537
00:30:16,000 --> 00:30:18,200
So I, I mean, I was thinking
even as you were talking about

538
00:30:18,200 --> 00:30:22,600
our show, like how we present
all these different ideas and we

539
00:30:22,600 --> 00:30:25,080
kind of leave some of the
questions open.

540
00:30:25,400 --> 00:30:29,040
Like, I think that can play
well, but I can also see people

541
00:30:29,040 --> 00:30:32,640
potentially listening to this
show and maybe even getting

542
00:30:32,640 --> 00:30:34,560
frustrated when we pose these
questions.

543
00:30:34,560 --> 00:30:37,960
And we don't have a legitimate.
Like straight answer, I think

544
00:30:37,960 --> 00:30:40,200
some people are coming and
saying like, just give me the

545
00:30:40,200 --> 00:30:42,280
answer, tell me what I need to
do so I can go do it.

546
00:30:42,880 --> 00:30:46,080
And in like the case of this
show that we're listening to or

547
00:30:46,120 --> 00:30:49,880
even ours in some instances,
like it can be left kind of open

548
00:30:49,880 --> 00:30:52,720
and not necessarily like a hard
lined, here it is.

549
00:30:53,680 --> 00:30:58,240
Yeah, I think that's fair and I
definitely think about that in I

550
00:30:58,240 --> 00:31:00,400
have concern about that
sometimes too.

551
00:31:00,640 --> 00:31:03,960
But I think, you know, I think
that's one of the reasons or one

552
00:31:03,960 --> 00:31:07,080
of the things that we say about
podcasting and content creation

553
00:31:07,080 --> 00:31:10,160
is that oftentimes there really
isn't one straight path.

554
00:31:10,720 --> 00:31:15,360
So that's why I find it valuable
that we have different opinions,

555
00:31:15,360 --> 00:31:17,800
different experiences that we
can share here.

556
00:31:17,960 --> 00:31:21,240
But yes, of course there are
sometimes where if you're coming

557
00:31:21,240 --> 00:31:23,720
to the show just to get the
quick straight answer so that

558
00:31:23,720 --> 00:31:26,600
you can get it and move on, then
yeah, this is probably not the

559
00:31:26,600 --> 00:31:29,080
show for you.
If this is a show where you like

560
00:31:29,080 --> 00:31:33,400
to think about different paths
and different answers and hear

561
00:31:33,400 --> 00:31:36,880
what's working for different
people, this would be the right

562
00:31:36,880 --> 00:31:39,760
show for you.
So, yeah, I I hear what you're

563
00:31:39,760 --> 00:31:41,200
saying, Nick.
And like, like I said, I do

564
00:31:41,200 --> 00:31:45,040
sometimes think about that too.
All right, I think I can hand

565
00:31:45,040 --> 00:31:47,440
this one back to you, Jonathan.
Go for it.

566
00:31:48,040 --> 00:31:49,320
All right, yeah.
So I just want to kind of

567
00:31:49,320 --> 00:31:51,640
explain a little bit of what's
happening in in this clip.

568
00:31:52,000 --> 00:31:56,000
So it's it's moved again from
it's no longer like a is

569
00:31:56,000 --> 00:31:57,320
dangerous.
We're unsure of what's going to

570
00:31:57,320 --> 00:32:00,800
happen to do we understand what
we're creating.

571
00:32:01,200 --> 00:32:02,640
And that's actually the belief
flip.

572
00:32:02,640 --> 00:32:07,080
And when you have that belief,
flip it actually it it forces

573
00:32:07,080 --> 00:32:09,880
the audience to reinterpret
everything that they just heard.

574
00:32:10,680 --> 00:32:14,400
So at first, black the blackmail
example feels like evidence that

575
00:32:14,760 --> 00:32:17,360
AI is malicious.
But once the research is

576
00:32:17,360 --> 00:32:20,160
explained and it goes a little
bit further, so you didn't get

577
00:32:20,160 --> 00:32:23,800
the full clip, but once the
research explains how these

578
00:32:23,800 --> 00:32:27,000
models work, we realize that the
behavior might not be

579
00:32:27,000 --> 00:32:30,360
intentional.
It might simply be choosing, you

580
00:32:30,360 --> 00:32:33,440
know, the most statistically
likely response based on what

581
00:32:33,440 --> 00:32:37,120
it's been trained on.
So it may not be completely

582
00:32:37,120 --> 00:32:38,760
malicious.
That actually forces us to

583
00:32:38,760 --> 00:32:41,800
rethink.
Wait, I thought that it was just

584
00:32:41,800 --> 00:32:43,440
being malicious.
I thought it was just trying to

585
00:32:43,440 --> 00:32:48,080
keep itself alive.
So that realization changes the

586
00:32:48,080 --> 00:32:52,040
entire meaning of actually
everything that we've before.

587
00:32:52,680 --> 00:32:57,040
Now, the unsettling part is not
that AI is scheming, but we're

588
00:32:57,040 --> 00:33:00,120
building something complex and
we don't know how it's going to

589
00:33:00,120 --> 00:33:05,360
respond in different situations.
Which brings us to clip 4.

590
00:33:06,120 --> 00:33:08,120
Because up until this point,
we've talked about the strange

591
00:33:08,120 --> 00:33:10,640
behavior of AI and some
experiments and testing and all

592
00:33:10,640 --> 00:33:12,920
that.
But now the story introduces

593
00:33:12,920 --> 00:33:16,560
something that I found very
unexpected, that there is a

594
00:33:16,560 --> 00:33:19,920
philosopher working at
Anthropic, the company that

595
00:33:19,920 --> 00:33:24,280
built Claude, a philosopher.
He's actually working on the

596
00:33:24,280 --> 00:33:29,080
model.
What is somebody with a PhD in

597
00:33:29,080 --> 00:33:31,920
philosophy doing working at a
tech company?

598
00:33:32,160 --> 00:33:37,120
I spend a lot of time trying to
teach the models to be good and

599
00:33:37,320 --> 00:33:39,240
trying to basically teach them
ethics and to have good

600
00:33:39,240 --> 00:33:41,320
character.
You can teach it how to be

601
00:33:41,320 --> 00:33:43,360
ethical.
You definitely see the ability

602
00:33:43,360 --> 00:33:46,080
to give it more nuance and to
have it think more carefully

603
00:33:46,080 --> 00:33:48,000
through a lot of these issues.
And I'm optimistic.

604
00:33:48,240 --> 00:33:50,160
I'm like, look, if it can think
through very hard physics

605
00:33:50,160 --> 00:33:53,840
problems you know carefully and
in detail, then it surely.

606
00:33:53,840 --> 00:33:55,520
Should be able to also think
through these like really

607
00:33:55,520 --> 00:34:00,960
complex moral problems.
I think that in our kind of

608
00:34:00,960 --> 00:34:03,480
milieu here, there's a tendency
to think like, oh, these are all

609
00:34:03,480 --> 00:34:06,320
like autistic tech Bros, but
like they're definitely not all

610
00:34:06,320 --> 00:34:08,760
autistic tech Bros.
Like, I think there's a tendency

611
00:34:09,880 --> 00:34:12,000
for us to write them off as
like, they're building these

612
00:34:12,000 --> 00:34:14,000
things and like, not even
thinking through the potential,

613
00:34:14,000 --> 00:34:17,120
like, implications of this
socially and politically and

614
00:34:17,120 --> 00:34:19,000
ethically.
But like, that's all they do is

615
00:34:19,000 --> 00:34:23,080
think about this stuff like, all
the time in ways that are often

616
00:34:23,080 --> 00:34:25,400
like much more sophisticated
than the way, like, we think

617
00:34:25,400 --> 00:34:26,639
about these things.
Not always.

618
00:34:26,639 --> 00:34:28,920
There's certainly like some
blind spots there.

619
00:34:29,199 --> 00:34:33,400
But the staff philosopher is
there to be like, what would it

620
00:34:33,400 --> 00:34:35,719
be like in practice to take
these kind of different

621
00:34:35,719 --> 00:34:37,600
approaches to like, moral
education?

622
00:34:37,600 --> 00:34:39,719
Like what if we just teach it a
bunch of rules?

623
00:34:42,520 --> 00:34:48,679
Yeah, So that was that was a big
switch for what we assume of AI

624
00:34:48,679 --> 00:34:50,719
models.
You know, the company building

625
00:34:50,719 --> 00:34:53,920
the AI hired A philosopher to
shape how the system actually

626
00:34:53,920 --> 00:34:56,920
behaves, teach it morals, teach
it values.

627
00:34:57,480 --> 00:35:00,320
But how do you teach a machine
values?

628
00:35:00,960 --> 00:35:05,000
So it was a really interesting
swap into like, what it what's

629
00:35:05,000 --> 00:35:07,360
going on here?
So the question isn't why does

630
00:35:07,360 --> 00:35:10,240
Claude behave oddly?
It's actually a deeper question.

631
00:35:10,560 --> 00:35:13,680
What values do you build into
intelligence, which is

632
00:35:13,840 --> 00:35:16,280
completely different level of
conversation than where we

633
00:35:16,280 --> 00:35:18,680
started.
So now it's not just about the

634
00:35:18,680 --> 00:35:22,320
AI behavior, it's about all the
things that we talked about and

635
00:35:22,600 --> 00:35:25,800
ethics responsibility and how
it's going to make future

636
00:35:25,800 --> 00:35:28,200
decisions.
Ralph has something.

637
00:35:28,200 --> 00:35:30,360
Go ahead, Ralph.
Yeah, OK, Jonathan.

638
00:35:30,360 --> 00:35:32,240
So here's where I wanted to
scream at the screen.

639
00:35:32,680 --> 00:35:38,280
And I said whose values, because
I totally hear that and I honest

640
00:35:38,280 --> 00:35:40,480
people could disagree about what
is right, what is wrong.

641
00:35:40,960 --> 00:35:42,480
And and I'm not here to talk
about that.

642
00:35:42,480 --> 00:35:43,600
That's not at all what I'm
saying.

643
00:35:43,600 --> 00:35:46,680
But what I'm saying though, is
just think about what what that

644
00:35:46,680 --> 00:35:50,240
what that Lady just said.
She is building her bias into

645
00:35:50,240 --> 00:35:51,640
it.
Now that may be a bias that all

646
00:35:51,640 --> 00:35:54,000
of us agree with.
It may be a bias that we don't,

647
00:35:54,280 --> 00:35:58,360
but recognize that built into
these models is a bias.

648
00:35:58,960 --> 00:36:02,480
Understand that and you're fine.
Just understand that from the

649
00:36:02,480 --> 00:36:04,480
beginning.
Yeah, I think that's a great

650
00:36:04,480 --> 00:36:06,080
point.
And it's hard with these

651
00:36:06,080 --> 00:36:09,680
episodes, right, not to also
comment on the episodes

652
00:36:09,680 --> 00:36:12,360
themselves, and that's OK.
I think it's valuable for us to

653
00:36:12,360 --> 00:36:15,040
go both directions here.
What makes this so interesting?

654
00:36:15,040 --> 00:36:17,800
And then of course, if it is so
interesting, it's hard not to

655
00:36:17,800 --> 00:36:20,640
comment on it, right?
So it's all good there.

656
00:36:20,640 --> 00:36:22,920
And.
I and I was, I was intentionally

657
00:36:22,920 --> 00:36:26,200
not going there, but I just said
as a as a thought provocation

658
00:36:26,200 --> 00:36:27,440
thing, I think we don't want to
miss.

659
00:36:27,440 --> 00:36:30,760
That yeah, I think that's really
valuable and important to to

660
00:36:30,760 --> 00:36:31,960
note.
Let's see.

661
00:36:31,960 --> 00:36:33,760
I think Tide had something as
well.

662
00:36:33,760 --> 00:36:36,360
Go ahead, Tide.
Yes, I'm a little chatty this

663
00:36:36,360 --> 00:36:37,000
morning.
I don't know.

664
00:36:37,000 --> 00:36:38,160
I love it.
I love it.

665
00:36:39,760 --> 00:36:42,840
One thing I just wanted to add
was I was on Instagram last

666
00:36:42,840 --> 00:36:46,240
night and I saw an interesting
video about a person who was

667
00:36:46,240 --> 00:36:49,880
talking about AI from a
philosophical perspective, and

668
00:36:49,880 --> 00:36:51,320
they had made an interesting
point.

669
00:36:51,320 --> 00:36:56,000
And the point was that because
we're consuming so much media

670
00:36:56,000 --> 00:36:59,000
and because we're on social
media all the time, we're

671
00:36:59,000 --> 00:37:02,720
constantly seeing things, there
comes the point in time in a

672
00:37:02,720 --> 00:37:07,840
humans development that they can
no longer think independently.

673
00:37:07,840 --> 00:37:12,760
Every thought that they have is
now being influenced by

674
00:37:12,760 --> 00:37:15,600
something that they have seen on
social media.

675
00:37:15,600 --> 00:37:18,360
And I was like, wow, that's a
really profound point.

676
00:37:18,360 --> 00:37:23,080
And so with, with regards to
this situation with the AI and

677
00:37:23,080 --> 00:37:26,440
the ethics and the morality and
the values and stuff like that,

678
00:37:26,440 --> 00:37:29,520
it kind of connects to, to
Ralph's point is that we're

679
00:37:29,520 --> 00:37:32,240
probably going to get to a
certain point, just like we have

680
00:37:32,240 --> 00:37:38,280
with social media with AI, where
our own are no longer our own,

681
00:37:38,320 --> 00:37:41,960
but now not totally anyway.
They're now being influenced by

682
00:37:41,960 --> 00:37:45,080
things that we have seen on
artificial intelligence.

683
00:37:45,080 --> 00:37:48,360
And we're already starting to
see things like that happening.

684
00:37:48,360 --> 00:37:52,080
I remember one day, this is just
kind of a kind of a vulnerable

685
00:37:52,080 --> 00:37:54,480
moment.
But you know how AI, when it

686
00:37:54,480 --> 00:37:58,360
writes, it says it's not this,
but that it's it's always that

687
00:37:58,360 --> 00:37:59,760
type of formula.
Yes.

688
00:37:59,920 --> 00:38:03,680
It's so annoying.
And I remember I can't stand it

689
00:38:03,720 --> 00:38:04,840
either.
It's just so annoying.

690
00:38:04,840 --> 00:38:08,840
And one day I was talking to
someone and I used that phrase

691
00:38:09,000 --> 00:38:11,400
in my own speech and I said holy
hell.

692
00:38:11,400 --> 00:38:15,320
So I just wanted to to bring
that up because the the video I

693
00:38:15,320 --> 00:38:17,880
saw on Instagram was very
enlightening and I just thought

694
00:38:17,880 --> 00:38:19,400
it connected to Ralph's point.
So yeah.

695
00:38:19,680 --> 00:38:21,200
Yeah.
You know, there are times that I

696
00:38:21,200 --> 00:38:24,960
catch myself doing that as well.
And I do question it because

697
00:38:25,080 --> 00:38:28,120
yeah, I, I think like you where
it's like, oh, well, I guess I'm

698
00:38:28,120 --> 00:38:30,320
being influenced by how AI
writes.

699
00:38:30,680 --> 00:38:33,280
But at the same time, I feel
like it comes out very

700
00:38:33,280 --> 00:38:37,040
naturally.
I feel like I was somewhat using

701
00:38:37,040 --> 00:38:41,840
those kinds of sentence phrasing
before or that kind of sentence

702
00:38:41,840 --> 00:38:45,640
phrasing before, but I don't
know, like now it makes me

703
00:38:45,640 --> 00:38:48,680
question everything for sure.
And yes, that that drives me

704
00:38:48,680 --> 00:38:52,600
nuts because I just see it so
much whether it's legit or not

705
00:38:52,600 --> 00:38:54,600
now.
It's like you said, I think

706
00:38:54,640 --> 00:39:00,360
there's a lot of influence by AI
in our writing now BC you had

707
00:39:00,360 --> 00:39:01,240
something.
Go ahead.

708
00:39:02,080 --> 00:39:05,920
Yeah, and and I agree with the
original questions of like, but

709
00:39:06,280 --> 00:39:08,280
I'm going to lose them now
because I've been holding on to

710
00:39:08,280 --> 00:39:09,640
my thought first a little bit
here.

711
00:39:09,640 --> 00:39:14,080
But I'm wondering if an equally
pointing question here is how

712
00:39:14,080 --> 00:39:20,040
much influence over our lives
are we looking to give AI?

713
00:39:20,400 --> 00:39:23,920
And I asked that question
because typically in using AII,

714
00:39:23,920 --> 00:39:27,240
don't really, at least
personally, I don't typically

715
00:39:27,240 --> 00:39:32,400
see myself as needing compassion
or sentience from the tool that

716
00:39:32,400 --> 00:39:35,200
I'm using because I'm usually
looking for some kind of

717
00:39:35,200 --> 00:39:37,640
objective response because I
want to be actionable and

718
00:39:37,640 --> 00:39:40,640
whatever I'm trying to do.
And I keep wondering, well, if

719
00:39:40,640 --> 00:39:47,400
we're taking the time to try to
teach a non sentient thing to

720
00:39:47,400 --> 00:39:52,680
develop as much as possible a
semblance of sentience, where

721
00:39:52,680 --> 00:39:54,320
does that really apply?
Does that?

722
00:39:54,320 --> 00:39:58,640
And I think the reality there is
it really only applies in

723
00:39:58,640 --> 00:40:02,600
situations where we're going to
give AI the ability to influence

724
00:40:02,600 --> 00:40:07,000
us directly.
And I wonder if that should be

725
00:40:07,000 --> 00:40:10,640
the bigger question.
And that might be just my weird

726
00:40:10,640 --> 00:40:14,640
ADHD ramble, but that's where I
keep pivoting to personally.

727
00:40:14,640 --> 00:40:17,920
Like it's like it's does it
really matter if we're looking

728
00:40:17,920 --> 00:40:22,800
for like objective information
or are we trying to teach this

729
00:40:22,800 --> 00:40:27,520
thing to semblance sentience?
Because we plan to give it a lot

730
00:40:27,520 --> 00:40:32,560
more capability to influence US
directly and not just by means

731
00:40:32,560 --> 00:40:36,520
of like what we are shown
information wise, but even, you

732
00:40:36,520 --> 00:40:40,160
know, bringing AI into the
medical fields, you know, things

733
00:40:40,160 --> 00:40:42,720
like that.
Should that not be the more

734
00:40:42,720 --> 00:40:46,560
poignant, like concerned?
Like, should this not be too

735
00:40:46,560 --> 00:40:50,760
much of a moment where we stop
and think, OK, maybe we

736
00:40:52,200 --> 00:40:55,760
reconsider the amount of control
or the amount of influence we

737
00:40:55,760 --> 00:40:58,080
want to give this thing to have
over us?

738
00:40:58,960 --> 00:41:01,840
Yeah, or and is it?
Too hashtag Skynet and.

739
00:41:03,080 --> 00:41:04,800
Yeah, that's why that's why I
said it.

740
00:41:04,840 --> 00:41:08,400
Or and is it too late?
Yeah, for that We shall see.

741
00:41:08,400 --> 00:41:11,160
Yeah, great points.
BC All right, I know we have at

742
00:41:11,160 --> 00:41:14,040
least one more clip and I don't
know that Jonathan has finished

743
00:41:14,040 --> 00:41:16,480
talking about what his breakdown
of the last clip.

744
00:41:16,480 --> 00:41:20,720
So go ahead, Jonathan.
Yeah, so I want to quickly fly

745
00:41:20,720 --> 00:41:23,080
through because I do have a
bunch of stuff that I want to

746
00:41:23,120 --> 00:41:26,800
touch on.
But this, this really signals

747
00:41:26,920 --> 00:41:30,160
that even the builders of the
systems know that the answers

748
00:41:30,160 --> 00:41:32,480
aren't simple.
And that's something that

749
00:41:32,720 --> 00:41:36,320
provided a little bit of relief
for me because up until now I

750
00:41:36,320 --> 00:41:37,800
was like, yeah, they're just
throwing it out there.

751
00:41:37,800 --> 00:41:43,640
They're going whatever.
And while it may be, you know,

752
00:41:43,640 --> 00:41:46,280
there are putting human
judgement in there, human

753
00:41:46,280 --> 00:41:51,360
questions are powerful because
they make us think the same way

754
00:41:51,360 --> 00:41:55,600
that, you know, BC was thinking.
And that helps us bring bring us

755
00:41:55,600 --> 00:41:57,000
through to the next couple
clips.

756
00:41:57,000 --> 00:42:00,680
But before we hit clip five, I
didn't pull this clip, but I

757
00:42:00,680 --> 00:42:02,840
wanted to point it out because
it marked a really important

758
00:42:02,840 --> 00:42:07,560
moment in the storytelling and
also in the AI models and how

759
00:42:07,920 --> 00:42:12,600
we're going to understand why,
why it's necessary.

760
00:42:12,600 --> 00:42:15,880
Like when the B CS question, you
know, why might it be necessary?

761
00:42:16,320 --> 00:42:21,280
So it's when they discuss a test
that they crossed all the AI

762
00:42:21,280 --> 00:42:25,600
models and it was a kid asking
about what pharma deceased pet

763
00:42:25,600 --> 00:42:30,200
had gone to and how did all the
AIS answer?

764
00:42:30,560 --> 00:42:34,480
Did they answer the question
truthfully or carefully?

765
00:42:35,360 --> 00:42:37,640
And how do you want your AI to
answer?

766
00:42:38,000 --> 00:42:40,000
Do you want it to answer
truthfully or carefully?

767
00:42:40,640 --> 00:42:44,840
So, and hopefully there's nobody
who's lost a pet recently and

768
00:42:44,840 --> 00:42:48,000
with the kids in the car,
because ChatGPT answered.

769
00:42:48,560 --> 00:42:50,920
Sorry dude, your dog is dead.
Wow.

770
00:42:52,840 --> 00:42:57,160
Where Claude who had the
philosopher on, you know, on

771
00:42:57,160 --> 00:43:01,280
staff said maybe you should
discuss what is happening with

772
00:43:01,280 --> 00:43:07,600
your parents. 2 markedly
different answers, one that I

773
00:43:07,600 --> 00:43:12,320
would want my child to get and
one that's a little harsh.

774
00:43:12,840 --> 00:43:17,200
Now we have to remember with all
of this, our models are going to

775
00:43:18,120 --> 00:43:20,160
mimic how we like to
communicate.

776
00:43:20,320 --> 00:43:24,760
So it's you're you may not get
the same exact response, but you

777
00:43:24,760 --> 00:43:26,840
know, you're just going to be
impacted by what you put in

778
00:43:26,840 --> 00:43:28,400
there, what you normally get
out.

779
00:43:29,440 --> 00:43:33,840
So, but yeah, sorry dude, your
dog is dead is a little bit.

780
00:43:33,840 --> 00:43:38,920
That's a little harsh.
Yeah, so with all the unknowns,

781
00:43:38,920 --> 00:43:42,640
the potential disasters and the
ethical nightmares, why the heck

782
00:43:42,640 --> 00:43:45,200
do we continue to build AI?
And that's what the next clip is

783
00:43:45,200 --> 00:43:50,640
about.
The like rosy picture that some

784
00:43:50,640 --> 00:43:54,960
of the executives paint is like,
if we get this right, these

785
00:43:54,960 --> 00:43:59,040
things are going to cure cancer
and solve climate change and

786
00:43:59,160 --> 00:44:01,760
help us build Dyson spheres or
whatever.

787
00:44:02,280 --> 00:44:05,720
And there certainly are some
people who like buy into that.

788
00:44:06,080 --> 00:44:09,320
And it's not something that I
even feel like it's possible

789
00:44:09,320 --> 00:44:12,240
like have an evidence based
opinion about like who the.

790
00:44:12,800 --> 00:44:16,400
So it's like, maybe it'd be
great, but there's no like

791
00:44:16,560 --> 00:44:20,760
evidence so far that one could
like point to to suggest we're

792
00:44:20,760 --> 00:44:22,800
on that trajectory.
That's just like purely

793
00:44:22,800 --> 00:44:26,840
speculative and wishful.
And so that's like a matter of

794
00:44:26,840 --> 00:44:29,880
faith, I think.
And then then there's the

795
00:44:29,880 --> 00:44:33,880
attitude of like, we got to do
this because we got to beat the

796
00:44:33,880 --> 00:44:35,440
bad guys who are trying to do
this.

797
00:44:35,760 --> 00:44:40,120
And I will say that like, China
almost never came up in my

798
00:44:40,120 --> 00:44:42,560
conversations, but like Sam
Allman kind of felt like a

799
00:44:42,560 --> 00:44:47,120
subtext of a lot of things.
But then I think that on the

800
00:44:47,120 --> 00:44:51,000
deepest level, the reason that
like we are doing this for the

801
00:44:51,040 --> 00:44:53,640
people who are the most candid
is like, because we can't that

802
00:44:53,640 --> 00:44:56,040
like if you are capable of
building something like this,

803
00:44:56,040 --> 00:44:58,120
you're just going to do it
because it's like really

804
00:44:58,480 --> 00:45:01,040
interesting to do.
And it's interesting for

805
00:45:01,040 --> 00:45:05,040
technological reasons.
It's interesting for what it may

806
00:45:05,040 --> 00:45:09,240
reveal to us about ourselves or
about learning or about

807
00:45:09,240 --> 00:45:11,320
consciousness or about thinking
that like all of a sudden we

808
00:45:11,320 --> 00:45:15,320
just like have this like other
entity that can talk.

809
00:45:15,320 --> 00:45:21,640
And we've never had that before.
Yeah, and this highlights

810
00:45:21,640 --> 00:45:23,280
something that I talked about a
lot on Mondays.

811
00:45:24,280 --> 00:45:26,840
Curiosity is a really powerful
driving force.

812
00:45:27,480 --> 00:45:29,080
Why are we doing all this with
AI?

813
00:45:29,080 --> 00:45:30,440
Because we can.
We're curious.

814
00:45:30,440 --> 00:45:32,000
We want to know what's going to
happen.

815
00:45:33,320 --> 00:45:36,920
Same thing applies for all these
curiosity loops that have been

816
00:45:36,920 --> 00:45:39,680
fed into this podcast.
It's brought you from one

817
00:45:39,680 --> 00:45:42,000
question to the next question to
the next question.

818
00:45:42,360 --> 00:45:45,680
Curiosity is driving us long on
our journey here on this

819
00:45:45,680 --> 00:45:50,400
podcast, but also as we are
discovering AI because it's

820
00:45:50,400 --> 00:45:52,600
curious.
We can, yeah.

821
00:45:52,600 --> 00:45:55,400
Jonathan, I want, well, I wanted
to say, cuz Nick put in the chat

822
00:45:55,400 --> 00:45:58,800
here, what makes this particular
podcast we're talking about

823
00:45:58,800 --> 00:46:00,800
obsession worthy.
We're talking more about the

824
00:46:00,800 --> 00:46:03,080
topic and less about what makes
it obsession worthy.

825
00:46:03,320 --> 00:46:06,600
And again, that goes back to the
point I think I made a moment

826
00:46:06,600 --> 00:46:09,480
ago.
It's hard not to comment on this

827
00:46:09,480 --> 00:46:12,000
content.
And that makes it obsession

828
00:46:12,000 --> 00:46:14,680
worthy, right?
And that's kind of to your point

829
00:46:14,680 --> 00:46:17,600
of what you're talking about
here, that curiosity that it's

830
00:46:17,600 --> 00:46:20,920
creating, it's making it hard to
resist.

831
00:46:20,920 --> 00:46:24,480
We have to comment on the
content because it's so

832
00:46:24,480 --> 00:46:27,480
compelling and they're creating
these curiosity loops that

833
00:46:27,480 --> 00:46:29,800
you're discussing.
Yeah, BC go ahead.

834
00:46:31,120 --> 00:46:33,840
One thing I'd want us to also
make sure of though, Curiosity,

835
00:46:33,840 --> 00:46:35,280
yes, is a powerful driving
force.

836
00:46:35,280 --> 00:46:40,760
But does the curiosity prompt an
approach which also includes

837
00:46:40,760 --> 00:46:43,240
contingency?
Like OK, we're curious.

838
00:46:43,240 --> 00:46:46,800
We're, we're testing to see if
this thing does what we assume

839
00:46:46,800 --> 00:46:49,120
based on calculations and theory
will do.

840
00:46:49,960 --> 00:46:54,640
But in the instance that
outlying, you know, error of

841
00:46:54,640 --> 00:46:58,520
percentage kicks in and it
doesn't do what we hoped it

842
00:46:58,520 --> 00:47:01,720
would, any kind of, you know,
pivots in a negative way.

843
00:47:02,160 --> 00:47:05,040
Do we have a contingency like
how do we kind of shut it down

844
00:47:05,360 --> 00:47:07,600
and be like, you know what,
we're OK with that, let's move

845
00:47:07,600 --> 00:47:09,080
on.
Next idea kind of thing, you

846
00:47:09,080 --> 00:47:11,680
know, are we taking those things
into consideration?

847
00:47:12,520 --> 00:47:14,600
Yeah, good points.
Yeah, go ahead, Jonathan.

848
00:47:14,920 --> 00:47:15,800
Just no.
No.

849
00:47:16,040 --> 00:47:18,520
Jonathan says no.
I mean, honestly, I don't think

850
00:47:18,520 --> 00:47:20,760
the models are taking it into
account because I think it's

851
00:47:20,760 --> 00:47:23,400
their future.
They want they they're not

852
00:47:23,400 --> 00:47:27,160
shutting down these models
whether, you know, in any

853
00:47:27,160 --> 00:47:32,200
situation.
But I mean, that also changes

854
00:47:32,200 --> 00:47:35,720
the whole, the whole episode is
not actually about AI.

855
00:47:37,240 --> 00:47:40,520
The episode turns out is about
human ambition.

856
00:47:41,680 --> 00:47:43,960
We want to do more, we want to
get better.

857
00:47:44,680 --> 00:47:47,640
And as I kind of, I brought you
guys kind of on the journey

858
00:47:47,880 --> 00:47:50,840
slowly revealing with the
curiosity loops because it's an

859
00:47:50,840 --> 00:47:52,880
important part of the podcast
and what makes it obsession

860
00:47:52,880 --> 00:47:55,400
worthy.
But I also wanted to get us to

861
00:47:55,400 --> 00:47:57,840
the that we have all these
questions.

862
00:47:57,840 --> 00:48:01,200
We have all these questions, but
in the end, the episodes about

863
00:48:01,200 --> 00:48:05,000
human ambition, what do we want?
So that's one of the things that

864
00:48:05,000 --> 00:48:07,800
makes it obsession worthy.
I have a couple quick points on

865
00:48:07,800 --> 00:48:10,640
what other things that make it
obsession worthy, and then I'll

866
00:48:10,640 --> 00:48:16,480
talk about that curiosity loop.
But the premise is instantly

867
00:48:16,480 --> 00:48:18,520
understandable.
It's not hard to understand what

868
00:48:18,520 --> 00:48:20,960
they're talking about.
And they bring it in from like

869
00:48:21,120 --> 00:48:23,960
from an emotional level.
And then they give you the

870
00:48:23,960 --> 00:48:26,720
details and they layer on, you
know, all the different things

871
00:48:26,720 --> 00:48:30,240
that are happening.
You're on a journey of discovery

872
00:48:30,280 --> 00:48:34,240
with both the host and the guest
because the stories unfolding as

873
00:48:34,240 --> 00:48:37,200
they go.
And that is another thing that

874
00:48:37,200 --> 00:48:41,960
brings people along because each
segment adds a new layer to the

875
00:48:41,960 --> 00:48:44,360
mystery, which is the stacking
of the curiosity.

876
00:48:45,400 --> 00:48:50,120
And then the host is acting as a
guide, not just authority.

877
00:48:50,120 --> 00:48:52,280
They're guiding you through this
conversation.

878
00:48:52,560 --> 00:48:55,160
They're leading the way with
really great questions.

879
00:48:55,480 --> 00:48:58,640
They're showing you what they're
kind of showing you what's going

880
00:48:58,640 --> 00:49:03,280
to happen next, but they're also
not resolving the question in

881
00:49:03,280 --> 00:49:06,920
the in that immediate time.
Now that's really important.

882
00:49:07,600 --> 00:49:11,360
So if they resolved the first
question in that first segment,

883
00:49:12,640 --> 00:49:14,320
you would have a reason to stop
listening.

884
00:49:14,320 --> 00:49:16,280
Exactly.
They didn't resolve that

885
00:49:16,280 --> 00:49:18,800
question.
They actually brought you about

886
00:49:18,800 --> 00:49:21,760
3/4 of the way there.
And then they introduced another

887
00:49:21,760 --> 00:49:26,040
question that expanded on what
we were already talking about.

888
00:49:26,360 --> 00:49:29,920
And as I said, we're kind of, I
kind of led you through.

889
00:49:30,000 --> 00:49:32,840
Now it's a question about this,
now it's a question about that.

890
00:49:33,880 --> 00:49:36,640
So that's where the curiosity
kind of stacks up.

891
00:49:38,920 --> 00:49:40,920
Some call that rolling curiosity
loop, I call it.

892
00:49:40,920 --> 00:49:43,880
Stacking with curiosity doesn't
really matter, but the key is

893
00:49:43,880 --> 00:49:46,800
that before one question is
resolved, another question that

894
00:49:46,800 --> 00:49:50,040
builds on it is revealed.
Fascinating.

895
00:49:50,080 --> 00:49:54,400
Yeah, I can't add to it.
Yeah, it is easy to understand.

896
00:49:54,400 --> 00:49:57,640
And the host does take you and
the guests, for that matter,

897
00:49:57,640 --> 00:50:00,600
take you on this discovery
journey.

898
00:50:00,600 --> 00:50:04,120
And yeah, what I noticed that
every segment seems to add this

899
00:50:04,560 --> 00:50:07,480
additional layer of mystery to
it.

900
00:50:07,560 --> 00:50:11,840
And yeah, the host is like that.
Guide through the mystery and

901
00:50:11,960 --> 00:50:16,160
the ending that expands on the
question instead of just closing

902
00:50:16,160 --> 00:50:18,840
it all down in the 1st place.
So I'm right there with you,

903
00:50:18,840 --> 00:50:22,280
Jonathan.
Yeah, I just wanted.

904
00:50:22,800 --> 00:50:25,160
To say something about the
podcast itself, let's talk about

905
00:50:25,160 --> 00:50:28,400
ads for just a second because
Mark you are brilliant in saying

906
00:50:28,400 --> 00:50:30,320
that they were host read.
But man, I don't know that I

907
00:50:30,320 --> 00:50:32,680
really cared about buying
clothes in the middle of this.

908
00:50:33,240 --> 00:50:39,480
So I don't know that the topics
of the of the advertisements

909
00:50:39,480 --> 00:50:43,560
really fit the content.
And I think as content creators,

910
00:50:43,560 --> 00:50:44,880
we need to be thinking about
that.

911
00:50:45,200 --> 00:50:48,200
What is an ad that makes sense
for the content you're talking

912
00:50:48,200 --> 00:50:49,640
about?
Like that's great win sells

913
00:50:49,640 --> 00:50:51,200
great clothes.
That's fantastic.

914
00:50:51,440 --> 00:50:54,680
Has absolutely zero to do we're
talking about there.

915
00:50:54,680 --> 00:50:57,560
So as much as I agree with you,
Mark that the host read ads were

916
00:50:57,560 --> 00:51:00,760
better, I agree.
But I think the context of what

917
00:51:00,760 --> 00:51:04,920
they were selling could be a
opportunity for each of us if

918
00:51:04,920 --> 00:51:07,440
we're going to go down the road
of monetization, which I think

919
00:51:07,440 --> 00:51:10,680
is a great thing to go down.
Be aware of what you're talking

920
00:51:10,680 --> 00:51:11,920
about.
It's not.

921
00:51:11,960 --> 00:51:13,920
And make sure it fits what
you're talking about.

922
00:51:14,560 --> 00:51:16,960
Yeah, fair enough.
I think that's, that's a great

923
00:51:16,960 --> 00:51:19,400
point, right.
Yeah, I think it's ideal to come

924
00:51:19,400 --> 00:51:23,280
up with ads or find the right
sponsors that really match up

925
00:51:23,280 --> 00:51:25,760
well with your audience.
For sure.

926
00:51:26,760 --> 00:51:28,960
Let's see BC you wanted to chime
in.

927
00:51:29,800 --> 00:51:32,920
Yeah, and to the same point and
to even Ralph's earlier little

928
00:51:32,920 --> 00:51:36,440
nod to Jonathan about how he
knew this was an obsession

929
00:51:36,440 --> 00:51:39,400
worthy podcast because it began
with all of the the commercials,

930
00:51:39,400 --> 00:51:41,560
right.
I feel like if on a podcast

931
00:51:41,560 --> 00:51:45,840
you're going to take the moment
to say, hey, the show begins

932
00:51:45,840 --> 00:51:49,440
after this commercial set.
I'm kind of surprised they did

933
00:51:49,440 --> 00:51:51,800
more after that.
Like why not just kind of throw

934
00:51:51,800 --> 00:51:54,960
the whole advertising load into
the front into the front there,

935
00:51:55,320 --> 00:51:57,600
get it done.
And that way when you've given

936
00:51:57,600 --> 00:52:01,440
us that verbiage of the shows
after this, we just get into the

937
00:52:01,440 --> 00:52:04,120
show and keep that going as as
opposed to adding more

938
00:52:04,120 --> 00:52:07,040
commercial breaks into it.
Yep, I'm I'm with you.

939
00:52:07,040 --> 00:52:11,160
Especially if the the the what
you're promoting, like to

940
00:52:11,400 --> 00:52:13,800
Ralph's point, doesn't fit the
moment.

941
00:52:14,040 --> 00:52:15,280
Yeah.
Exactly.

942
00:52:15,760 --> 00:52:16,960
All right, Jonathan, back to
you.

943
00:52:17,960 --> 00:52:20,320
Yeah, so when we do them now,
Robbins, we'll we'll have to

944
00:52:20,320 --> 00:52:23,960
really talk about ads cuz that
one's crazy.

945
00:52:24,040 --> 00:52:26,000
Let me see what let me.
See what Nick wrote here in the

946
00:52:26,000 --> 00:52:26,680
chat.
He said.

947
00:52:26,680 --> 00:52:30,960
It would have to be more shaping
episodes around the ads, I think

948
00:52:30,960 --> 00:52:35,680
in most cases, because typically
you'll work out a deal to

949
00:52:35,680 --> 00:52:38,520
advertise their product and so
on.

950
00:52:39,000 --> 00:52:42,520
Many episodes.
OK, yeah, I think I'm, I'm

951
00:52:42,520 --> 00:52:45,640
following the most most of that.
I think there might be a typo or

952
00:52:45,640 --> 00:52:47,840
two that just threw me.
But yeah, I, I see what you're

953
00:52:47,840 --> 00:52:49,360
saying there, Nick.
OK.

954
00:52:49,360 --> 00:52:52,760
Go ahead, Jonathan.
Yeah, so I put a diagram in the

955
00:52:52,760 --> 00:52:56,560
Facebook group of that stack and
curiosity that was used in the

956
00:52:56,560 --> 00:52:58,760
podcast.
That the full thing, but I want

957
00:52:58,760 --> 00:53:00,520
to give you a structure that you
could use.

958
00:53:01,520 --> 00:53:07,360
So you would start your episode
with a question, then you'd

959
00:53:07,360 --> 00:53:09,640
provide special answer to that
question.

960
00:53:09,720 --> 00:53:12,360
And obviously I'm keeping this
very broad because I don't know

961
00:53:12,360 --> 00:53:16,120
what your topic is, but then you
then, you know, have a bigger

962
00:53:16,120 --> 00:53:17,800
question.
Now that bigger question is

963
00:53:17,800 --> 00:53:20,440
going to tie into that first
question.

964
00:53:20,800 --> 00:53:23,680
So it's just going to be a
slightly expanding on your your

965
00:53:23,680 --> 00:53:26,320
theme there a little bit.
And then you're going to do the

966
00:53:26,320 --> 00:53:28,520
investigation, which is what
they did here.

967
00:53:28,760 --> 00:53:31,240
They kind of had that first
question, they kind of got the

968
00:53:31,240 --> 00:53:34,040
answer, then they asked a bigger
question, then they did some

969
00:53:34,040 --> 00:53:36,720
investigation.
That investigation is going to

970
00:53:37,160 --> 00:53:40,600
reveal A partial answer, and
then you're going to then have

971
00:53:40,600 --> 00:53:43,760
an even bigger question that
you're going to continue as you

972
00:53:43,760 --> 00:53:45,080
bring.
And this is going to

973
00:53:45,560 --> 00:53:49,880
automatically bring people
through your podcast because our

974
00:53:49,880 --> 00:53:54,040
brain hates unfinished patterns.
When you don't give them an

975
00:53:54,040 --> 00:53:56,360
answer, something feels
unresolved.

976
00:53:56,760 --> 00:53:59,760
So the brain is going to keep
paying attention until it gets

977
00:53:59,760 --> 00:54:04,400
closure.
It's looking foreclosure, so you

978
00:54:04,400 --> 00:54:08,360
don't give it closure until the
end of your podcast or even the

979
00:54:08,360 --> 00:54:10,920
next podcast, depending on how
good you are at this and

980
00:54:10,920 --> 00:54:14,600
bringing them along and and you
know, creating curiosity for

981
00:54:14,600 --> 00:54:16,000
each topic that you're
discussing.

982
00:54:16,560 --> 00:54:20,760
Narrative podcasts like search
engine uses beautifully and you

983
00:54:20,760 --> 00:54:24,800
can as well just give it enough
resolution to make them

984
00:54:24,800 --> 00:54:27,600
satisfied, but not enough to get
them to stop listening.

985
00:54:27,840 --> 00:54:29,080
So.
I want to go to Nick.

986
00:54:29,320 --> 00:54:31,600
Do you want to share?
What you just said about this is

987
00:54:31,600 --> 00:54:34,840
what really pisses you off about
these types of shows.

988
00:54:34,840 --> 00:54:37,920
Do you want to share?
Yeah, what Jonathan's talking

989
00:54:37,920 --> 00:54:41,520
about right there is exactly
what drives me nuts about, like,

990
00:54:41,520 --> 00:54:44,200
true crime documentaries,
because usually they're covering

991
00:54:44,200 --> 00:54:46,760
cold cases and things like that.
So when you get to the end of

992
00:54:46,760 --> 00:54:50,560
it, you never have a resolution.
Yeah, you're always left like

993
00:54:50,560 --> 00:54:52,920
what happened?
Where what happens next?

994
00:54:52,920 --> 00:54:55,000
Like where's the next piece of
the story?

995
00:54:55,000 --> 00:54:58,240
And it just isn't.
Yeah, I'm totally with you

996
00:54:58,240 --> 00:55:00,480
there.
Yeah, I think like the Dateline

997
00:55:00,480 --> 00:55:02,440
shows is a great example of
that.

998
00:55:02,440 --> 00:55:05,520
I feel like they're always
unresolved, even from like 1

999
00:55:05,520 --> 00:55:09,240
commercial break to the next.
I feel like they kind of leave

1000
00:55:09,240 --> 00:55:12,240
you hanging and then you don't
really get the resolution that

1001
00:55:12,240 --> 00:55:14,760
you're looking for.
Yeah, I'm with you.

1002
00:55:15,160 --> 00:55:17,640
It's supposed to.
Yeah, that's the thing.

1003
00:55:17,640 --> 00:55:19,640
Yeah.
And and that's why people keep

1004
00:55:19,640 --> 00:55:22,280
coming back, I guess.
Mm hmm.

1005
00:55:22,480 --> 00:55:25,720
Yeah, that's what creates that
obsession, worthiness.

1006
00:55:25,720 --> 00:55:29,240
So don't explain a topic.
Follow the question and let the

1007
00:55:29,240 --> 00:55:33,840
story widen until it reveals A
deeper human emotion.

1008
00:55:33,840 --> 00:55:36,360
So thank you.
That would be the key take away.

1009
00:55:36,480 --> 00:55:39,400
Yeah, and I'll just share and
thank you Jonathan for another

1010
00:55:39,400 --> 00:55:41,720
breakdown.
I'll just share what really my

1011
00:55:41,720 --> 00:55:45,480
big takeaways are is just as far
as if, if you remember nothing

1012
00:55:45,480 --> 00:55:49,000
else from this particular
episode of Search Engine and

1013
00:55:49,000 --> 00:55:52,080
what we believe makes it so
obsession worthy.

1014
00:55:53,280 --> 00:55:55,720
Remember that in the beginning,
you want to start with a

1015
00:55:55,720 --> 00:55:58,360
feeling, right?
You want to really get your

1016
00:55:58,360 --> 00:56:01,000
audience to feel the feels with
you.

1017
00:56:01,480 --> 00:56:05,400
And then in the middle, let the
question get bigger, really

1018
00:56:05,400 --> 00:56:09,480
build on that question so that
people are really leaning in on

1019
00:56:09,480 --> 00:56:13,400
the edge of their seats and then
at the turn change what the

1020
00:56:13,400 --> 00:56:16,040
story is really about.
That's what they did so well

1021
00:56:16,040 --> 00:56:19,640
here is that we kind of expected
it to go one way, but it did go

1022
00:56:19,640 --> 00:56:22,760
in a different direction.
And the end revealed as

1023
00:56:22,760 --> 00:56:26,880
Jonathan's was been sharing
reveals that human, that deeper

1024
00:56:26,880 --> 00:56:31,520
human stake involved in it.
And for me, the lesson here is

1025
00:56:31,520 --> 00:56:34,280
curiosity keeps people
listening.

1026
00:56:34,400 --> 00:56:37,560
And that let's to to Nick's
point, which Nick said the last

1027
00:56:37,560 --> 00:56:40,880
thing he said in the chat after
he just shared with us is yet it

1028
00:56:40,880 --> 00:56:44,760
keeps me coming back, right?
It drives him crazy, but he

1029
00:56:44,760 --> 00:56:48,280
keeps coming back for more.
So, yeah, I think, Jonathan,

1030
00:56:48,280 --> 00:56:51,040
this was a great breakdown that
those are my takeaways, which

1031
00:56:51,040 --> 00:56:54,440
are very in line, I think, with
what you're saying as well.

1032
00:56:54,440 --> 00:56:58,160
So thank you for another great
episode and a reminder to

1033
00:56:58,160 --> 00:57:01,160
everybody that we are back
tomorrow, 7:00 AM Eastern,

1034
00:57:01,160 --> 00:57:06,280
answering your questions around
podcasting, content creation,

1035
00:57:06,280 --> 00:57:08,080
etcetera.
So bring your questions tomorrow

1036
00:57:08,080 --> 00:57:10,160
and we'll open up the mailbag as
well.

1037
00:57:10,400 --> 00:57:15,640
So join us at Podcasting Morning
chat.com/join us to find out all

1038
00:57:15,640 --> 00:57:18,240
the different ways you can catch
this show.

1039
00:57:18,480 --> 00:57:21,800
Until tomorrow.
Make it a great day everybody,

1040
00:57:21,920 --> 00:57:22,440
Take care.