1
00:00:00,000 --> 00:00:14,380
Today is the 30th of August 2024 and we will start in a few minutes when everybody has a chance to log on.
2
00:00:30,000 --> 00:00:59,980
Thank you.
3
00:01:00,000 --> 00:01:29,980
Thank you.
4
00:01:30,000 --> 00:01:59,980
Thank you.
5
00:02:00,000 --> 00:02:08,880
All right, everybody, we will get started here.
6
00:02:09,660 --> 00:02:22,340
So it's not to be confused with Khal, be within a few minutes of our starting time, not 45 minutes or an hour like he likes to do.
7
00:02:23,660 --> 00:02:27,380
Anyway, hope everybody is well.
8
00:02:27,840 --> 00:02:29,480
We have a lot going on.
9
00:02:29,480 --> 00:02:31,700
A lot of different things taking place.
10
00:02:32,260 --> 00:02:36,480
Last week we had Vaultec on talking about VSC.
11
00:02:37,460 --> 00:02:40,680
That was an informative conversation.
12
00:02:40,940 --> 00:02:42,760
I hope you found it to be also.
13
00:02:44,060 --> 00:02:46,440
And today we'll get back to Leo.
14
00:02:47,120 --> 00:02:49,240
And this actually is a follow up.
15
00:02:50,020 --> 00:02:51,080
I think it was two weeks ago.
16
00:02:51,180 --> 00:02:52,820
Maybe it was three weeks ago.
17
00:02:53,420 --> 00:02:54,780
I don't remember exactly.
18
00:02:54,780 --> 00:02:58,780
But we did a lion's den on AI.
19
00:02:59,480 --> 00:03:00,660
LeoAI.
20
00:03:01,880 --> 00:03:08,580
And the reason why I wanted to do a follow up is I'm continually researching this subject.
21
00:03:08,980 --> 00:03:18,520
I'm continually talking to other people who are involved in it, continually looking at other things and just always stepping back and seeing what's going on.
22
00:03:18,520 --> 00:03:28,520
If you go into the technology blog, you see or the technology Threadcast, excuse me, you see videos posted that.
23
00:03:28,520 --> 00:03:36,460
That myself and primarily Mightpossibly watch, but there's some others.
24
00:03:37,300 --> 00:03:43,060
We also have articles that are related to technology.
25
00:03:43,060 --> 00:03:50,600
It's not exclusive to that, but there's also articles related to AI and some of the things going on there.
26
00:03:51,700 --> 00:03:57,000
And so it's really becoming a resource of what's taking place.
27
00:03:57,500 --> 00:03:58,340
Also.
28
00:03:58,340 --> 00:03:58,500
Okay.
29
00:03:58,520 --> 00:04:11,900
I recently started adding some of my conversations with Claude and Grok and things of that nature into there for a couple of reasons.
30
00:04:11,900 --> 00:04:15,100
The first reason is the basic premise of this call.
31
00:04:15,560 --> 00:04:22,800
But the second reason is, again, as an information source, as an educational source.
32
00:04:23,580 --> 00:04:28,500
And I find that this is a great use case for the daily thread.
33
00:04:28,520 --> 00:04:51,660
Cast that if you are doing a daily Threadcast about a specific topic, like this one with technology could do it with history, you could do it with there's one with news that shows up pretty much anything that falls into that category, make it a information source.
34
00:04:51,660 --> 00:04:55,660
I've mentioned this to the Brazilian community a number of times.
35
00:04:56,340 --> 00:04:58,340
Put anything related to Brazil.
36
00:04:58,340 --> 00:05:03,740
Brazil in their Brazilian news, Brazilian information, stuff that Brazilians would be interested in.
37
00:05:06,800 --> 00:05:12,620
That's the first level, the engagement, what we comment on, what we read, all of that.
38
00:05:13,160 --> 00:05:21,460
The second layer is one that I feel is more important, and this is the money printer.
39
00:05:23,020 --> 00:05:25,160
And I dropped in the Threadcast.
40
00:05:25,160 --> 00:05:48,720
About wanting $100,000,000 market cap, and that's a level Khal has thrown out there, I think it's a level that's completely attainable, and if we understand how all of this feeds in, we can understand what it takes to get there.
41
00:05:48,720 --> 00:05:53,120
The world of AI is exploding.
42
00:05:53,120 --> 00:05:54,940
Now, people talk about hype.
43
00:05:55,160 --> 00:05:57,160
People talk about bubbles.
44
00:05:57,160 --> 00:05:59,160
People talk about a lot of stuff.
45
00:05:59,160 --> 00:06:05,160
And maybe there's a degree of truth to it.
46
00:06:05,160 --> 00:06:09,160
When you talk bubbles, now you're talking finance, you're talking markets.
47
00:06:09,160 --> 00:06:11,160
They can get bubblicious.
48
00:06:11,160 --> 00:06:15,160
So I don't even engage in that conversation.
49
00:06:15,160 --> 00:06:17,160
Hype is related to technology.
50
00:06:17,160 --> 00:06:19,160
There could be hype.
51
00:06:19,160 --> 00:06:21,160
There's been a lot of hype cycles.
52
00:06:21,160 --> 00:06:23,160
I've been through a lot of hype cycles in 25 years.
53
00:06:23,160 --> 00:06:25,160
I've been through a lot of hype cycles in 25 years.
54
00:06:25,160 --> 00:06:39,160
In terms of watching technology and technological trends, I think I've mentioned on this broadcast that I've been through at least three hype cycles with virtual reality.
55
00:06:39,160 --> 00:06:46,160
And we're still really not much closer to a VR world than we were years ago.
56
00:06:46,160 --> 00:06:48,160
So they do happen.
57
00:06:48,160 --> 00:06:51,160
And maybe there is a degree of that with AI.
58
00:06:51,160 --> 00:06:53,160
But let's not mistake that.
59
00:06:53,160 --> 00:06:58,160
The underlying premise of what's taking place.
60
00:06:58,160 --> 00:07:03,160
There is radical advancement in technologies.
61
00:07:03,160 --> 00:07:10,160
You can see it looking at specifically generative AI because that's what most of us encounter.
62
00:07:10,160 --> 00:07:14,160
You look at image generators as an example.
63
00:07:14,160 --> 00:07:20,160
Look at how much they've improved in the last year, in the last six months.
64
00:07:20,160 --> 00:07:22,160
Hell, in the last three months.
65
00:07:23,160 --> 00:07:25,160
Look at video generators.
66
00:07:25,160 --> 00:07:27,160
Now, many of us haven't played with a lot of them.
67
00:07:27,160 --> 00:07:29,160
Mightpossibly has done a good bit of it.
68
00:07:29,160 --> 00:07:38,160
But there are some videos online about what things such as Sora are capable of doing now compared to where things stood a year ago.
69
00:07:38,160 --> 00:07:42,160
Very famous is Will Smith eating spaghetti.
70
00:07:42,160 --> 00:07:45,160
You look at the video a year, year and a half ago, whatever it was.
71
00:07:45,160 --> 00:07:47,160
He looked like Plastic Man.
72
00:07:47,160 --> 00:07:50,160
His face was all messed up and everything else.
73
00:07:50,160 --> 00:07:52,160
And the movement wasn't smooth.
74
00:07:52,160 --> 00:07:55,160
And it was just a train wreck.
75
00:07:55,160 --> 00:07:57,160
You look at what Sora presented.
76
00:07:57,160 --> 00:07:58,160
It was much improved.
77
00:07:58,160 --> 00:08:01,160
Still not perfect, but much improved.
78
00:08:01,160 --> 00:08:07,160
So this is what we're dealing with AI.
79
00:08:07,160 --> 00:08:17,160
And so what we have going on is we need to let's start with what is AI.
80
00:08:17,160 --> 00:08:22,160
And there's a lot of different definitions.
81
00:08:22,160 --> 00:08:25,160
Things people put up.
82
00:08:25,160 --> 00:08:30,160
But here it is in the most simplistic terms.
83
00:08:30,160 --> 00:08:40,160
AI is data wrapped in algorithms and fed through compute.
84
00:08:40,160 --> 00:08:42,160
That's AI.
85
00:08:42,160 --> 00:08:44,160
That's what comes out, AI.
86
00:08:44,160 --> 00:08:48,160
Now, depending on what your algorithms are training on, it could be images.
87
00:08:48,160 --> 00:08:49,160
It could be video.
88
00:08:49,160 --> 00:08:51,160
It could be the ability to drive a car.
89
00:08:51,160 --> 00:08:53,160
Ability to run a robot.
90
00:08:53,160 --> 00:08:54,160
A lot of different things.
91
00:08:54,160 --> 00:08:57,160
But it all comes back to that same premise.
92
00:08:57,160 --> 00:09:05,160
And so there's also a lot of variables that expand the capabilities.
93
00:09:05,160 --> 00:09:08,160
So, for example, algorithms.
94
00:09:08,160 --> 00:09:13,160
And writing better algorithms, which is being done.
95
00:09:13,160 --> 00:09:20,160
And so there's a growth curve just from the algorithms if everything else stayed constant.
96
00:09:20,160 --> 00:09:23,160
But everything else doesn't stay constant.
97
00:09:23,160 --> 00:09:28,160
We see the amount of compute that is being put in the field is growing.
98
00:09:28,160 --> 00:09:34,160
In fact, AI compute is doubling every six months.
99
00:09:34,160 --> 00:09:37,160
Keep in mind, Moore's law, which was a little different.
100
00:09:37,160 --> 00:09:38,160
It wasn't total compute.
101
00:09:38,160 --> 00:09:42,160
That was the transmitters per wafer.
102
00:09:42,160 --> 00:09:48,160
But basically, when you look at Moore's law, a simple way of looking at it is the cost per compute doubled about every two years.
103
00:09:48,160 --> 00:09:51,160
18 months, two years, something like that.
104
00:09:51,160 --> 00:09:57,160
Well, we have the amount of AI compute going in doubling every six months.
105
00:09:57,160 --> 00:10:00,160
So we have bigger amounts of compute.
106
00:10:00,160 --> 00:10:02,160
The other thing is data.
107
00:10:02,160 --> 00:10:08,160
More data, more compute means more AI output.
108
00:10:08,160 --> 00:10:11,160
That's a simple way of looking at it.
109
00:10:11,160 --> 00:10:17,160
Now, data also can be broken down into structured data, which is a lot easier.
110
00:10:17,160 --> 00:10:22,160
Labeled data, which is a lot easier to use, versus raw data.
111
00:10:22,160 --> 00:10:29,160
Well, they get a lot of time structured data and labeled data out of raw data.
112
00:10:29,160 --> 00:10:38,160
And what they do is the data scientists and data science, there's a thing called knowledge graphs.
113
00:10:38,160 --> 00:10:47,160
And basically, that's a way of linking this data together so that it's able to spit out information.
114
00:10:47,160 --> 00:10:53,160
And these are the three phases that we go to.
115
00:10:53,160 --> 00:10:59,160
And all of this is applicable to Leo, I promise you.
116
00:10:59,160 --> 00:11:03,160
When we go to the first phase, it's data.
117
00:11:03,160 --> 00:11:05,160
We'll just take raw data.
118
00:11:05,160 --> 00:11:07,160
You have raw data.
119
00:11:07,160 --> 00:11:08,160
That gets processed.
120
00:11:08,160 --> 00:11:09,160
It gets labeled.
121
00:11:09,160 --> 00:11:11,160
It gets structured.
122
00:11:11,160 --> 00:11:14,160
Maybe it is somewhat structured, let's say a table.
123
00:11:14,160 --> 00:11:17,160
Because keep in mind, all this has to be converted to data.
124
00:11:17,160 --> 00:11:22,160
It has to be converted to computer language, which computer language is binary code.
125
00:11:22,160 --> 00:11:24,160
It's ones and zeros.
126
00:11:24,160 --> 00:11:30,160
So even though coders use like Rush and Python and other things, that's not computer code.
127
00:11:30,160 --> 00:11:32,160
Computer code is binary.
128
00:11:32,160 --> 00:11:38,160
So the computer has to take this and convert it to binary code.
129
00:11:38,160 --> 00:11:41,160
So they do better with tables.
130
00:11:41,160 --> 00:11:44,160
Tables can structure better because they're kind of structured.
131
00:11:44,160 --> 00:11:46,160
And of course, there are a lot of numbers.
132
00:11:46,160 --> 00:11:56,160
And binary code works well with numbers, as we've seen over the last 60 years since that's where computers first and foremost started to penetrate.
133
00:11:56,160 --> 00:11:58,160
So you have the data.
134
00:11:58,160 --> 00:12:05,160
You have these knowledge graphs that are created out of this data, which brings us to information.
135
00:12:05,160 --> 00:12:13,160
And this is, as I explained on the previous call, this is where generative AI is a little bit different.
136
00:12:13,160 --> 00:12:16,160
People go to these chatbots and expect a mirror image.
137
00:12:16,160 --> 00:12:20,160
Or a mirror output of what they put in, like a search engine.
138
00:12:20,160 --> 00:12:23,160
That's not what this is doing.
139
00:12:23,160 --> 00:12:25,160
This is taking in data.
140
00:12:25,160 --> 00:12:26,160
It's being trained on data.
141
00:12:26,160 --> 00:12:28,160
And it spits out information.
142
00:12:28,160 --> 00:12:33,160
And it spits out information it thinks is applicable to what you want.
143
00:12:33,160 --> 00:12:41,160
Now, a lot of it, especially in the early stages, is a guessing game, hence where the hallucinations come in.
144
00:12:41,160 --> 00:12:45,160
And the system will just kind of make stuff up.
145
00:12:46,160 --> 00:12:49,160
You know, think of it as a politician.
146
00:12:49,160 --> 00:12:52,160
Politicians never say, I don't know, they just make crap up.
147
00:12:52,160 --> 00:12:54,160
Well, that's what the AI is doing.
148
00:12:54,160 --> 00:13:01,160
But as the compute gets bigger and as they're able to train on more data, the hallucinations decline over time.
149
00:13:01,160 --> 00:13:05,160
They don't necessarily go away, but they will decline over time.
150
00:13:05,160 --> 00:13:12,160
Now, the third thing in this, and this is very important, is the next step is knowledge.
151
00:13:12,160 --> 00:13:14,160
Now, you have data.
152
00:13:14,160 --> 00:13:15,160
You have information.
153
00:13:15,160 --> 00:13:16,160
You have knowledge.
154
00:13:16,160 --> 00:13:22,160
For knowledge, we require context.
155
00:13:22,160 --> 00:13:27,160
For example, 4% means nothing.
156
00:13:27,160 --> 00:13:28,160
That's data.
157
00:13:28,160 --> 00:13:29,160
Something's 4%.
158
00:13:29,160 --> 00:13:31,160
Well, what the hell does it apply to?
159
00:13:31,160 --> 00:13:32,160
It's just data, 4%.
160
00:13:32,160 --> 00:13:33,160
It could mean anything.
161
00:13:33,160 --> 00:13:45,160
The information is 4% APR, or even more specifically, 4% APR on a 10-year bond.
162
00:13:45,160 --> 00:13:46,160
That's information.
163
00:13:46,160 --> 00:13:47,160
No data allowed through an opponent.
164
00:13:47,160 --> 00:13:48,160
No bank account.
165
00:13:48,160 --> 00:13:49,160
No code.
166
00:13:49,160 --> 00:13:50,160
There's no information.
167
00:13:50,160 --> 00:13:51,160
There's only a little bit of information.
168
00:13:51,160 --> 00:13:52,160
But that's it.
169
00:13:52,160 --> 00:13:53,160
Knowledge is understanding what the hell that means.
170
00:13:53,160 --> 00:13:54,160
I wasn't even happy with just a little bit of Axial.
171
00:13:54,160 --> 00:14:10,260
…
172
00:14:10,260 --> 00:14:11,160
Knowledge is understanding what the hell that means.
173
00:14:11,160 --> 00:14:12,160
It means 4% as opposed to 10%, 6%, 1%, 0%.
174
00:14:12,160 --> 00:14:13,160
Knowledge means a context of 4% as opposed to 10%, 6%, 1%, half a percent, 0%.
175
00:14:13,160 --> 00:14:14,160
APR, or APY, on a bank?
176
00:14:14,160 --> 00:14:15,160
On a bond?
177
00:14:15,160 --> 00:14:22,840
Versus what? Treasury bonds versus corporate bonds versus stocks. What does it mean?
178
00:14:23,540 --> 00:14:31,140
This is context. And this is very important. And this is the challenge that we are at right now
179
00:14:31,140 --> 00:14:39,860
with LeoAI or with AI. So that said, what does all this have to do with Leo? What does this have
180
00:14:39,860 --> 00:14:45,640
to do with Leo and $100 million market cap? Well, the reason why I call this a money printer
181
00:14:45,640 --> 00:14:55,900
is it's very simple. More data with improving algorithms plus more compute equals better AI.
182
00:14:56,360 --> 00:15:06,940
Better AI has a greater value than not as great AI. Because once you take the better AI
183
00:15:06,940 --> 00:15:09,840
and you start to build and incorporate it,
184
00:15:09,860 --> 00:15:22,900
you end up with improvements in your platform. You end up with products that have increased
185
00:15:22,900 --> 00:15:35,900
value. It's really very simple. You see increased utility. Now, one of the things we have to keep
186
00:15:35,900 --> 00:15:39,060
in mind, I'll kind of tangent here a little bit into Web3.
187
00:15:39,860 --> 00:15:52,620
One of the factors of technology is demonetization. So for example, for those of you who remember
188
00:15:52,620 --> 00:16:01,860
before the smartphone, you go back, what did it take to add up numbers? You go way back, it was
189
00:16:01,860 --> 00:16:08,700
the old mechanical adding machines that CPA used. Then we got digital calculators. And they were big,
190
00:16:08,700 --> 00:16:09,460
they were expensive.
191
00:16:09,860 --> 00:16:14,680
They were like $100. You plugged them into the wall. Then in the 80s, you ended up with
192
00:16:14,680 --> 00:16:19,660
these little pocket solar calculators that cost like 20 bucks. Then they went down in
193
00:16:19,660 --> 00:16:25,180
price in the 90s. And you know, they were giving them away for free. If you, you know,
194
00:16:25,180 --> 00:16:30,360
did something at the shopping mall, they gave you a little pocket calculator or whatever.
195
00:16:30,360 --> 00:16:38,660
And now today, you have a smartphone, it has a calculator on it. It costs you nothing.
196
00:16:39,860 --> 00:16:42,560
These things used to cost 100 bucks to add up columns.
197
00:16:42,560 --> 00:16:43,940
Bruce Jentleson,ığın.com
198
00:16:43,940 --> 00:16:47,980
And, by the way, if you're doing an add up of a column online in like an Excel spreadsheet
199
00:16:47,980 --> 00:16:53,460
or something like that, an algorithm, you can put in an algorithm, it'll do all of that
200
00:16:53,460 --> 00:16:54,200
for you.
201
00:16:55,680 --> 00:17:01,660
So not only do you have the device being free by doing it in a spreadsheet, your time is
202
00:17:01,660 --> 00:17:05,340
safe. You don't even have to enter all these numbers, make a mistake. Try to figure out
203
00:17:05,340 --> 00:17:08,840
where did you reverse a number? Where did you put in the wrong number? It's creative, hardcore,
204
00:17:08,840 --> 00:17:09,060
Yeah.
205
00:17:09,060 --> 00:17:09,860
Bruce Jentleson,own.com
206
00:17:09,860 --> 00:17:18,020
wrong number all of that it's just here it is the demonetization so now if we
207
00:17:18,020 --> 00:17:23,660
have demonetization of stuff how does value grow how do you capture value in a
208
00:17:23,660 --> 00:17:30,380
world where revenues and profits disappear because of demonetization this
209
00:17:30,380 --> 00:17:36,980
is where I believe tokenization comes in because even though it's demonetized it
210
00:17:36,980 --> 00:17:48,920
it adds utility to whatever it's on and so anything that's added to Leo even if it is demonetized
211
00:17:50,960 --> 00:17:57,380
adds value to Leo and tokenization the idea of digital real estate in my opinion is what
212
00:17:57,380 --> 00:18:06,440
captures this so how do we drive up the value if we look past the engagement
213
00:18:06,980 --> 00:18:10,940
and I'm not saying that's not important I'm not saying that having people read what we
214
00:18:10,940 --> 00:18:15,380
write isn't important I'm not saying that having people watch the shorts isn't important people
215
00:18:15,380 --> 00:18:21,500
posting shorts isn't important all of that is I'm just dealing at the data level I'm dealing
216
00:18:21,500 --> 00:18:31,400
at the AI level I'm dealing at feeding this engine and so obviously the more data that we have in our
217
00:18:32,900 --> 00:18:34,700
blockchain on our blockchain
218
00:18:36,260 --> 00:18:36,920
the more
219
00:18:36,980 --> 00:18:42,620
valuable it becomes now it's not captured because it's permissionless
220
00:18:43,940 --> 00:18:50,840
but see I I notice uh Lucille mentioned X before and she how she's upset that every day she's
221
00:18:50,840 --> 00:18:56,360
locked out of her X account and has to reset it well what the hell you doing on X you know what
222
00:18:56,360 --> 00:19:03,680
you're doing on X you're feeding Elon every time you hit a vote every time you submit something
223
00:19:03,680 --> 00:19:04,880
you're feeding Elon
224
00:19:06,080 --> 00:19:06,560
now
225
00:19:06,980 --> 00:19:14,180
you're feeding Elon with this space why because he's the one who offers space and we don't have
226
00:19:14,180 --> 00:19:21,320
an alternative could go to YouTube kind of but that's feeding Google whatever
227
00:19:22,640 --> 00:19:28,640
you know we're feeding these big tech companies and obviously getting away from them I mean you
228
00:19:28,640 --> 00:19:33,560
know you can go with everything to our mobile carriers you can go to everything we do Google
229
00:19:33,560 --> 00:19:35,840
you go to a website Google's has
230
00:19:35,840 --> 00:19:40,820
it's bots all through the website anyway so even if it's not monitoring you it's monitoring the
231
00:19:40,820 --> 00:19:47,240
site which is seeing what you're doing so obviously there's a whole system eradicated uh of surveillance
232
00:19:47,240 --> 00:19:52,820
that they've set up but the point being is if we can start taking data and putting it on the
233
00:19:52,820 --> 00:20:04,940
blockchain that gets it away from big tech and if we can feed that data onto Hive and Leo
234
00:20:05,840 --> 00:20:12,260
the Leo team starts to write better algorithms and more compute
235
00:20:14,240 --> 00:20:22,640
you end up with an improving AI and an improving AI means more value because
236
00:20:22,640 --> 00:20:26,660
then you can start to do other things such as been mentioned in the past as
237
00:20:26,660 --> 00:20:34,220
co-pilots as build image generators you know you can feed all this stuff in ultimately video
238
00:20:34,220 --> 00:20:41,500
generators you can start potentially starting to build ai agents that do different things that are
239
00:20:41,500 --> 00:20:52,300
tied to LEO all of this becomes possible and it becomes part of a larger decentralized ecosystem
240
00:20:52,860 --> 00:21:00,060
some are calling a decentralized brain where ai and blockchain start to merge they converge
241
00:21:00,060 --> 00:21:06,620
and it multiplies at a much greater rate than the any individual centralized entity
242
00:21:08,380 --> 00:21:13,660
because don't forget when you go to a ChatGPT when you go to a Grok when you go to
243
00:21:13,660 --> 00:21:19,420
you know any of these chat bots claude they have their models
244
00:21:21,500 --> 00:21:27,980
they don't share those models nobody can add to those mouse you add the data you use it you
245
00:21:27,980 --> 00:21:29,980
engage the data into the data you use it you engage it
246
00:21:30,060 --> 00:21:37,140
with it you feed it more data more synthetic data but nobody adds to the
247
00:21:37,140 --> 00:21:45,180
models other than anthropic or Google or XAI or whoever that's not the same when
248
00:21:45,180 --> 00:21:51,760
you get into blockchain with blockchain a you have your data which is
249
00:21:51,760 --> 00:21:56,760
permissionless so any blockchain that's out there public blockchain any of the
250
00:21:56,760 --> 00:22:02,100
data is permissionless anybody can can scrape that data but also when you have
251
00:22:02,100 --> 00:22:08,340
AI models that are built on these blockchains there's the potentiality
252
00:22:08,340 --> 00:22:12,760
and I say the potentiality because some of them can be closed models LeoAI may
253
00:22:12,760 --> 00:22:17,760
be a close is probably going to be a closed model that's based upon an open
254
00:22:17,760 --> 00:22:24,360
LLM which I'll explain all that in a second but the point being is when you
255
00:22:24,360 --> 00:22:26,760
have these open models tied to blockchain
256
00:22:26,760 --> 00:22:36,300
other people can use them fork them build upon them add to them make them do
257
00:22:36,300 --> 00:22:41,940
different things build different agents tied to these models and then these
258
00:22:41,940 --> 00:22:47,640
agents can be connected together just like personal computers were connected
259
00:22:47,640 --> 00:22:54,560
together by the internet and all of that grew the value of this thing called
260
00:22:54,560 --> 00:22:56,560
computing
261
00:22:56,760 --> 00:23:02,260
how much is computing worth today in other words how much is the internet
262
00:23:02,260 --> 00:23:07,900
worth anybody guess because that's all it'll be thirty trillion fifty trillion
263
00:23:07,900 --> 00:23:16,580
hundred trillion who knows who has any idea about that so we've all heard the
264
00:23:16,580 --> 00:23:25,960
saying data is the new oil right so if you want a big oil well have a lot of
265
00:23:25,960 --> 00:23:26,340
data
266
00:23:26,760 --> 00:23:40,260
this is something very important and it's important because we need to secure
267
00:23:40,260 --> 00:23:45,660
a lot of data to make sure it's available to everyone this is what
268
00:23:45,660 --> 00:23:51,840
decentralization is all about folks so the question I always have it boils down
269
00:23:51,840 --> 00:23:56,340
to this do you want a future run by big tech who controls everything
270
00:23:56,760 --> 00:24:03,760
and the rest of us are only allocated based upon what the regular you say that
271
00:24:03,760 --> 00:24:10,640
Sam Altman and Mark Zuckerberg and all these other tech billionaires what they
272
00:24:10,640 --> 00:24:19,080
allow us to do or do you want this available to everybody and it all starts
273
00:24:19,080 --> 00:24:26,380
with the data and from Leo's perspective Leo's slice of the pie
274
00:24:26,760 --> 00:24:35,820
comes from the data we create notice how you have not seen Google X AI which
275
00:24:35,820 --> 00:24:44,540
owns Twitter or X well I guess actually X calm those acts but you get the point
276
00:24:44,540 --> 00:24:54,060
or meta so you see anthropic suit you see open a I sued by publishers you
277
00:24:54,060 --> 00:24:56,680
notice how those other social media and platforms
278
00:24:56,680 --> 00:25:07,000
not sued. Why? They keep generating their own data. Well, their users do. The users
279
00:25:07,000 --> 00:25:12,860
show up every day when Lucille goes to Twitter and does whatever she does. After
280
00:25:12,860 --> 00:25:20,020
she resets her account on Twitter, she is feeding X's database. When somebody goes
281
00:25:20,020 --> 00:25:26,980
to Instagram and puts up a photo, they are feeding Zuckerberg's database. They
282
00:25:26,980 --> 00:25:33,640
don't have to go out and scour the internet for data. They have it growing
283
00:25:33,640 --> 00:25:40,600
every day. In fact, other companies are trying to scrape their data. We heard of
284
00:25:40,600 --> 00:25:46,300
X locking down their data. Reddit locked down their data. Reddit now sells their
285
00:25:46,300 --> 00:25:49,960
data. Two AI companies.
286
00:25:50,020 --> 00:26:04,420
So, if we look at data as the new oil, why aren't we creating more oil? And
287
00:26:04,420 --> 00:26:10,220
that's exactly what I've been doing. I posted a tweet here not long ago that I
288
00:26:10,220 --> 00:26:15,720
talked about Web3. We are in the competition mindset. That's what we were
289
00:26:15,720 --> 00:26:20,000
reared in. That's what we dealt with. When you're in the virtual world, you're
290
00:26:20,000 --> 00:26:23,180
in the virtual world, when you're in the world of abundance, when you're in the
291
00:26:23,180 --> 00:26:30,440
world that isn't limited other than by bandwidth and compute, it's not to out
292
00:26:30,440 --> 00:26:35,900
compete others. It's out contribute. That's what open source is. Out
293
00:26:35,900 --> 00:26:41,240
contribute. That's what gets you recognized. And I've mentioned it before.
294
00:26:41,240 --> 00:26:45,500
I always mention them. Phish. I see Phish every day. Phish is in the technology
295
00:26:45,500 --> 00:26:49,980
blog Threadcast. He's dropping them. Phish is in when he's around. He's in the
296
00:26:49,980 --> 00:26:53,660
town. They're not on vacation in the Alps. By the way, I hope you had a great
297
00:26:53,660 --> 00:27:07,700
trip. He's in Threadcast. He's contributing. So, this is the big thing.
298
00:27:07,700 --> 00:27:15,060
He's contributing. And contribute the data. And so, one of the things that we've
299
00:27:15,060 --> 00:27:19,040
seen a lot of conversation about is the recent addition of
300
00:27:19,040 --> 00:27:19,980
. . .
301
00:27:19,980 --> 00:27:27,200
Shorts. Wonderful addition. Not even from this perspective, it's a wonderful
302
00:27:27,200 --> 00:27:35,840
addition from the utility standpoint. We know shorts is very popular. TikTok is an
303
00:27:35,840 --> 00:27:44,080
example of that. So, now we have shorts. Now, does shorts help the data? Questionable.
304
00:27:44,080 --> 00:27:49,980
Why is it questionable? Well, right now, as far as I understand, there's a lot of
305
00:27:49,980 --> 00:27:50,980
jatangles exist.
306
00:27:50,980 --> 00:27:51,980
So, that information could be valuable.
307
00:27:51,980 --> 00:27:52,980
I see about as much advertising for Rogers.
308
00:27:52,980 --> 00:27:53,980
I see a lot of advertising for Rogers.
309
00:27:53,980 --> 00:27:54,980
After all, bliruthing.
310
00:27:54,980 --> 00:27:55,980
One-for-all.
311
00:27:55,980 --> 00:27:56,980
But why?
312
00:27:56,980 --> 00:27:57,980
Because video is so tissue-rich.
313
00:27:57,980 --> 00:27:58,980
LeoAI is not multi-modal.
314
00:27:58,980 --> 00:27:59,980
It's text.
315
00:27:59,980 --> 00:28:00,980
It's only fed text.
316
00:28:00,980 --> 00:28:01,980
So, the video is not feeding LeoAI.
317
00:28:01,980 --> 00:28:05,800
I say right now, I presume this is a cost factor. brown-cast-s weave communion
318
00:28:05,800 --> 00:28:09,800
I say right now, I presume this is a cost factor.
319
00:28:09,800 --> 00:28:12,740
Video is very expensive in all capacities.
320
00:28:12,740 --> 00:28:14,420
mar 복 consorti
321
00:28:14,420 --> 00:28:15,980
but cost come down over time.
322
00:28:15,980 --> 00:28:17,400
So, I would not be surprised if, in a year.
323
00:28:17,400 --> 00:28:18,400
You want to be cheap?
324
00:28:18,400 --> 00:28:18,640
You want to be super mówi is super, super, super cheap.
325
00:28:18,640 --> 00:28:19,100
Yeah.
326
00:28:19,100 --> 00:28:28,160
maybe year and a half, where LeoAI is multimodal, where it is capable of being trained on all of
327
00:28:28,160 --> 00:28:37,420
what we've done, all these shorts. Wouldn't surprise me. In fact, there will be a certain
328
00:28:37,420 --> 00:28:42,520
point in time where that's the case, because the compute will get down low enough where we can
329
00:28:42,520 --> 00:28:49,020
afford to train it on the video we have. Whether it's a year, whether it's three years, I don't
330
00:28:49,020 --> 00:29:01,040
know. Now, what I've done is I've taken to having a script, and thank you, Mike, possibly, he put
331
00:29:01,040 --> 00:29:11,600
together a script for me, and I started this yesterday, where I take my shorts and I feed the
332
00:29:11,600 --> 00:29:12,480
video through.
333
00:29:12,520 --> 00:29:21,740
This script, and it gives a transcript. Now that's on the blockchain. The video's not, as we know,
334
00:29:21,800 --> 00:29:29,780
the video's off on IPFS, but now the transcript is on the blockchain. So what I said is actually
335
00:29:29,780 --> 00:29:38,320
there. And I've actually been, in some instances, contextualizing it by running it through Leo
336
00:29:38,320 --> 00:29:42,320
Linker. And then that...
337
00:29:42,520 --> 00:29:58,780
That links the words to the particular page. So now there is a direct path to the word technology or to the
338
00:29:58,780 --> 00:30:06,360
word artificial intelligence, words artificial intelligence, or whatever it is. Hive. It helps to
339
00:30:06,360 --> 00:30:10,180
contextualize just what I was saying before.
340
00:30:12,520 --> 00:30:18,580
I've also taken my chats, as I said, with Chat Claude and some of them, and posted them on
341
00:30:18,580 --> 00:30:26,240
chain. Yes, it's synthetic data, but it's synthetic data that's on Anthropic servers if I'm on
342
00:30:26,240 --> 00:30:35,040
Claude. Now, it's synthetic data that's on the blockchain. And I do this mostly in the Threadcast,
343
00:30:36,080 --> 00:30:41,840
because I just happen to do it with technology. But I did it in the Brazilian one. I did it... I
344
00:30:41,840 --> 00:30:42,320
forget.
345
00:30:42,320 --> 00:30:43,960
oh, I did it on the GDP of Brazil
346
00:30:43,960 --> 00:30:46,620
and what are the biggest businesses.
347
00:30:46,800 --> 00:30:48,920
I just did a couple of quick questions
348
00:30:48,920 --> 00:30:51,200
and pasted them in the Brazil Threadcast.
349
00:30:54,000 --> 00:30:55,080
There's the information.
350
00:30:55,680 --> 00:30:56,380
There's the data.
351
00:30:57,120 --> 00:30:58,600
That'll be feeding LeoAI.
352
00:30:59,740 --> 00:31:02,120
Now, a lot of that stuff's already out there.
353
00:31:03,940 --> 00:31:06,280
But see, again, we have to remember
354
00:31:06,280 --> 00:31:08,020
it's one thing about data.
355
00:31:08,940 --> 00:31:11,380
It's another thing to get to the point
356
00:31:11,380 --> 00:31:13,860
where we go from information to context.
357
00:31:14,780 --> 00:31:18,560
And context requires a wide variety of stuff.
358
00:31:19,300 --> 00:31:21,880
These models are learning the same way we learn.
359
00:31:22,640 --> 00:31:24,540
If you're just given a bit of information
360
00:31:24,540 --> 00:31:27,260
and you have no context to anything,
361
00:31:28,420 --> 00:31:32,620
let's say a word from, let's say, astrophysics.
362
00:31:34,000 --> 00:31:36,140
Well, if you don't know anything about astrophysics,
363
00:31:36,180 --> 00:31:36,820
which I don't,
364
00:31:37,400 --> 00:31:39,920
all of a sudden somebody throws out some dude's name
365
00:31:39,920 --> 00:31:41,360
or some theorem or something
366
00:31:41,360 --> 00:31:42,940
and it means nothing to us.
367
00:31:43,060 --> 00:31:44,020
We have no reference.
368
00:31:47,200 --> 00:31:47,600
Astro.
369
00:31:47,720 --> 00:31:49,320
Oh, that was the dog on the Jetsons.
370
00:31:49,420 --> 00:31:51,260
Okay, I know what you're talking about.
371
00:31:51,580 --> 00:31:52,920
I mean, that's my reference point.
372
00:31:54,820 --> 00:31:57,940
So without any reference points, we're lost.
373
00:31:58,340 --> 00:31:58,900
Well, guess what?
374
00:31:58,960 --> 00:31:59,540
So is the AI.
375
00:32:01,200 --> 00:32:04,360
And that's where not only do we want more data,
376
00:32:04,360 --> 00:32:07,900
but we want different types of data.
377
00:32:08,980 --> 00:32:11,040
And also, I've said this before in different languages.
378
00:32:11,360 --> 00:32:12,660
I've said this before in different languages.
379
00:32:12,700 --> 00:32:14,880
Because even if you take something in English
380
00:32:14,880 --> 00:32:19,140
and translate it to Spanish or Portuguese or Italian or French or whatever,
381
00:32:20,760 --> 00:32:23,180
there's nuances in each of these languages
382
00:32:23,180 --> 00:32:27,320
that Google Translate or whatever is used may not pick up.
383
00:32:28,320 --> 00:32:30,300
And so if somebody takes like my transcript,
384
00:32:30,460 --> 00:32:32,980
like Coyote Alation was taking my transcripts
385
00:32:32,980 --> 00:32:35,140
and translating them into Portuguese,
386
00:32:36,140 --> 00:32:39,220
he can skim through and see where it makes sense
387
00:32:39,220 --> 00:32:40,980
and where, okay, it's screwed up here.
388
00:32:40,980 --> 00:32:41,280
This.
389
00:32:41,480 --> 00:32:42,760
This is the proper work.
390
00:32:43,440 --> 00:32:45,300
Because even though technically that's correct,
391
00:32:45,380 --> 00:32:46,420
that's not how people speak.
392
00:32:46,760 --> 00:32:47,840
That's not how people read.
393
00:32:47,920 --> 00:32:49,180
That's not how the language goes.
394
00:32:50,600 --> 00:32:51,900
We see that all the time.
395
00:32:52,900 --> 00:32:55,160
You know, you say stuff in English.
396
00:32:55,580 --> 00:32:58,460
Hell, I won't understand it if you're using Shakespeare in English.
397
00:32:59,380 --> 00:33:00,580
What the hell are you talking about?
398
00:33:02,220 --> 00:33:04,020
I mean, it's not that the person's incorrect.
399
00:33:04,020 --> 00:33:05,240
They're speaking English.
400
00:33:05,740 --> 00:33:09,720
They're just speaking a form of English that hasn't been really used.
401
00:33:10,040 --> 00:33:10,280
Prevent.
402
00:33:11,360 --> 00:33:13,400
Primarily in the last 400 years.
403
00:33:16,980 --> 00:33:21,640
So that's one thing to keep in mind is all that we're doing
404
00:33:21,640 --> 00:33:25,460
is we're feeding context.
405
00:33:26,520 --> 00:33:30,320
And this is one of the reasons why I'm excited for News2Thread 2.0.
406
00:33:31,040 --> 00:33:32,040
One of the reasons.
407
00:33:32,840 --> 00:33:36,020
And what News2Thread 2.0 is, if anybody doesn't know,
408
00:33:36,020 --> 00:33:40,020
is that's basically going to be a command.
409
00:33:41,360 --> 00:33:42,180
On InLeo.
410
00:33:42,740 --> 00:33:49,320
That allows you to essentially create thread storms out of articles.
411
00:33:50,740 --> 00:33:52,600
Non-Leo articles.
412
00:33:53,740 --> 00:33:56,420
So you could go to ESPN.
413
00:33:56,760 --> 00:34:01,340
You could go to Sky Sports, Sky News, Al Jazeera.
414
00:34:01,340 --> 00:34:05,840
Your, you know, Medium, Substack, wherever.
415
00:34:06,140 --> 00:34:09,420
You could go to any website that has articles on it.
416
00:34:10,520 --> 00:34:11,340
And you...
417
00:34:11,360 --> 00:34:15,820
You could take these articles, prompt it, and it'll create a thread storm.
418
00:34:17,160 --> 00:34:18,700
Won't reproduce the article.
419
00:34:18,920 --> 00:34:21,620
It'll create a thread storm, then put a link back to the article.
420
00:34:22,300 --> 00:34:26,200
So it'll create three, four, five threads underneath the main thread,
421
00:34:26,960 --> 00:34:28,360
including the link to the article.
422
00:34:30,780 --> 00:34:34,640
But if people are submitting that about a wide range of topics,
423
00:34:34,640 --> 00:34:39,820
from Medium, from Substack, from HomeandGarden.com, or wherever,
424
00:34:40,740 --> 00:34:41,060
you know,
425
00:34:42,260 --> 00:34:44,120
Rollingstone Magazine, you name it.
426
00:34:44,120 --> 00:34:48,200
It's getting some of that information.
427
00:34:48,540 --> 00:34:50,800
Albeit it's short little threads.
428
00:34:52,300 --> 00:34:56,860
But it's getting that information into the database and LeoAI is training on it.
429
00:34:59,240 --> 00:35:04,400
So, think of that capability where not only can we share the news,
430
00:35:04,400 --> 00:35:08,060
not only can we share the information, it helps to feed into it.
431
00:35:08,060 --> 00:35:09,460
And all of this is in an overall notebook.
432
00:35:09,460 --> 00:35:10,360
And then a different board like that.
433
00:35:10,360 --> 00:35:10,860
And that's...
434
00:35:10,860 --> 00:35:19,480
All of this is part of a larger initiative, as I said before, the digital, the decentralized
435
00:35:19,480 --> 00:35:22,560
digital brain.
436
00:35:22,560 --> 00:35:27,540
We are taking and putting stuff in digital form, whether it's in video, whether it's
437
00:35:27,540 --> 00:35:33,460
in audio, whatever our thoughts are, we're putting them in digital form.
438
00:35:33,460 --> 00:35:40,260
The question is, who owns it?
439
00:35:40,260 --> 00:35:45,980
Now personally, I want Leo to have a piece of this pie.
440
00:35:45,980 --> 00:35:49,520
That's why the technology Threadcast, not all me.
441
00:35:49,520 --> 00:35:52,680
As I said, Godfish is there every day.
442
00:35:52,680 --> 00:35:58,380
Other people are in there, Mike Possibly is in there, Coyotilation goes in there, Lisa
443
00:35:58,380 --> 00:36:00,860
Genta goes in there.
444
00:36:00,860 --> 00:36:04,940
But last time I checked, we were at about 120 comments.
445
00:36:04,940 --> 00:36:07,500
Some of them are just links to articles.
446
00:36:07,500 --> 00:36:09,300
Okay, that's information.
447
00:36:09,300 --> 00:36:10,240
It's cool.
448
00:36:10,240 --> 00:36:19,020
That's for utility for users, but a lot of it's information, a lot of it's data.
449
00:36:19,020 --> 00:36:26,160
And as I said, some of it's from Claude.
450
00:36:26,160 --> 00:36:34,140
So this is the context that I operate from, and I hope other people start to operate from.
451
00:36:34,140 --> 00:36:39,760
And it's hard to see because we don't know what Khal's doing on the back end, but I did
452
00:36:39,760 --> 00:36:42,640
a short, which we'll end up posting later today.
453
00:36:42,640 --> 00:36:45,420
AI should be the focus of Leo.
454
00:36:45,420 --> 00:36:52,040
Now I say that, and I will readily admit, I don't know if it is or isn't.
455
00:36:52,040 --> 00:36:54,500
I don't know the training schedule.
456
00:36:54,500 --> 00:37:00,120
I don't know how long it takes them for them to train their model.
457
00:37:00,120 --> 00:37:01,880
I have no idea.
458
00:37:01,880 --> 00:37:06,880
So I cannot speak intelligently about that.
459
00:37:06,880 --> 00:37:07,640
So maybe they are rushing it out and putting it in.
460
00:37:07,640 --> 00:37:08,640
I don't know.
461
00:37:08,640 --> 00:37:09,640
I don't know.
462
00:37:09,640 --> 00:37:13,100
Maybe they're rushing it out as quickly as they can.
463
00:37:13,100 --> 00:37:15,880
They just have to wait for the training.
464
00:37:15,880 --> 00:37:16,880
I mean, that's the way it goes.
465
00:37:16,880 --> 00:37:18,760
You can't speed this stuff up.
466
00:37:18,760 --> 00:37:25,000
I mean, unless you just go get more compute, but that requires more money.
467
00:37:25,000 --> 00:37:31,660
So there's always a cost benefit factor there.
468
00:37:31,660 --> 00:37:39,400
So when we start to look at this and we start to look at the value of Leo, first off, I
469
00:37:39,400 --> 00:37:48,000
will tell you, in three years, at most, any platform that does not have AI interwoven
470
00:37:48,000 --> 00:37:52,460
through it is dead.
471
00:37:52,460 --> 00:38:01,840
They're not even going to be useful because it's going to be the norm.
472
00:38:01,840 --> 00:38:08,360
And the earlier you get in, the easier it is, the more you can do, the more advanced
473
00:38:08,360 --> 00:38:09,360
you are.
474
00:38:09,360 --> 00:38:14,460
The more data you provide, the better it is.
475
00:38:14,460 --> 00:38:21,700
And so as the value of Leo goes up, the ability to scale becomes greater.
476
00:38:21,700 --> 00:38:24,500
It becomes a self-feeding process.
477
00:38:24,500 --> 00:38:31,340
That's one of the reasons why I was so excited when Khal said, LeoAI will have a prompt where
478
00:38:31,340 --> 00:38:36,980
you can prompt, just like you do on ChatGPT or any of the others, because that creates
479
00:38:36,980 --> 00:38:38,600
more data, right?
480
00:38:38,600 --> 00:38:39,600
Within their ecosystem.
481
00:38:39,600 --> 00:38:47,200
And of course, we can take that and just paste that onto Leo so it's out there for everybody.
482
00:38:47,200 --> 00:38:50,880
But at least within Leo, we have this.
483
00:38:50,880 --> 00:38:54,680
And yes, it's synthetic data, and there's arguments between synthetic data and non-synthetic
484
00:38:54,680 --> 00:39:00,800
data where the synthetic data kind of diminishes in value over time each time you train through
485
00:39:00,800 --> 00:39:06,440
it, because you lose the real-world applicability.
486
00:39:06,440 --> 00:39:08,600
And nobody's really sure.
487
00:39:08,600 --> 00:39:09,720
If that's correct or not.
488
00:39:09,720 --> 00:39:12,920
But that's one of the theories out there, and there's a logic to it.
489
00:39:12,920 --> 00:39:17,600
So I'm not going to say that's untrue.
490
00:39:17,600 --> 00:39:22,220
But we also keep adding more real-world data.
491
00:39:22,220 --> 00:39:23,720
And that's the difference.
492
00:39:23,720 --> 00:39:27,400
Let me catch up on some of these questions, because there are some good ones earlier that
493
00:39:27,400 --> 00:39:29,600
I wanted to touch upon.
494
00:39:29,600 --> 00:39:35,320
Nifty, how do you think LeoAI could impact short content, if at all?
495
00:39:35,320 --> 00:39:36,600
I mentioned that shorts can actually affect short content.
496
00:39:36,600 --> 00:39:37,600
I think that's a good question.
497
00:39:37,600 --> 00:39:38,600
I think that's a good question.
498
00:39:38,600 --> 00:39:43,360
It could affect LeoAI over time when we go multimodal.
499
00:39:43,360 --> 00:39:51,100
The other way AI could affect shorts is if there is the capability to have every short
500
00:39:51,100 --> 00:39:55,460
that goes in have a transcript, just like YouTube does.
501
00:39:55,460 --> 00:39:56,840
Have a transcript.
502
00:39:56,840 --> 00:40:01,520
See, a lot of companies have scraped YouTube, although they don't admit it, because Google
503
00:40:01,520 --> 00:40:04,120
will sue the shit out of them.
504
00:40:04,120 --> 00:40:07,400
And they didn't scrape YouTube for the videos, because they don't have the money for that.
505
00:40:07,400 --> 00:40:08,400
They went and grabbed the transcript.
506
00:40:08,400 --> 00:40:11,240
Think of that.
507
00:40:11,240 --> 00:40:19,880
So this is a way that LeoAI could impact shorts.
508
00:40:19,880 --> 00:40:25,940
The other way that I envision in the future is video generator, where at some point in
509
00:40:25,940 --> 00:40:35,780
time it will be cost-effective enough to have an image generator on Leo, where you can create
510
00:40:35,780 --> 00:40:37,400
a 30-second AI video.
511
00:40:38,400 --> 00:40:41,360
It's not here now.
512
00:40:41,360 --> 00:40:47,520
In fact, that's mostly what you get for free on these larger video generation sites, which
513
00:40:47,520 --> 00:40:51,060
they've spent who knows how many billions getting to that point.
514
00:40:51,060 --> 00:40:54,740
And for free, they'll give you a 30-second video.
515
00:40:54,740 --> 00:40:56,900
You want like three minutes or five minutes.
516
00:40:56,900 --> 00:41:00,120
You got to pay them whatever dollars a month.
517
00:41:00,120 --> 00:41:05,960
But again, just like we've seen with all these others, over time that comes down.
518
00:41:05,960 --> 00:41:07,240
So maybe two years.
519
00:41:07,240 --> 00:41:08,240
Or maybe two years.
520
00:41:08,240 --> 00:41:13,200
Or maybe a year from now on those same platforms, you'll get a three-minute video for free.
521
00:41:13,200 --> 00:41:16,480
And you'll be able to do a 15-minute video for whatever the monthly fee is.
522
00:41:16,480 --> 00:41:20,860
Well, at some point in time, there's going to be open-source stuff that can be incorporated
523
00:41:20,860 --> 00:41:26,260
into LeoAI that, OK, maybe you can't create a 15-minute video, but you can create something
524
00:41:26,260 --> 00:41:28,620
for 30 seconds.
525
00:41:28,620 --> 00:41:33,820
So that's the second way that I see.
526
00:41:33,820 --> 00:41:36,200
The third way affecting shorts in particular?
527
00:41:36,200 --> 00:41:37,200
I don't know.
528
00:41:37,200 --> 00:41:38,200
I don't know.
529
00:41:38,200 --> 00:41:49,860
If it has the transcripts, it'll affect shorts in the ability to search, to find, and also
530
00:41:49,860 --> 00:41:51,960
feed in for the generation.
531
00:41:51,960 --> 00:41:57,960
So I think all of that will be helpful.
532
00:41:57,960 --> 00:42:04,800
Because don't forget, LeoAI will index every threat.
533
00:42:04,800 --> 00:42:08,200
And that's another thing compared to the external world.
534
00:42:08,200 --> 00:42:12,960
The external world is not indexing threats for the most part.
535
00:42:12,960 --> 00:42:14,780
LeoAI is.
536
00:42:14,780 --> 00:42:20,700
So if you have a short, and it has its own URL, which that's what Khal said will be out
537
00:42:20,700 --> 00:42:21,960
in the next day or so.
538
00:42:21,960 --> 00:42:23,440
And he said that yesterday.
539
00:42:23,440 --> 00:42:24,440
So it may be out.
540
00:42:24,440 --> 00:42:26,240
I haven't looked.
541
00:42:26,240 --> 00:42:30,960
That's going to be very important for LeoAI, because it'll be able to reference it.
542
00:42:30,960 --> 00:42:34,860
It'll be able to direct people to that.
543
00:42:34,860 --> 00:42:37,200
So that's how I see it.
544
00:42:37,200 --> 00:42:38,200
Tokenized.
545
00:42:38,200 --> 00:42:42,820
Using normal AI agents to ask stuff about Hive is not very good, because it gets confused
546
00:42:42,820 --> 00:42:46,740
with other blockchain company called Hive.
547
00:42:46,740 --> 00:42:49,020
LeoAI should be a lot more accurate in that sense.
548
00:42:49,020 --> 00:42:52,700
It'll be trained with more directed data.
549
00:42:52,700 --> 00:43:02,260
Yes, that is true based AI.
550
00:43:02,260 --> 00:43:05,800
They got an image generator, which isn't very good, and they got a chatbot.
551
00:43:05,800 --> 00:43:06,800
Yeah.
552
00:43:06,800 --> 00:43:11,580
And I asked them about Hive Blockchain.
553
00:43:11,580 --> 00:43:16,220
And they were right on about not the Hive Blockchain LTD.
554
00:43:16,220 --> 00:43:22,580
They came back exactly what Hive is, a fork of steam, decentralized, blah, blah, blah.
555
00:43:22,580 --> 00:43:23,580
Why?
556
00:43:23,580 --> 00:43:27,380
This was trained on blockchain data.
557
00:43:27,380 --> 00:43:29,040
So they grabbed it directly.
558
00:43:29,040 --> 00:43:32,840
I can guarantee you they grabbed it directly from Hive or steam.
559
00:43:32,840 --> 00:43:36,420
Well, probably Hive because steam didn't really touch.
560
00:43:36,420 --> 00:43:41,720
So they grabbed it right from our blockchain, because they trained it on all blockchain,
561
00:43:41,720 --> 00:43:43,960
I presume all blockchain data.
562
00:43:43,960 --> 00:43:45,520
And it was spot on.
563
00:43:45,520 --> 00:43:52,320
It gave me an itemized list of it, decentralized, proof of stake, witness, blah, blah, blah.
564
00:43:52,320 --> 00:43:57,980
Now to explain how this works and the updates, and I think there was a question about the
565
00:43:57,980 --> 00:44:00,320
update.
566
00:44:00,320 --> 00:44:05,200
If I'm correct, LeoAI is being trained on Claude 3.0.
567
00:44:05,200 --> 00:44:06,200
Not Claude.
568
00:44:06,200 --> 00:44:08,480
Excuse me, Lama 3.1.
569
00:44:08,480 --> 00:44:10,960
Lama 3.1 is meta.
570
00:44:10,960 --> 00:44:15,380
Lama 3.1 updates immediately.
571
00:44:15,380 --> 00:44:22,120
It's continually updated on metadata.
572
00:44:22,120 --> 00:44:29,380
It's not updated immediately on internet data.
573
00:44:29,380 --> 00:44:35,100
That's why when you go to a Claude, when you go to some of these others and you ask a question,
574
00:44:35,100 --> 00:44:36,100
okay.
575
00:44:36,100 --> 00:44:43,380
They'll say, as of the last time I was updated, or as of the last update I got in April of
576
00:44:43,380 --> 00:44:52,040
2024, their training, they only go through, they only update when it's updated.
577
00:44:52,040 --> 00:44:55,180
They don't have real time data.
578
00:44:55,180 --> 00:44:59,860
Meta does off Meta's platform.
579
00:44:59,860 --> 00:45:04,380
An update to Wikipedia is not going to be in Meta 3.1.
580
00:45:04,380 --> 00:45:05,420
Okay.
581
00:45:05,420 --> 00:45:11,520
It'll be in Meta 3.0, but Meta 3.1, it won't update.
582
00:45:11,520 --> 00:45:17,200
Now if you want the score of last night's baseball game between team A and team B, if
583
00:45:17,200 --> 00:45:26,740
it was posted somewhere on one of the Meta apps, which likely was, it's in there.
584
00:45:26,740 --> 00:45:32,900
So you could say, what's the score between team A and team B, and it'll give it to you.
585
00:45:32,900 --> 00:45:34,740
That's on the large language model side.
586
00:45:34,740 --> 00:45:41,940
On LeoAI side, every single thread instantly updates.
587
00:45:41,940 --> 00:45:47,800
And when I say instantly, it may be 30 seconds or whatever, but the information is in there.
588
00:45:47,800 --> 00:45:56,000
So if LeoAI was live and we were able to use it, you would be able to ask it, what
589
00:45:56,000 --> 00:46:02,540
was the score of last night's ball game, even if it did not appear on Facebook, if tokenized
590
00:46:02,540 --> 00:46:03,740
society put it on in a thread.
591
00:46:03,740 --> 00:46:04,740
Okay.
592
00:46:04,740 --> 00:46:09,620
It's already there.
593
00:46:09,620 --> 00:46:16,740
And yes, Leo stuff, stuff about Leo power, stuff about delegated proof of staff, stuff
594
00:46:16,740 --> 00:46:29,940
about a lot of that stuff may not be in Meta, but it's InLeo because we train on it.
595
00:46:29,940 --> 00:46:33,940
Lucio, we talked about her being fed up.
596
00:46:33,940 --> 00:46:34,540
Okay.
597
00:46:34,540 --> 00:46:35,540
Let's see what Morty has to say.
598
00:46:35,540 --> 00:46:36,540
I got to clear my screen here.
599
00:46:36,540 --> 00:46:37,540
Okay.
600
00:46:37,540 --> 00:46:51,960
Morty, also known as Liquid Eye, will InLeo AI, it's actually LeoAI, output be in any
601
00:46:51,960 --> 00:46:58,620
way or form restricted and can it be used for all sorts of NSFW violence, political
602
00:46:58,620 --> 00:47:01,540
and racist stuff like Grok?
603
00:47:01,540 --> 00:47:04,540
Well, it's trained up.
604
00:47:04,540 --> 00:47:13,180
So that stuff is restricted on Meta, I presume.
605
00:47:13,180 --> 00:47:18,240
I guess Meta is restricting stuff.
606
00:47:18,240 --> 00:47:28,360
On Leo, I don't know what the parameters are that they're setting when they write the algorithm.
607
00:47:28,360 --> 00:47:32,540
So that's going to be, A, dependent on that.
608
00:47:32,540 --> 00:47:33,540
B, it's also dependent on what they're doing.
609
00:47:33,540 --> 00:47:34,540
Okay.
610
00:47:34,540 --> 00:47:39,180
So it's going to be dependent on what is on Leo, well, on the high blockchain.
611
00:47:39,180 --> 00:47:51,320
So if there is that type of stuff on the high blockchain, then that might be fed into Leo
612
00:47:51,320 --> 00:47:53,000
AI.
613
00:47:53,000 --> 00:47:59,080
If it's not the high blockchain, which there's not a ton of it, there might be some NSFW
614
00:47:59,080 --> 00:48:01,140
stuff, but not even a lot of that.
615
00:48:01,140 --> 00:48:03,180
And don't forget, NSFW.
616
00:48:03,180 --> 00:48:04,180
Okay.
617
00:48:04,180 --> 00:48:05,100
It's a deal in picture.
618
00:48:05,100 --> 00:48:09,660
So if somebody puts up a nudie pic, that's not seen.
619
00:48:09,660 --> 00:48:10,940
That's not even on the blockchain.
620
00:48:10,940 --> 00:48:12,840
That's in the image server.
621
00:48:12,840 --> 00:48:20,400
So it'd have to be, I don't know, some type of erotic story or something.
622
00:48:20,400 --> 00:48:21,840
And maybe there's a few of them.
623
00:48:21,840 --> 00:48:29,900
Maybe there's a few erotic fiction writers who wrote on Peak D or something.
624
00:48:29,900 --> 00:48:31,360
That's about it.
625
00:48:31,360 --> 00:48:32,640
So the answer to the question is, I don't know exactly.
626
00:48:32,640 --> 00:48:33,640
I don't know exactly.
627
00:48:33,640 --> 00:48:37,500
I don't know exactly on the LeoAI end.
628
00:48:37,500 --> 00:48:38,620
There is the restriction.
629
00:48:38,620 --> 00:48:44,280
Whatever meta restricts, that's going to be restricted on Leo.
630
00:48:44,280 --> 00:48:47,080
You just can't get around.
631
00:48:47,080 --> 00:48:49,420
Winanda has experienced the same thing on X.
632
00:48:49,420 --> 00:48:57,000
So stay the hell off X, unless you're doing a spaces or just dropping links.
633
00:48:57,000 --> 00:48:58,180
I am the first one in.
634
00:48:58,180 --> 00:48:59,560
Hopefully, you answered the Threadcast.
635
00:48:59,560 --> 00:49:00,560
Do you start doing thread casts?
636
00:49:00,560 --> 00:49:01,560
Yes.
637
00:49:01,560 --> 00:49:02,560
Yeah.
638
00:49:02,560 --> 00:49:03,560
There you go.
639
00:49:03,560 --> 00:49:09,560
to see how this ties into the 100 million dollar market cap and how we have control of steering it
640
00:49:10,920 --> 00:49:17,560
lucille is Ecency down I no longer get noticed from there again uh okay that brings up what I
641
00:49:17,560 --> 00:49:23,400
wanted to mention we we've talked about the numbers and I threw out numbers earlier in
642
00:49:23,400 --> 00:49:30,520
the week I went in containers account eric threw out the latest numbers that he pulled through
643
00:49:30,520 --> 00:49:37,720
Hive sql and they were different they were dramatically lower in the container if you
644
00:49:37,720 --> 00:49:47,320
respond to something through Ecency or through peak d it's not InLeo so it doesn't count towards
645
00:49:47,320 --> 00:49:53,800
LEO numbers monthly weekly or any of it and it also doesn't count towards rewards so if you get
646
00:49:53,800 --> 00:50:00,120
an upvote your LEO is not paid out your Hive will be any Hive associated with now
647
00:50:00,520 --> 00:50:04,120
or any other token associated with count but LEO is not paid out
648
00:50:06,760 --> 00:50:11,720
and so you're doing yourself and the platform a disservice
649
00:50:12,760 --> 00:50:19,560
if you're if you're operating through Ecency and not through LEO if you can avoid it
650
00:50:19,560 --> 00:50:24,280
now obviously there's times that LEO's down there's times and this is one of the reasons
651
00:50:24,280 --> 00:50:30,120
why I I know at times if somebody replies to me out of Ecency and I'll click on it
652
00:50:30,760 --> 00:50:34,680
it gives me an error message I can't reply to I can't on the threat
653
00:50:37,400 --> 00:50:39,800
or the reply and I certainly can't vote on