You are viewing a single comment's thread from:

RE: Lion’s Den – September 20, 2024 [Full Episode w/ Subtitles & Transcript]

in Threespeak3 months ago

1
00:00:00,000 --> 00:00:13,520
Today's date is September 20th, 2024, and we will get started in a couple minutes.

2
00:00:30,000 --> 00:00:59,980
Thank you.

3
00:01:00,000 --> 00:01:00,620
you

4
00:01:30,000 --> 00:01:58,900
All right, everybody.

5
00:02:00,000 --> 00:02:20,020
Getting started on time today, and good to see some of the crowd showing up, and I am sure others will be floating in over the next few minutes as they start to see that we are live, get their reminders, and all kinds of good stuff.

6
00:02:21,020 --> 00:02:29,020
A couple things of housekeeping that Sunseeker put in the Threadcast.

7
00:02:30,000 --> 00:02:42,040
I just wanted to touch upon, really not touching upon what I'm going to go over, but this is basically Leo business, if you will.

8
00:02:42,840 --> 00:02:50,800
She said, not all feedback has to be an error. You can give suggestions well-formulated, even with models.

9
00:02:51,660 --> 00:02:57,860
Now, with models, I don't think she means in bikinis, but that's a valid point.

10
00:02:57,860 --> 00:02:59,860
The feedback tag can be used.

11
00:03:00,000 --> 00:03:03,740
If you have a suggestion, if you have an idea, certainly drop it.

12
00:03:03,820 --> 00:03:08,100
We know Khal reads them, and if it's a good idea, he'll add it to the list.

13
00:03:08,780 --> 00:03:13,820
Sunseeker goes on to say, feedback with the correct tag and all the info is so important and helpful.

14
00:03:14,300 --> 00:03:18,600
Imagine you were the one having to fix the problem when doing a report.

15
00:03:21,040 --> 00:03:23,680
And I see SimpleGame is listening.

16
00:03:24,100 --> 00:03:29,880
He is of the coding ilk, and I'm sure he can identify that.

17
00:03:30,160 --> 00:03:38,160
When people only give use a little bit of the information about the bugs, it's like, okay, well, what the hell are you talking about?

18
00:03:38,300 --> 00:03:52,960
So, again, with feedback, the more you give, especially if you give screenshots, here's a use case, maybe record a short.

19
00:03:53,580 --> 00:03:57,020
If you're on your laptop, I've done this before where I'm having an issue.

20
00:03:57,140 --> 00:03:58,880
I'll take out my phone and I'll record.

21
00:03:59,280 --> 00:03:59,980
What is good?

22
00:04:00,040 --> 00:04:01,440
What's going on on my laptop?

23
00:04:02,300 --> 00:04:05,160
And then I'll include that with the feedback tag.

24
00:04:06,160 --> 00:04:08,380
That way they can see what's going on.

25
00:04:09,200 --> 00:04:14,260
Something simple, something easy, but that at least gives them a direction to look.

26
00:04:14,260 --> 00:04:20,620
So, those are a couple suggestions from Sunseeker, which are valid suggestions.

27
00:04:21,760 --> 00:04:25,000
So, please pay attention.

28
00:04:25,240 --> 00:04:28,920
And, yes, consider if you were on the other side.

29
00:04:28,920 --> 00:04:37,140
I mean, when you just simply say site not working feedback, what's a developer read that's supposed to do?

30
00:04:39,320 --> 00:04:42,060
I mean, can you narrow it down?

31
00:04:43,840 --> 00:04:48,380
So, the more we give, the better the results will be.

32
00:04:51,100 --> 00:04:53,200
Yes, Brando, there is the link.

33
00:04:53,200 --> 00:04:58,200
Can somebody drop the link in the Threadcast?

34
00:04:58,920 --> 00:05:03,400
Last to the end space, so Brando and some of the others.

35
00:05:03,800 --> 00:05:10,020
It's already I put it down lower, but I guess I could probably bring it back up to the top.

36
00:05:10,200 --> 00:05:10,680
Okay.

37
00:05:22,860 --> 00:05:24,060
With that done.

38
00:05:24,080 --> 00:05:25,260
I can turn the That's it.

39
00:05:25,280 --> 00:05:26,820
Every Alphabet Awesome question, any time, please.

40
00:05:26,820 --> 00:05:27,700
Yes, thank You.

41
00:05:27,700 --> 00:05:28,200
Let's have some questions.

42
00:05:28,200 --> 00:05:46,140
all right I dropped it in there and MoreTea did too Walter J did too so Brando if you

43
00:05:46,140 --> 00:05:57,360
can't find it it's there three times all right what I want to discuss today and this is kind

44
00:05:57,360 --> 00:06:07,860
of an extension of what we talked about last week it is also something that we discussed Lisa

45
00:06:07,860 --> 00:06:18,900
and Gentile and myself a show we did on Monday and this I think is one of the most important

46
00:06:18,900 --> 00:06:24,920
things that we have going right now a lot of you have heard the

47
00:06:27,360 --> 00:06:35,340
discussion around AI the excitement around AI the debate around AI

48
00:06:37,320 --> 00:06:43,680
and in my time watching technology and technological Trends which I can tell you

49
00:06:43,680 --> 00:06:51,000
I started this in 1999 this is the fastest moving most impactful thing I've ever seen

50
00:06:51,840 --> 00:06:56,940
this is even faster moving than the smartphone which was pretty fast moving

51
00:06:56,940 --> 00:06:57,340
than the United States

52
00:06:57,360 --> 00:07:06,540
considering that went from pretty much zero market share to over 70 in a little over six years

53
00:07:07,260 --> 00:07:10,320
that's pretty fast and this is going a lot quicker

54
00:07:11,580 --> 00:07:18,900
and so what I think and this is credit to Nifty because he kind of dropped this

55
00:07:19,740 --> 00:07:27,060
moniker if you will I didn't think of it in these terms but he said LeoAI as a service

56
00:07:27,920 --> 00:07:36,960
laptop something like that KR L

57
00:07:39,000 --> 00:07:45,360
software used to be something you bought And it used to come on a disc and you would load

58
00:07:45,360 --> 00:07:52,620
it on your computer and then it was also something in the internet age you would download you pay for

59
00:07:52,620 --> 00:07:55,260
it you download it it's now on your computer well software as a service then advance for a lifetime

60
00:07:55,260 --> 00:07:57,260
okay now that's kind of a spin on software-as-a-service and the community has been involved in

61
00:07:57,260 --> 00:08:07,340
service became something completely different and suddenly companies were paying for their enterprise

62
00:08:07,340 --> 00:08:13,580
software people were used up in the cloud and it wasn't something you owned it was something you

63
00:08:13,580 --> 00:08:20,620
used it was a service and you paid for and there were different models of pricing and people paid

64
00:08:20,620 --> 00:08:29,180
based upon seeds and things of that name that these companies had had developed and so this

65
00:08:29,180 --> 00:08:47,260
is an extension of that idea and my theory is as an owner social media is dead as a destination

66
00:08:48,220 --> 00:08:49,980
social media is dead

67
00:08:50,620 --> 00:08:59,180
as a product again I say as an owner so as an owner we're looking at google

68
00:08:59,180 --> 00:09:07,180
we're looking at xai we're looking or x we're looking at meta we are looking at the entities

69
00:09:07,180 --> 00:09:15,260
that own these platforms what are they looking at and it may be the same thing they've always

70
00:09:15,260 --> 00:09:19,420
been looking at but it's on steroids they're looking at data

71
00:09:21,340 --> 00:09:24,940
social media is a pathway to more data

72
00:09:27,260 --> 00:09:35,260
exclamation point people show up each day to social media x youtube facebook instagram

73
00:09:35,260 --> 00:09:41,420
whatever it is snapchat whatsapp uh tick tock and they provide data

74
00:09:44,460 --> 00:09:50,140
this treasure in the modern era and they just provide it to these

75
00:09:50,700 --> 00:09:54,460
big tech companies the big tech companies

76
00:09:58,300 --> 00:10:03,180
are very appreciative I'm sure even though they don't send out thank you notes

77
00:10:04,780 --> 00:10:07,660
because you are feeding their coffers

78
00:10:08,860 --> 00:10:14,540
and the reason why you are feeding their coffers and why they are so

79
00:10:14,540 --> 00:10:16,540
so

80
00:10:20,620 --> 00:10:28,940
they are so benefited from it is what I'm posting in the room in the Threadcast now

81
00:10:31,500 --> 00:10:32,940
the data feedback loop

82
00:10:35,180 --> 00:10:42,620
this is the modern digital platform model in my opinion I've thought a lot about this

83
00:10:43,420 --> 00:10:49,420
and it's very simple the data that people place in social media

84
00:10:50,620 --> 00:11:01,980
the neural network it's used to train the models from these models these companies these companies

85
00:11:01,980 --> 00:11:12,060
that own these digital platforms they create and will be creating more ai features that they are

86
00:11:12,060 --> 00:11:17,500
going to be placing ai features and services that they will be placing on their platforms

87
00:11:20,620 --> 00:11:28,160
which generate more data and as people utilize them they will create more data

88
00:11:28,160 --> 00:11:33,720
and that data is then fed back into the neural network

89
00:11:33,720 --> 00:11:41,780
to train further models and it's a rinse and repeat method and I'll give you an example

90
00:11:46,640 --> 00:11:50,600
youtube youtube is still by far the most

91
00:11:50,600 --> 00:11:57,480
dominant video platform. There are some very successful channels on YouTube, Mr. Beast being

92
00:11:57,480 --> 00:12:04,080
the big one. But there's a lot of people who have pretty big followings and done very well

93
00:12:04,080 --> 00:12:14,780
in different niches. And they do fantastic. Google has added YouTube

94
00:12:14,780 --> 00:12:27,580
a community feature which allows these content creators to engage with their audience.

95
00:12:29,720 --> 00:12:40,520
Why? What is it like? They are targeting Discord. They are targeting, perhaps, Reddit.

96
00:12:41,500 --> 00:12:44,760
They are trying to keep those people who are on YouTube, who are on YouTube,

97
00:12:44,760 --> 00:12:50,280
conversations that happen between the content creators and their audience from

98
00:12:50,280 --> 00:12:55,440
help happening elsewhere now obviously you could say well they want time on

99
00:12:55,440 --> 00:13:03,100
site they want they want that and that's probably true but let me ask you how

100
00:13:03,100 --> 00:13:15,480
much data does discord generate that is what Google is after they want people

101
00:13:15,480 --> 00:13:24,460
talking on their application because any conversation that takes place with any

102
00:13:24,460 --> 00:13:32,560
of these content creators is our on Google server more data so they bring

103
00:13:32,560 --> 00:13:33,060
out a social media site and they put it on their website and they put it on their

104
00:13:33,060 --> 00:13:33,080
social media site and they put it on their social media site and they put it on their

105
00:13:33,080 --> 00:13:37,760
social media font feature which is okay for the users hey it has utility it's a

106
00:13:37,760 --> 00:13:44,500
social media feature what is Google's goal more data

107
00:13:45,880 --> 00:13:50,620
more data and you could say well they want the advertising yes they do they're

108
00:13:50,620 --> 00:13:56,300
so big advertising company but you know what the money they are going to make

109
00:13:56,300 --> 00:14:03,040
off Gemini most likely will to walk their advertising record

110
00:14:03,040 --> 00:14:03,160
off Gemini most likely will to walk their advertising record

111
00:14:03,160 --> 00:14:06,800
off Gemini most likely will to walk their advertising record

112
00:14:06,800 --> 00:14:11,000
open AI is going for another round of funding their

113
00:14:11,000 --> 00:14:11,040
open AI is going for another round of funding their

114
00:14:11,040 --> 00:14:18,680
open AI is going for another round of funding their valuation is 150 billion dollars

115
00:14:18,680 --> 00:14:18,700
valuation is 150 billion dollars

116
00:14:18,700 --> 00:14:20,760
valuation is 150 billion dollars

117
00:14:20,760 --> 00:14:23,160
think of that

118
00:14:23,160 --> 00:14:27,620
is that based on their revenue you know

119
00:14:27,620 --> 00:14:27,640
is that based on their revenue you know

120
00:14:27,640 --> 00:14:29,400
is that based on their revenue you know is it based on their profitability know

121
00:14:29,400 --> 00:14:29,420
is it based on their profitability know

122
00:14:29,420 --> 00:14:31,640
is it based on their profitability know they're losing an estimated like eight

123
00:14:31,640 --> 00:14:31,660
they're losing an estimated like eight

124
00:14:31,660 --> 00:14:32,720
they're losing an estimated like eight billion dollars a yeardar's worth but

125
00:14:32,720 --> 00:14:32,740
billion dollars a yeardar's worth but

126
00:14:32,740 --> 00:14:35,500
a year or something.

127
00:14:35,500 --> 00:14:39,620
It's based on the fact that they have these models, and they have other products in the

128
00:14:39,620 --> 00:14:43,480
pipeline like Sora and some of these other things that they're going to be rolling out

129
00:14:43,480 --> 00:14:45,160
over time.

130
00:14:45,160 --> 00:14:47,220
And it's going to be enormous.

131
00:14:47,220 --> 00:14:53,240
And one of the reasons from Leo's perspective that I keep harping upon this is if you want

132
00:14:53,240 --> 00:15:00,840
$100, you want $150 million, you want a $200 million token market cap value, I don't think

133
00:15:00,840 --> 00:15:02,640
you get it there with social media.

134
00:15:02,740 --> 00:15:09,960
Because people who are on social media are happy on social media.

135
00:15:09,960 --> 00:15:16,880
And we can con ourselves into believing, oh, they'll leave Twitter and come to Leo.

136
00:15:16,880 --> 00:15:20,980
They'll leave Medium and post on Leo.

137
00:15:20,980 --> 00:15:23,160
No.

138
00:15:23,160 --> 00:15:25,960
We've tried that for years.

139
00:15:25,960 --> 00:15:28,300
You can talk about censorship-resistant.

140
00:15:28,300 --> 00:15:30,860
You can talk about immutable data.

141
00:15:30,860 --> 00:15:32,700
Most people don't know.

142
00:15:32,700 --> 00:15:33,740
They don't care.

143
00:15:33,740 --> 00:15:39,240
In fact, over the years, we see people on Hive bitching and moaning when Twitter closed

144
00:15:39,240 --> 00:15:42,700
their account, or YouTube closed their account, or whoever closed their account.

145
00:15:42,700 --> 00:15:49,820
And they stayed on Hive, posted on Hive, till the second their account was reactivated.

146
00:15:49,820 --> 00:15:50,820
And then what happens?

147
00:15:50,820 --> 00:15:53,140
They're right back to YouTube.

148
00:15:53,140 --> 00:15:54,140
No more three-speed.

149
00:15:54,140 --> 00:16:01,300
Now, I'm not saying that we don't want to draw people with the social media to Leo.

150
00:16:01,300 --> 00:16:02,220
But again, we are owners.

151
00:16:02,220 --> 00:16:08,340
If you have the Leo token, you are an owner.

152
00:16:08,340 --> 00:16:09,480
You are not a user.

153
00:16:09,480 --> 00:16:13,600
You are an owner.

154
00:16:13,600 --> 00:16:20,740
It's a much different perspective.

155
00:16:20,740 --> 00:16:25,260
Another example about the data feedback loop.

156
00:16:25,260 --> 00:16:28,180
Alibaba.

157
00:16:28,180 --> 00:16:31,260
Chinese version of Amazon.

158
00:16:31,260 --> 00:16:35,840
Amazon slash some of the other companies, because Alibaba does a lot of different things.

159
00:16:35,840 --> 00:16:38,620
Alibaba has a cloud service.

160
00:16:38,620 --> 00:16:48,200
They just released over 100 open source AI applications.

161
00:16:48,200 --> 00:16:58,140
And they're hoping that this will stimulate companies and people around the world to use

162
00:16:58,140 --> 00:16:59,380
Alibaba's cloud.

163
00:16:59,380 --> 00:17:00,380
All right.

164
00:17:00,380 --> 00:17:00,880
Thank you.

165
00:17:01,260 --> 00:17:05,940
Now, obviously, it's a revenue generator when you get into enterprise and stuff like

166
00:17:05,940 --> 00:17:08,140
that.

167
00:17:08,140 --> 00:17:15,180
But what about all the free space that these apps or these cloud services give?

168
00:17:15,180 --> 00:17:17,720
What about like Google Photos or whatever the hell it's called?

169
00:17:17,720 --> 00:17:22,020
And all these people loading up Google and putting their photos on Google.

170
00:17:22,020 --> 00:17:26,200
They give them two gigabytes or five gigabytes or whatever the hell they give for free.

171
00:17:26,200 --> 00:17:27,200
Right?

172
00:17:27,200 --> 00:17:29,260
What do you think happens with those photos?

173
00:17:29,260 --> 00:17:30,260
You don't think they're useful?

174
00:17:30,260 --> 00:17:31,260
You don't think they're useful?

175
00:17:31,260 --> 00:17:32,260
You don't think they're useful?

176
00:17:32,260 --> 00:17:33,260
They don't have the tools to train?

177
00:17:33,260 --> 00:17:39,140
You put stuff on their server they have access to.

178
00:17:39,140 --> 00:17:43,540
And that's what they want, more data.

179
00:17:43,540 --> 00:17:51,340
So they go and design all these AI functions, all these AI features that, by the way, are

180
00:17:51,340 --> 00:17:55,680
on their system to get people to utilize it.

181
00:17:55,680 --> 00:17:58,320
And what happens when people use it?

182
00:17:58,320 --> 00:17:59,140
More data is created.

183
00:17:59,140 --> 00:18:00,140
That's the goal.

184
00:18:00,140 --> 00:18:01,140
That's the goal.

185
00:18:01,140 --> 00:18:02,300
That's the feedback loop.

186
00:18:02,540 --> 00:18:04,340
And it actually accelerates.

187
00:18:04,560 --> 00:18:17,280
And one of the reasons why I'm so keen on LeoAI and why I've hammered this time and time and time again is you get a LeoAI chatbot.

188
00:18:18,060 --> 00:18:24,300
Even if it's not posted on the blockchain, it's still on Leo's server.

189
00:18:24,960 --> 00:18:26,780
It's still data for Leo.

190
00:18:26,780 --> 00:18:33,880
So every time somebody prompts a chatbot, more information is created.

191
00:18:35,720 --> 00:18:37,940
That information is fed back in.

192
00:18:38,920 --> 00:18:43,040
So even though it's synthetic, it's fed right back in.

193
00:18:44,860 --> 00:18:48,640
Now, what is the advantage to Leo?

194
00:18:51,080 --> 00:18:54,180
We can take that data and post it on the blockchain.

195
00:18:54,180 --> 00:18:55,900
First off, it makes it public.

196
00:18:56,780 --> 00:18:59,280
Because now it's part of the high blockchain.

197
00:18:59,880 --> 00:19:00,860
Anybody can see it.

198
00:19:00,900 --> 00:19:01,620
It's not only Leo.

199
00:19:03,360 --> 00:19:04,580
But what can happen?

200
00:19:05,480 --> 00:19:09,420
People can actually interact with that data, with that information.

201
00:19:09,860 --> 00:19:11,140
They can comment on it.

202
00:19:11,900 --> 00:19:13,100
They can respond to it.

203
00:19:13,540 --> 00:19:14,580
Say what they agree with.

204
00:19:14,640 --> 00:19:16,140
They can say what they disagree with.

205
00:19:16,240 --> 00:19:17,760
They can say where it's incorrect.

206
00:19:17,940 --> 00:19:20,960
They can feed back the static data.

207
00:19:20,960 --> 00:19:26,740
And what you have now is you have synthetic and human data combined.

208
00:19:26,780 --> 00:19:31,080
To me, that's more powerful.

209
00:19:33,900 --> 00:19:35,700
And that really helps the knowledge graph.

210
00:19:36,120 --> 00:19:40,400
And if you dig in a little to knowledge graph, which is part of data science,

211
00:19:40,800 --> 00:19:53,320
and vector databases is what AI uses, it all ties in how the models start to connect information

212
00:19:54,200 --> 00:19:56,660
or connect data and spit out information.

213
00:19:57,020 --> 00:19:59,480
And it's a proper way of phrasing it.

214
00:20:05,480 --> 00:20:06,640
And this is huge.

215
00:20:08,780 --> 00:20:16,240
And one thing I learned, and Khal commented on this the other day on the AMA,

216
00:20:17,980 --> 00:20:20,480
or on the marketing call, excuse me.

217
00:20:20,720 --> 00:20:26,480
I made a reference that Khal was the only one to mention AI in any of the presentations I watched.

218
00:20:27,340 --> 00:20:29,380
He was the only one.

219
00:20:32,560 --> 00:20:34,540
Which tells me nobody is doing this.

220
00:20:35,500 --> 00:20:40,460
And that was validated because Khal said he met with a couple people after he mentioned that.

221
00:20:40,800 --> 00:20:48,400
And he explained what he was doing and how we are, how Leo is taking the high blockchain data and training model.

222
00:20:49,780 --> 00:20:51,880
And people say, I hope somebody would have done that.

223
00:20:51,880 --> 00:21:00,960
LEO's going to stand out because this is not a cheap process and Khal said that it's very

224
00:21:00,960 --> 00:21:05,200
expensive and you need you need to have the right people to do this

225
00:21:05,200 --> 00:21:10,940
and uh I guess since the model's being trained evidently they do

226
00:21:10,940 --> 00:21:15,480
but as a little side note just to give you an idea

227
00:21:15,480 --> 00:21:22,380
of what some of the potentiality is and this isn't even talking in the data sense how

228
00:21:22,380 --> 00:21:25,760
I'm referring to it but it all feeds in together

229
00:21:25,760 --> 00:21:32,660
a company let's say a crypto company wants to advertise on LEO

230
00:21:32,660 --> 00:21:42,520
and we all know the the LEO ad platform's being worked on so or we were told so

231
00:21:42,520 --> 00:21:45,340
a crypto company wants to advertise

232
00:21:45,480 --> 00:21:50,800
do they just throw up a banner and pay whatever and there it is

233
00:21:50,800 --> 00:22:02,120
this is what LeoAI can do LeoAI can start to draw a list

234
00:22:02,120 --> 00:22:06,120
of those people who regularly talk about bitcoin

235
00:22:06,120 --> 00:22:09,300
who talk about crypto who post about crypto

236
00:22:09,300 --> 00:22:13,320
if you post about food and you don't post about crypto

237
00:22:13,320 --> 00:22:15,460
you are probably not going to be able to do that

238
00:22:15,460 --> 00:22:15,480
you are probably not going to be able to do that

239
00:22:15,480 --> 00:22:16,760
you are probably not going to be able to do that

240
00:22:16,760 --> 00:22:21,480
part of their target audience you might be

241
00:22:21,480 --> 00:22:28,040
but from a statistical standpoint a probability

242
00:22:28,040 --> 00:22:30,600
standpoint usually the people who are talking about

243
00:22:30,600 --> 00:22:35,080
crypto are the ones who are interested in crypto

244
00:22:35,800 --> 00:22:39,320
those are the people who are going to be targeted with the ads

245
00:22:39,320 --> 00:22:45,460
which of course as an owner and many people who've been on LEO said

246
00:22:45,460 --> 00:22:48,560
I miss the ads because you know damn well as an owner

247
00:22:48,560 --> 00:22:52,500
that's to our benefit but even as a user

248
00:22:52,500 --> 00:22:56,400
I mean if you're a guy

249
00:22:56,400 --> 00:23:05,580
do you want ads about the barbie movie now you may be into that type of thing

250
00:23:05,580 --> 00:23:09,120
but odds are you're not that's not going to be necessarily your forte

251
00:23:09,120 --> 00:23:12,860
it's not going to be your deal so if you're going to see ads you might want

252
00:23:12,860 --> 00:23:15,380
to see ads at least of stuff that you could be

253
00:23:15,380 --> 00:23:20,620
conceivably interested in and of course the youtubes and the or google

254
00:23:20,620 --> 00:23:27,240
meta they've spent who knows how much money fine-tuning all this stuff in their

255
00:23:27,240 --> 00:23:33,600
algorithms over the years but this is part of LeoAI but how does

256
00:23:33,600 --> 00:23:38,700
that work how does that happen how does LEO as a platform offer that

257
00:23:38,700 --> 00:23:44,460
ability to the advertisers because we put in data

258
00:23:45,380 --> 00:23:50,460
because the data was put onto the blockchain

259
00:23:50,460 --> 00:23:58,760
and that is enormous

260
00:23:58,760 --> 00:24:12,560
but again it all comes back to that initial thrust of data here's another just side bit with this idea

261
00:24:15,380 --> 00:24:21,360
2015 open ai was started by I think elon moss altman and

262
00:24:21,360 --> 00:24:25,100
what was the other guy's name brockman or somebody else but anyway

263
00:24:25,100 --> 00:24:28,900
open ai was started in 2015

264
00:24:28,900 --> 00:24:34,060
now I don't know how they proceeded along

265
00:24:34,060 --> 00:24:42,060
they hired some people they had people who coded they bought computers they bought servers

266
00:24:42,060 --> 00:24:45,340
they wrote algorithms they did all this good stuff right

267
00:24:45,380 --> 00:24:55,820
what they have zip you are not doing anything in ai without data so open ai had to go get data

268
00:24:55,820 --> 00:25:01,640
and we now know they went and scraped every site that they could scrape they sent out robot.txt

269
00:25:01,640 --> 00:25:09,020
crawler and crawled whatever they could and now open ai is being sued by everybody and their

270
00:25:09,020 --> 00:25:15,100
brother because they didn't have any of their data on their own but they had to go get data before

271
00:25:15,380 --> 00:25:18,200
they could do anything

272
00:25:18,200 --> 00:25:22,580
write all the algos in the world if you have no data

273
00:25:22,580 --> 00:25:25,500
you have nothing to put into the compute

274
00:25:25,500 --> 00:25:27,840
and it's the same thing here

275
00:25:27,840 --> 00:25:30,440
we need the data

276
00:25:30,440 --> 00:25:31,780
LEO has

277
00:25:31,780 --> 00:25:37,200
a social media platform where we can put data in

278
00:25:37,200 --> 00:25:41,440
we can generate it we can keep feeding it

279
00:25:41,440 --> 00:25:44,380
and so this is an important

280
00:25:44,380 --> 00:25:45,380
important feature

281
00:25:45,380 --> 00:25:53,240
brando asked what gives LeoAI that competitive edge

282
00:25:53,240 --> 00:25:56,040
very simple brando

283
00:25:56,040 --> 00:25:58,280
it is

284
00:25:58,280 --> 00:26:01,280
the only

285
00:26:01,280 --> 00:26:05,040
model that I know of

286
00:26:05,040 --> 00:26:07,040
that will be trained

287
00:26:07,040 --> 00:26:09,000
on everything

288
00:26:09,000 --> 00:26:10,460
on hot

289
00:26:10,460 --> 00:26:15,060
and

290
00:26:15,380 --> 00:26:16,620
covid

291
00:26:16,620 --> 00:26:22,640
episc

292
00:26:22,640 --> 00:26:26,480
pigs

293
00:26:26,480 --> 00:26:27,340
roe

294
00:26:27,340 --> 00:26:30,120
bear

295
00:26:30,120 --> 00:26:32,780
they over

296
00:26:32,780 --> 00:26:34,880
I doubt at this point anybody has scraped the Hive database

297
00:26:34,880 --> 00:26:40,040
now

298
00:26:40,040 --> 00:26:44,120
huh data is incorporated in other models because you can go to meta and you can video

299
00:26:44,120 --> 00:26:49,720
and you will get information on this high block not the high blockchain mining company you also

300
00:26:49,720 --> 00:26:56,120
get that so if you are specific it has the information in there where do you get that

301
00:26:56,120 --> 00:27:04,600
information hard to know the source but I would believe a lot of the data came from indexed pages

302
00:27:04,600 --> 00:27:20,440
from Hive so from peakd hot blog pages that were indexed to google you know what's not indexed threads

303
00:27:24,440 --> 00:27:30,280
that's not in there you know what's also not on the high blockchain images

304
00:27:33,080 --> 00:27:34,440
you know it's also not on there

305
00:27:34,920 --> 00:27:39,800
video now at this point in time I don't believe

306
00:27:41,480 --> 00:27:48,840
I could be mistaken but I don't believe that LeoAI is being trained on any of the images

307
00:27:48,840 --> 00:27:55,640
on the image keep them LEO has its own image server so any image you upload through LEO

308
00:27:56,440 --> 00:28:04,040
is on that image but I don't believe because of the cost that images or video are being used to

309
00:28:05,320 --> 00:28:11,080
I could be mistaken and they could be training on images too but I I would say that they're not

310
00:28:12,440 --> 00:28:18,120
the key is at this time over the next year to year and a half

311
00:28:19,080 --> 00:28:26,360
the cost of computer will continue to come down what we are probably seeing if I had a guess

312
00:28:26,360 --> 00:28:34,040
InLeo ai's training based upon what I know about the cost you're probably looking at about a chatgpt

313
00:28:34,600 --> 00:28:39,000
in the so this is a link in your environment or on youtube if you want to look at yourself

314
00:28:39,560 --> 00:28:53,320
your3614b of the model right here when you create a how I kalau

315
00:28:53,320 --> 00:28:59,400
is distributed to this emily peter I'm one of my best messengers but if you're watching from my

316
00:28:59,400 --> 00:29:00,840
home market I'd recommend that you open up my он reca k20 on 내일

317
00:29:00,840 --> 00:29:04,500
I checked out from da alمكن and they're

318
00:29:04,500 --> 00:29:12,200
In terms of the training of Hive data,

319
00:29:12,200 --> 00:29:17,920
they're using slower processing because Leo

320
00:29:17,920 --> 00:29:25,200
doesn't have the resources to go out and get 10,000 GPUs.

321
00:29:25,200 --> 00:29:27,300
It's not cheap to do this.

322
00:29:27,300 --> 00:29:33,800
So the fact that Leo is going to be

323
00:29:33,800 --> 00:29:37,820
able to answer questions about anything that's

324
00:29:37,820 --> 00:29:43,100
put on the Hive Blockchain and do it almost instantaneously

325
00:29:43,100 --> 00:29:45,320
is a big competitive edge.

326
00:29:45,320 --> 00:29:49,460
So you are going to be able to go and ask

327
00:29:49,460 --> 00:29:55,700
AI about your voting, or about who voted on your stuff,

328
00:29:55,700 --> 00:30:01,580
or about which post got the most comments, which post

329
00:30:01,580 --> 00:30:03,700
got the most upvotes.

330
00:30:03,700 --> 00:30:09,460
You can't ask that of Llama from Meta.

331
00:30:09,460 --> 00:30:12,800
Meta may have some of the articles that were indexed,

332
00:30:12,800 --> 00:30:19,940
but it's not looking at voting on the Hive Blockchain

333
00:30:19,940 --> 00:30:21,740
because it doesn't care.

334
00:30:21,740 --> 00:30:24,380
LeoAI does.

335
00:30:24,380 --> 00:30:27,460
And as Khal said with the advertising,

336
00:30:27,460 --> 00:30:33,500
it gives the ability to have each person create a profile.

337
00:30:33,700 --> 00:30:36,840
Well, LeoAI can create a profile for each person

338
00:30:36,840 --> 00:30:40,180
based upon what engagement and what you put on there.

339
00:30:40,180 --> 00:30:44,840
So if you're constantly talking about sports, and let's say EPL,

340
00:30:44,840 --> 00:30:46,300
and you're always talking about that,

341
00:30:46,300 --> 00:30:48,060
and you're posting about that, it's

342
00:30:48,060 --> 00:30:50,520
going to create a profile around that,

343
00:30:50,520 --> 00:30:55,280
knowing that you are big into the EPL.

344
00:30:55,280 --> 00:30:57,540
You are big into sports.

345
00:30:57,540 --> 00:30:58,220
Well, guess what?

346
00:30:58,220 --> 00:31:00,780
If there's an advertiser who is advertising something

347
00:31:00,780 --> 00:31:03,700
to do with sports, you may end up seeing that.

348
00:31:03,700 --> 00:31:10,300
And this is what Grok does, by the way.

349
00:31:10,300 --> 00:31:13,660
Anybody who has premium Twitter subscription,

350
00:31:13,660 --> 00:31:18,000
go and type your username into Grok, your Twitter handle.

351
00:31:18,000 --> 00:31:23,620
It'll give you a profile based upon what you've done on there.

352
00:31:23,620 --> 00:31:26,740
And the more active you are, the more accurate it is.

353
00:31:26,740 --> 00:31:28,960
I mean, obviously, over 10 years,

354
00:31:28,960 --> 00:31:31,480
you've done like 100 tweets.

355
00:31:31,480 --> 00:31:33,700
OK, it doesn't have a lot of data to compile on.

356
00:31:33,700 --> 00:31:38,080
I mean, it'll give you something,

357
00:31:38,080 --> 00:31:41,580
but it's like not much in there.

358
00:31:41,580 --> 00:31:42,700
And it's the same thing on Leo.

359
00:31:46,720 --> 00:31:48,880
But again, you have to put in the data,

360
00:31:48,880 --> 00:31:52,900
and the data has to be in text form.

361
00:31:52,900 --> 00:31:57,580
And so this brings up the next question of what kind of data?

362
00:32:03,700 --> 00:32:11,660
There's two things about posting on threads, in my view.

363
00:32:11,660 --> 00:32:14,800
The first thing is, you post something, you want engagement.

364
00:32:14,800 --> 00:32:17,420
You want others to see it, you want others to engage.

365
00:32:17,420 --> 00:32:17,960
It's a meme.

366
00:32:17,960 --> 00:32:21,180
You find it funny, you think other people find it funny.

367
00:32:21,180 --> 00:32:22,740
OK, wonderful.

368
00:32:22,740 --> 00:32:24,320
Doesn't do anything for the database,

369
00:32:24,320 --> 00:32:27,120
or doesn't do much for database, but you think it's funny.

370
00:32:27,120 --> 00:32:30,580
You post a short off YouTube because you think it's funny.

371
00:32:30,580 --> 00:32:33,680
Some guy riding a bicycle, and then he rides into a garbage

372
00:32:33,680 --> 00:32:34,180
can.

373
00:32:34,180 --> 00:32:36,280
So you post that, you think it'll be funny.

374
00:32:36,280 --> 00:32:37,600
You want other people to see it.

375
00:32:37,600 --> 00:32:42,140
OK, so you're sharing content that you think others will like,

376
00:32:42,140 --> 00:32:45,540
and maybe others can comment on, engage in, whatever.

377
00:32:45,540 --> 00:32:48,520
That's the first thing, engagement data,

378
00:32:48,520 --> 00:32:50,100
engagement content.

379
00:32:50,100 --> 00:32:54,780
The second thing is database content.

380
00:32:54,780 --> 00:32:57,820
This is totally different.

381
00:32:57,820 --> 00:33:02,740
A YouTube short, or worse, a YouTube full-length video

382
00:33:02,740 --> 00:33:03,660
doesn't do anything.

383
00:33:03,680 --> 00:33:11,700
Because all it is, is it's a stream in.

384
00:33:11,700 --> 00:33:14,720
It's not even, it's an embed.

385
00:33:14,720 --> 00:33:19,280
The video is not even on Leo or on IPFS.

386
00:33:19,280 --> 00:33:22,440
It's just an embed from YouTube.

387
00:33:22,440 --> 00:33:25,840
So you'll notice that at least once a day,

388
00:33:25,840 --> 00:33:27,740
I'll put up a YouTube video.

389
00:33:27,740 --> 00:33:30,320
I'll go grab the transcript from YouTube.

390
00:33:30,320 --> 00:33:33,200
I'll feed that through a chat bot and get a summary.

391
00:33:33,200 --> 00:33:33,600
OK.

392
00:33:33,600 --> 00:33:35,700
And then post the summary in the threads

393
00:33:35,700 --> 00:33:38,400
underneath the video.

394
00:33:38,400 --> 00:33:42,960
Because now, that's part of the ..

395
00:33:42,960 --> 00:33:47,100
So if you're sitting there, and you're just re-threading stuff,

396
00:33:47,100 --> 00:33:48,860
that's adding to the database.

397
00:33:48,860 --> 00:33:53,640
If you're just sitting there throwing up YouTube links,

398
00:33:53,640 --> 00:33:56,640
that's not adding to the database.

399
00:33:56,640 --> 00:33:59,880
I mean, now, if you put up something somebody might enjoy,

400
00:33:59,880 --> 00:34:01,600
OK, that's fine.

401
00:34:01,600 --> 00:34:02,600
But let's be honest.

402
00:34:02,600 --> 00:34:03,440
How many people?

403
00:34:03,440 --> 00:34:06,980
If you put up six one-hour-long videos,

404
00:34:06,980 --> 00:34:09,020
who the hell is going to watch that?

405
00:34:09,020 --> 00:34:11,980
I mean, shorts might get some views.

406
00:34:11,980 --> 00:34:16,100
I've seen people put up shorts and stuff like that.

407
00:34:16,100 --> 00:34:18,100
If I'm interested in by the title, I'll click on it.

408
00:34:18,100 --> 00:34:21,200
I'll watch it for 30, 45 seconds, whatever it is.

409
00:34:21,200 --> 00:34:22,260
And it's a YouTube embed.

410
00:34:22,260 --> 00:34:23,440
It's not even LeoShorts.

411
00:34:23,440 --> 00:34:24,440
It's a YouTube embed.

412
00:34:28,840 --> 00:34:29,960
And then tagging.

413
00:34:29,960 --> 00:34:33,240
Tagging is very important, because when we deal with the knowledge,

414
00:34:33,240 --> 00:34:35,900
graph, this is how things tie in.

415
00:34:35,900 --> 00:34:36,820
But what do people do?

416
00:34:36,820 --> 00:34:38,920
Try and scout tokens.

417
00:34:38,920 --> 00:34:43,240
If you take a sports video, or if you take a post about sports,

418
00:34:43,240 --> 00:34:47,400
and you use a tag like free compliments,

419
00:34:47,400 --> 00:34:51,720
and then you use a tag of free compliments for something

420
00:34:51,720 --> 00:34:55,480
for food, and then you use the tag for free compliments,

421
00:34:55,480 --> 00:34:57,480
and I'm using this because it's a top-trend in tech,

422
00:34:57,480 --> 00:35:02,620
so I'm just looking at it, for something to do with crypto, what the hell?

423
00:35:02,620 --> 00:35:06,080
What the hell are you doing?

424
00:35:06,080 --> 00:35:14,360
You're trying to scout whatever, but it's not training the AI on anything that matters.

425
00:35:14,360 --> 00:35:24,100
So now you have free compliments tied to data that has nothing to do with each other.

426
00:35:24,100 --> 00:35:25,700
And we see this all over the place.

427
00:35:25,700 --> 00:35:28,560
We see people tag spamming.

428
00:35:28,560 --> 00:35:31,280
They put up a video, and they put 10 fucking tags underneath.

429
00:35:31,280 --> 00:35:32,320
What the hell are you doing?

430
00:35:32,320 --> 00:35:38,940
And most of them aren't even about what the video's about.

431
00:35:38,940 --> 00:35:45,220
If the video, and you put up comedy, because it's a comedy skit or something,

432
00:35:45,220 --> 00:35:47,700
okay, well, at least that's something there.

433
00:35:47,700 --> 00:35:52,500
Comedy, Jim Carrey, The Tonight Show, television.

434
00:35:52,500 --> 00:35:53,940
Okay, well, at least that makes sense.

435
00:35:53,940 --> 00:35:59,660
All of this feeds in and ties in together.

436
00:36:02,320 --> 00:36:09,780
So it's understanding and trying to assimilate the data in a certain way that the machine

437
00:36:09,780 --> 00:36:18,260
learning and the models will improve upon.

438
00:36:18,260 --> 00:36:21,160
So we need stuff to engage with.

439
00:36:21,160 --> 00:36:23,060
We need people to engage.

440
00:36:23,060 --> 00:36:26,480
Don't overlook the fact of engagement.

441
00:36:26,480 --> 00:36:29,680
That's very important.

442
00:36:29,680 --> 00:36:31,320
But so is adding to the database.

443
00:36:31,320 --> 00:36:32,320
Okay.

444
00:36:32,320 --> 00:36:33,900
So that's the numbers that we have.

445
00:36:33,900 --> 00:36:35,940
Which brings me to another point.

446
00:36:35,940 --> 00:36:44,820
Do you want Leo to be worth a hell of a lot?

447
00:36:44,820 --> 00:36:50,280
Do you want Leo to be a high market cap token?

448
00:36:50,280 --> 00:36:57,980
Do you want your future of Leo financially to be very, very nice?

449
00:36:57,980 --> 00:37:01,820
Well, I'm going to frame this.

450
00:37:01,820 --> 00:37:08,420
frame this in a way we've all heard of tokens we all see if people want to get

451
00:37:08,420 --> 00:37:14,840
tokens from Leo how are you given to Leo

452
00:37:16,120 --> 00:37:23,580
think of that how many kids are you given to Leo how many tokens are you

453
00:37:23,580 --> 00:37:29,300
creating I'll drop a link here

454
00:37:37,540 --> 00:37:43,220
Ahmad says do you want Leo to be worth a lot yes and and with with Ahmad

455
00:37:43,220 --> 00:37:48,540
I'm I'm preaching to the choir here Ahmad understands this I'll made in fact

456
00:37:48,540 --> 00:37:53,520
most every day is setting up a Threadcast basically for people

457
00:37:53,520 --> 00:37:53,560
you're setting up a Threadcast basically for people you're setting up a Threadcast basically for people

458
00:37:53,560 --> 00:37:59,220
to post their chats with ChatGPT or Claude or wherever.

459
00:37:59,800 --> 00:38:03,440
He has a Threadcast where if you have a conversation with one of the chatbots,

460
00:38:04,000 --> 00:38:07,540
take it, copy it, and drop it in to his Threadcast.

461
00:38:08,480 --> 00:38:10,140
It's a great way to add data.

462
00:38:14,360 --> 00:38:18,000
Anybody who's been involved in computer science before crypto

463
00:38:18,000 --> 00:38:19,920
knows what I'm talking about.

464
00:38:19,920 --> 00:38:24,900
Those of us who weren't, to us a token is crypto.

465
00:38:25,180 --> 00:38:26,340
A token is not crypto.

466
00:38:27,320 --> 00:38:31,540
A token is actually a unit of data.

467
00:38:34,860 --> 00:38:37,840
Tokens are taking on different meaning in the AI age

468
00:38:37,840 --> 00:38:40,060
to most of us who've been in crypto.

469
00:38:41,000 --> 00:38:44,720
It's the same meaning of anybody who's been in the computer science development world

470
00:38:44,720 --> 00:38:47,000
because you've dealt with this for years.

471
00:38:48,740 --> 00:38:49,900
A token is a unit of data.

472
00:38:49,900 --> 00:38:51,060
A token is a unit of data.

473
00:38:52,520 --> 00:38:59,140
And so you'll hear things where they talk about the token window of a chatbot.

474
00:38:59,980 --> 00:39:01,320
How big's the token window?

475
00:39:02,380 --> 00:39:05,420
That's telling you how much data can be put in there.

476
00:39:06,760 --> 00:39:08,860
And so you'll go and you'll look at some of these

477
00:39:08,860 --> 00:39:12,040
and I think Claude is like 100 or 200,000 tokens.

478
00:39:12,640 --> 00:39:15,220
It's like off the charts compared to some of the rest of them.

479
00:39:15,800 --> 00:39:17,460
Meta is like 2,000.

480
00:39:19,900 --> 00:39:27,320
So, yeah, MoreTea seems to be,

481
00:39:27,780 --> 00:39:30,620
is anybody else having problems with me fading out?

482
00:39:36,060 --> 00:39:37,180
Let me know.

483
00:39:37,340 --> 00:39:40,280
I mean, MoreTea said that and I think Brando said it earlier.

484
00:39:40,280 --> 00:39:49,880
I don't know if it's something with my mic or if it's X or I don't know what it is.

485
00:39:49,900 --> 00:39:52,760
So let me know if it continues.

486
00:39:52,940 --> 00:39:55,540
I'll try to hold the phone still.

487
00:39:55,640 --> 00:39:57,220
Maybe that'll be a little better.

488
00:39:58,440 --> 00:40:05,220
But anyway, so basically what we're saying with how many tokens are you adding

489
00:40:05,220 --> 00:40:09,080
is how much data are you giving to Leo?

490
00:40:10,040 --> 00:40:12,140
And I'll frame this up for everybody.

491
00:40:15,980 --> 00:40:17,660
Lama 3.1.

492
00:40:17,660 --> 00:40:22,300
Was trained on 16 trillion tokens.

493
00:40:23,160 --> 00:40:30,620
Now, a token can mean a lot of different things in terms of data.

494
00:40:31,520 --> 00:40:32,840
A token can be a word.

495
00:40:33,260 --> 00:40:34,480
A token can be a space.

496
00:40:34,580 --> 00:40:35,620
A token can be a character.

497
00:40:35,740 --> 00:40:39,420
There's different types of tokens and they denote different things.

498
00:40:42,600 --> 00:40:45,800
But a simple framework to operate from.

499
00:40:45,800 --> 00:40:46,800
Thanks, Walter J.

500
00:40:46,900 --> 00:40:47,460
He said, okay.

501
00:40:47,660 --> 00:40:47,940
Thank you for him.

502
00:40:50,540 --> 00:40:52,600
Sunseeker has the volume going up and down.

503
00:40:53,200 --> 00:40:53,320
Okay.

504
00:40:55,960 --> 00:40:59,460
For a simple frame of reference, let's just say a token is a word.

505
00:41:01,320 --> 00:41:05,020
Now, I dropped the OpenAI tokenizer in the room.

506
00:41:05,020 --> 00:41:11,860
If you take one of your articles and you copy the text and you throw it in there,

507
00:41:12,560 --> 00:41:14,780
it will give you the number of tokens.

508
00:41:15,200 --> 00:41:16,360
That article.

509
00:41:17,660 --> 00:41:27,840
And you'll be able to toggle back and forth between GPT-3 and GPT-3.5 and 4.0.

510
00:41:28,000 --> 00:41:30,060
And you'll notice the tokens are a little different.

511
00:41:31,300 --> 00:41:36,780
And that's what I was talking about with Khal and training on the images and videos.

512
00:41:37,000 --> 00:41:45,480
You'll notice the same article will have more tokens on ChatGPT-3 than it does on ChatGPT-4.

513
00:41:45,820 --> 00:41:47,540
Because there's a little bit more.

514
00:41:47,540 --> 00:41:53,660
There's a lot more efficiency in GPT-4 than GPT-3.

515
00:41:54,520 --> 00:41:57,680
That's the advancement in the efficiency of these different models.

516
00:41:57,860 --> 00:41:59,580
They get more efficient over time.

517
00:42:00,160 --> 00:42:06,860
And so that's why over time it will get cheaper and cheaper to run these models where potentially next year,

518
00:42:06,940 --> 00:42:11,040
the end of next year, maybe in 2026, you can run a GPT.

519
00:42:11,040 --> 00:42:13,580
You can train a GPT-3 for probably $10,000.

520
00:42:14,520 --> 00:42:16,380
So it's going to keep dropping in price.

521
00:42:16,380 --> 00:42:17,360
But anyway.

522
00:42:17,540 --> 00:42:22,820
So you throw your article in and you see it's 800 tokens.

523
00:42:25,420 --> 00:42:28,040
That quantifies your article.

524
00:42:28,640 --> 00:42:39,200
When I do a transcript on my shorts, it's usually about 125, maybe 150 tokens, the transcript.

525
00:42:39,200 --> 00:42:46,440
Because I take my shorts, I run them through a script that Mightpossibly create it, and I get a short.

526
00:42:46,820 --> 00:42:47,520
I mean, I get a transcript.

527
00:42:47,840 --> 00:42:49,320
Which I post under my short.

528
00:42:50,240 --> 00:42:54,220
And it's about 125, maybe 150 tokens.

529
00:42:59,080 --> 00:43:01,760
Llama 3.1 was trained on $16 trillion.

530
00:43:02,280 --> 00:43:04,540
How many tokens are in the Hive database?

531
00:43:05,820 --> 00:43:06,640
I don't know.

532
00:43:06,900 --> 00:43:07,680
I have no clue.

533
00:43:08,340 --> 00:43:11,440
How many words are being created every day on Leo?

534
00:43:11,860 --> 00:43:14,200
How many more words can be created?

535
00:43:14,200 --> 00:43:16,200
And see, this is where Ahmad's Threadcast.

536
00:43:17,540 --> 00:43:20,040
Why I said I'm preaching to the choir.

537
00:43:20,580 --> 00:43:24,980
When he takes a chat that he had with, I think he uses GPT.

538
00:43:24,980 --> 00:43:28,860
So he's having a chat with ChatGPT.

539
00:43:29,380 --> 00:43:32,220
And he breaks it up into six or seven threads.

540
00:43:33,480 --> 00:43:41,980
He's instantly adding, you know, 700, 1,000 tokens to the database.

541
00:43:42,580 --> 00:43:46,360
So the mindset is everybody wants to get Leo tokens.

542
00:43:46,460 --> 00:43:47,520
And they want to take tokens.

543
00:43:47,520 --> 00:43:49,060
And I want to be paid out.

544
00:43:49,140 --> 00:43:49,880
I want to be upvoted.

545
00:43:49,980 --> 00:43:51,180
I want more and more tokens.

546
00:43:51,700 --> 00:43:53,360
Well, why don't you create more and more tokens?

547
00:43:54,760 --> 00:44:09,880
See, if Leo has a database or Hive has a database of billions and billions and billions and tens of billions and hundreds of billions of tokens, that's a valuable database for AI training.

548
00:44:12,440 --> 00:44:16,560
Which then brings us full circle to what I said about HiveFest.

549
00:44:17,520 --> 00:44:23,240
It's one thing to have an open database, a permissionless database, which anybody can utilize.

550
00:44:23,820 --> 00:44:26,340
It's another thing to actually utilize the data.

551
00:44:26,780 --> 00:44:29,160
And that's what LeoAI is being trained on.

552
00:44:30,020 --> 00:44:33,740
Leo is actually training a model utilizing the data.

553
00:44:34,440 --> 00:44:39,000
To my knowledge, and from what Khal said, nobody else is doing that.

554
00:44:40,380 --> 00:44:40,980
Nobody.

555
00:44:43,060 --> 00:44:44,420
That's a big difference.

556
00:44:44,980 --> 00:44:46,960
That's going to make it stand out.

557
00:44:47,520 --> 00:45:01,920
And it's also going to keep feeding into this system where more AI services can be created that can help increase the value of the system.

558
00:45:02,480 --> 00:45:06,820
And if you understand the tokenomics of Leo, we see how this enhances it.

559
00:45:07,900 --> 00:45:13,400
So, for example, LeoAI is going to be a part of premium.

560
00:45:14,980 --> 00:45:16,440
10HBD gets you premium.

561
00:45:16,440 --> 00:45:16,520
Yeah.

562
00:45:17,520 --> 00:45:22,100
So, whatever features you have now, LeoAI will be included.

563
00:45:22,840 --> 00:45:24,680
Now, what will be included in that?

564
00:45:24,780 --> 00:45:25,360
I don't know.

565
00:45:25,560 --> 00:45:26,680
We'll have to wait and see.

566
00:45:26,840 --> 00:45:32,540
But it will have some type of chatbot, first and foremost.

567
00:45:33,860 --> 00:45:37,520
So, now you have chatbot capabilities right on Leo.

568
00:45:38,480 --> 00:45:43,700
You have a co-pilot type capability right on Leo.

569
00:45:43,700 --> 00:45:49,740
You have recommend ends right on Leo.

570
00:45:51,580 --> 00:45:58,600
I mean, I've talked about how, you know, right now, with data in, you put data in on Leo, it's lost forever.

571
00:45:58,760 --> 00:46:00,580
You even create shorts, it's lost.

572
00:46:01,940 --> 00:46:04,780
I mean, it's on your shorts page on your profile.

573
00:46:05,180 --> 00:46:07,320
And you can go to the short screen and scroll down.

574
00:46:07,580 --> 00:46:11,940
But, I mean, now if something posted two days ago, it's missing in action.

575
00:46:12,380 --> 00:46:13,680
You know, you're going to scroll back.

576
00:46:13,680 --> 00:46:15,300
Two days.

577
00:46:17,540 --> 00:46:19,960
But what happens when there's a recommend engine?

578
00:46:21,020 --> 00:46:26,360
What happens when you click on it and play one and another one gets recommended to you?

579
00:46:27,180 --> 00:46:30,180
Isn't this what happens on all the major social media platforms?

580
00:46:30,460 --> 00:46:32,080
That'll be driven by LeoAI.

581
00:46:33,300 --> 00:46:36,800
And, of course, that part, I presume, will be available to everybody.

582
00:46:37,420 --> 00:46:38,720
Everybody will have that feature.

583
00:46:38,820 --> 00:46:40,900
But the chatbot and stuff like that, you pay for it.

584
00:46:42,160 --> 00:46:43,660
Have you seen what's happening with KPI?

585
00:46:43,680 --> 00:46:44,180
Canva.

586
00:46:44,880 --> 00:46:58,020
Canva is creating AI tools on their platform so people can take their images and their information, whatever they're creating, and use AI tools to enhance them and create them and do whatever, edit them, all that stuff.

587
00:47:00,440 --> 00:47:02,160
We need those tools on Leo.

588
00:47:03,300 --> 00:47:11,500
Now, you may end up seeing different layers built where, OK, basic Leo premium gives you the chatbot, gives you the copilot, gives you this.

589
00:47:11,740 --> 00:47:13,580
If you want these other tools, there will be an extra.

590
00:47:13,580 --> 00:47:16,060
We can have one or two businesses and you can have one or five HBD a month or 10 HBD a month or whatever.

591
00:47:17,060 --> 00:47:21,260
And you let people start scaling up and see whatever they want and whatever their needs are.

592
00:47:22,720 --> 00:47:26,600
But that's what I mean by LeoAI as a service.

593
00:47:27,100 --> 00:47:34,820
You start offering out AI tools and all these tools end up generating more data on the platform.

594
00:47:35,660 --> 00:47:39,360
And the more data that's generated on the platform, the more that is fed in.

595
00:47:42,260 --> 00:47:42,760
Yeah, see?

596
00:47:43,580 --> 00:47:51,600
there's one uh ahmed just did one he ran it through the the tokenizer and uh

597
00:47:51,600 --> 00:48:02,580
it was a uh it was a thread that he posted and you can see it's 153 tokens 702 characters

598
00:48:02,580 --> 00:48:10,380
and then you know it gives you some understanding underneath of how how if each token and every one

599
00:48:10,380 --> 00:48:17,920
of those different colors is a token so token r small chunks each of those is a token and then

600
00:48:17,920 --> 00:48:28,220
you'll see where he put a uh um what the hell is that called a hyphen uh hyphen and then one and

601
00:48:28,220 --> 00:48:35,960
then he put a quote that that's a token you know and then he put an explanation of what it is

602
00:48:35,960 --> 00:48:40,360
so if you if you use

603
00:48:40,380 --> 00:48:47,900
that concept if we utilize that concept we say okay how many tokens can we create a day

604
00:48:47,900 --> 00:48:57,320
tokens from the data science world not the crypto world how many tokens a day can we create

605
00:48:57,320 --> 00:49:01,180
that is what you want

606
00:49:01,180 --> 00:49:08,960
that's where activity comes in that's where posting stuff comes in and what ahmed did

607
00:49:10,380 --> 00:49:17,840
look this tokens explained by ChatGPT he just jumped on ChatGPT and probably prompted something

608
00:49:17,840 --> 00:49:22,800
like what something to that effect and he got it he just copied and pasted it

609
00:49:22,800 --> 00:49:33,120
now this is the point where is that data resident the second he prompted that that's on open ai

610
00:49:33,120 --> 00:49:39,040
servers ahmed just gave open ai some synthetic data

611
00:49:40,380 --> 00:49:47,740
open ai is going to feed that right through of course they are but because copied and pasted it

612
00:49:48,540 --> 00:49:55,340
now it's on the high blockchain now it's part of Hive now it's part of the public database

613
00:49:56,460 --> 00:50:04,700
open ai doesn't own it it's public it's on Hive anybody can use it anybody can see it

614
00:50:06,300 --> 00:50:08,700
but not everybody is feeding it in

615
00:50:10,380 --> 00:50:15,180
this is not going to be indexed it's a thread but you know who's going to end up you know who's going

616
00:50:15,180 --> 00:50:20,220
to process it LeoAI because LeoAI processes every thread

617
00:50:23,640 --> 00:50:26,940
the learning engine takes each thread and feeds it in

618
00:50:34,320 --> 00:50:38,040
so basically you're looking at the creation of data

619
00:50:39,120 --> 00:50:40,260
but then the

620
00:50:40,380 --> 00:50:48,540
visualization of data and Hive from day one has created data you know Khal often talks about the

621
00:50:48,540 --> 00:50:56,280
seven day window that that's how we've operated for for the last eight years create data seven

622
00:50:56,280 --> 00:51:01,320
days later paid on it it's gone the article's gone the information's gone nobody gives a crap

623
00:51:03,480 --> 00:51:08,640
and yeah that's going to change with the evergreen rewards and all that other stuff but

624
00:51:09,400 --> 00:51:14,320
to me the true value is not in any of that the true value is in the data

625
00:51:19,080 --> 00:51:24,000
that is going to change the entire economy ai is going to change the entire economy

626
00:51:26,440 --> 00:51:37,000
and so if we want LEO to be 100 150 200 million dollars which I just told you ChatGPT or open ais

627
00:51:37,960 --> 00:51:45,480
valuation you're looking at a smidgen of open ai's valuation which by the way that's just this round

628
00:51:46,200 --> 00:51:52,840
my guess is next round because they will go for more funding I bet they're going to be 250 300

629
00:51:52,840 --> 00:52:03,240
billion when they go another year from now open ai will be a trillion dollar company in a few years

630
00:52:04,040 --> 00:52:05,240
and they're not the only ones

631
00:52:08,600 --> 00:52:12,680
all right is LEO going to be another open ai no it's not

632
00:52:15,960 --> 00:52:26,280
but if we understand the tokenomics could LEO in a year get 10 000 users of LeoAI yes yes they could

633
00:52:30,440 --> 00:52:32,920
because you can sell

634
00:52:33,240 --> 00:52:36,520
the ai services

635
00:52:37,640 --> 00:52:41,000
now you have a lot to offer

636
00:52:42,280 --> 00:52:47,560
and you can keep creating more ai services because the more people use it

637
00:52:47,560 --> 00:52:51,400
the more data is created the more data that's created

638
00:52:51,400 --> 00:52:55,560
essentially the more tokens you have the more tokens you feed into these

639
00:52:55,560 --> 00:53:00,520
models the more your models improve

640
00:53:03,240 --> 00:53:08,440
let me scroll down to before I get

641
00:53:08,440 --> 00:53:10,760
uh

642
00:53:16,600 --> 00:53:20,280
all right there's links uh

643
00:53:23,400 --> 00:53:28,760
ben wants his thank you note I'm a good infograph where did you find

644
00:53:28,760 --> 00:53:33,000
it I created it on ideogram yeah the the the feedback

645
00:53:33,000 --> 00:53:33,240
data

646
00:53:33,240 --> 00:53:36,440
feedback loop I created that on ideogram

647
00:53:36,440 --> 00:53:40,040
it's ai wouldn't it have been nice to be able to

648
00:53:40,040 --> 00:53:44,120
create to create that using LeoAI

649
00:53:44,360 --> 00:53:49,240
and maybe some point we'll have uh maybe we'll have image generation too

650
00:53:49,240 --> 00:53:52,920
see you can get into image generation you can get a video generation anything

651
00:53:52,920 --> 00:53:57,480
that comes out we could incorporate

652
00:54:00,360 --> 00:54:03,000
the community comments features on youtube isn't very

653
00:54:03,000 --> 00:54:05,960
good if their aim is to keep people on their platform they should do better

654
00:54:05,960 --> 00:54:10,200
they got a whole new thing that they just came out with ahmed

655
00:54:10,200 --> 00:54:15,320
more data morty's right people comfortable with social media

656
00:54:15,320 --> 00:54:17,320
will never leave it the blockchain has much

657
00:54:17,320 --> 00:54:22,200
to offer better service for them to switch and make their sector place yeah

658
00:54:22,200 --> 00:54:27,720
I answered the competitive edge voice we're like five lines ahead I don't know

659
00:54:27,720 --> 00:54:30,680
what that means ben

660
00:54:30,680 --> 00:54:32,840
I don't know what that means ben

661
00:54:33,640 --> 00:54:36,840
LEO is promised to have a lot of specialized feature that depends on

662
00:54:36,840 --> 00:54:40,840
real-time data from Hive compared to ChatGPT that does have real-time data

663
00:54:42,120 --> 00:54:48,360
it's an advantage well see um ahmed there's another thing that comes into play with LEO

664
00:54:48,360 --> 00:55:01,800
ai it's trained it its core is meta uh llama llama does the same thing in that it's updated

665
00:55:02,520 --> 00:55:02,920
because

666
00:55:03,000 --> 00:55:09,960
of whatever's fed onto the facebook instagram

667
00:55:09,960 --> 00:55:14,840
so I've gone to meta and asked it about the score of a baseball game from last

668
00:55:14,840 --> 00:55:19,240
night and it spit it out I asked it to give me a summary and it spit it out

669
00:55:19,240 --> 00:55:23,720
because obviously that was posted on one of its apps

670
00:55:23,720 --> 00:55:27,160
I go to Grok with the q not with the k I go to

671
00:55:27,160 --> 00:55:32,440
Grok and I ask that and it gives me the same thing as claude well back as of my

672
00:55:32,440 --> 00:55:37,800
last update in april 2024 and I'm sure open ai is the same thing

673
00:55:38,440 --> 00:55:44,840
that's the advantage of these platforms that have real-time data being fed in

674
00:55:44,840 --> 00:55:50,760
this is why I always talk about having news taking news articles taking news headlines

675
00:55:50,760 --> 00:55:56,680
taking news the first paragraph of a news article and posting it on LEO this is one

676
00:55:56,680 --> 00:56:02,200
of the reasons why I've talked about news to LEO 2.0 is one of the most important features

677
00:56:02,440 --> 00:56:12,200
to bring out simply because this could add a huge amount of data to the blockchain and what it is

678
00:56:12,200 --> 00:56:21,000
for those who don't know is news to threads 2.0 it's going to command prompt of some kind on the

679
00:56:21,000 --> 00:56:27,960
UI so you'll go into the thread box and you will you'll somehow issue a command

680
00:56:29,960 --> 00:56:31,640
and you'll put in a link to an article

681
00:56:32,440 --> 00:56:36,280
and it will take that link from that article and essentially create a thread store

682
00:56:38,520 --> 00:56:45,720
it'll summarize the article and create a thread storm for you so now that article

683
00:56:46,680 --> 00:56:52,920
if it was from sports illustrated if it was from espn if it was from the bbc if it was

684
00:56:54,840 --> 00:56:59,960
cnn sky news whatever it was the wall street journal cnbc

685
00:57:01,240 --> 00:57:02,120
all of a sudden that

686
00:57:02,920 --> 00:57:09,720
article will will end up being fed into LeoAI so if you've got a lot of information that article

687
00:57:10,280 --> 00:57:20,760
about the fed rate cut or about the latest epl game or whatever it's now a the news the information

688
00:57:20,760 --> 00:57:27,560
is right there on LEO but b it's also fed into LeoAI so if that was out somebody could ask about it

689
00:57:28,360 --> 00:57:30,520
but even though LeoAI is a little bit down the road

690
00:57:30,520 --> 00:57:41,980
And another point to this, that news on thread I think is important, I know immutability is not important to most people, but a record of history is.

691
00:57:41,980 --> 00:57:58,960
And we see a lot of push to try to rewrite history and change history and people going back and saying, well, we didn't say that, we didn't push that, we didn't say that that wasn't effective.

692
00:57:58,960 --> 00:58:13,440
Having that type of stuff on blockchain, then it's right there and it's like, this is exactly what was written. You can't go back and change that article and, you know, edit it and change what it was.

693
00:58:14,160 --> 00:58:28,440
And so now the public version is completely different than what was ever before. Because it's on blockchain. And yeah, you see the latest version, but guess what, when you go to the block explorer, you get the link, and you just keep linking back to the blocks that are tied together.

694
00:58:28,440 --> 00:58:29,900
And all of a sudden, you get the original article.

695
00:58:34,520 --> 00:58:38,960
Chat GPT can be used to create Hive-specific applications so it knows enough about Hive.

696
00:58:39,580 --> 00:58:41,000
LeoAI should be a lot better.

697
00:58:41,460 --> 00:58:41,680
Yeah.

698
00:58:46,040 --> 00:58:52,680
I remember reading that grass scraped the whole Reddit, so I don't think it would take much to scrape Hive.

699
00:58:52,680 --> 00:58:55,980
That's true, but most of the data, they may not want.

700
00:58:56,920 --> 00:58:57,820
They're not going to want.

701
00:58:58,300 --> 00:58:58,420
I think.

702
00:58:58,440 --> 00:58:59,120
They're not going to want to do things like upvote.

703
00:59:00,320 --> 00:59:01,160
They don't really care.

704
00:59:06,500 --> 00:59:12,200
As far as I know, LeoAI is not trained on images. Too expensive. Yeah, that's my understanding, too.

705
00:59:13,840 --> 00:59:19,740
It's my presumption. I've never heard one way or another, but I would imagine it's just.