661
00:50:48,740 --> 00:50:54,480
Do you go to individual websites and pull it? No. You send out one of these

662
00:50:54,480 --> 00:51:02,320
bots. You just use the crawler. And that's what these entities would do. And so they would send

663
00:51:02,320 --> 00:51:07,880
this out, academics, researchers, all this stuff. Well, now they're shutting it down.

664
00:51:08,740 --> 00:51:15,040
We saw X made big news because they put a limit on their APIs. A week later, Reddit makes news

665
00:51:15,040 --> 00:51:19,480
doing the same damn thing. And of course, everybody starts screaming, how could they do it?

666
00:51:19,480 --> 00:51:26,480
Well, they were doing it for this exact reason. So the idea that Facebook, that meta, even Elon

667
00:51:26,480 --> 00:51:33,340
Musk, he was talking about, they were talking about X and Grok. And he made a joke. He said,

668
00:51:33,440 --> 00:51:38,560
do you think anybody out there really hasn't scraped all of Twitter yet? Is there anybody,

669
00:51:38,980 --> 00:51:43,380
any of the big companies that everybody scraped Twitter? He knows it. But the difference is,

670
00:51:43,400 --> 00:51:48,360
is they're not going to get the real time updated information because they're being blocked now.

671
00:51:49,480 --> 00:51:56,280
Same thing with meta. Now, meta may have always had it locked down, but they're not letting it

672
00:51:56,280 --> 00:52:09,160
out. And by the way, neither are 500 of the top 2000 websites. They have it blocked. Why?

673
00:52:10,160 --> 00:52:16,060
Mostly ChatGPT. They're the main culprit of this. Because they went out and they got ahead

674
00:52:16,060 --> 00:52:19,460
of everybody else. They've been scraping data for years. They have none of their,

675
00:52:19,480 --> 00:52:24,860
really. And so they've been scraping the data and then they get ahead and they start with these

676
00:52:24,860 --> 00:52:29,940
language models and they start with some of this stuff that impresses people. They become the big

677
00:52:29,940 --> 00:52:37,800
name. They become the Coke or the Kleenex of the industry, where instead of saying tissue,

678
00:52:38,020 --> 00:52:42,680
people just say, get me a Kleenex. Oh, what do you have? I'll have a Coke instead of saying a soda

679
00:52:42,680 --> 00:52:47,160
or a pop. They became the name for,

680
00:52:49,480 --> 00:52:54,700
well, just like Google became the name for search. So what did you do?

681
00:52:56,780 --> 00:53:02,920
They started shutting it down because that's what OpenAI uses. It's a robot text.

682
00:53:05,000 --> 00:53:10,260
And this is only going to get worse. So the question is, and now OpenAI,

683
00:53:11,040 --> 00:53:16,320
and they're tied to Microsoft. They keep getting billion dollar, multi-billion dollar funding

684
00:53:16,320 --> 00:53:19,460
rounds. So they have the money to go out and now start doing that. And they're not going to get

685
00:53:19,480 --> 00:53:27,380
and these other publishers for their data. But what do startups do? What do other entities do?

686
00:53:28,200 --> 00:53:31,940
And this is why I keep going back to the democratization of data

687
00:53:31,940 --> 00:53:36,880
and what Leo can do for it. By posting stuff on chain,

688
00:53:37,480 --> 00:53:41,420
anybody can read it. Anybody can set up an API and pull it. Anybody.

689
00:53:44,000 --> 00:53:49,120
But we have to put it on there. And so what type of data,

690
00:53:49,240 --> 00:53:49,460
do you have to put in there? And so, you know, I think that's the question. I think the question is,

691
00:53:49,460 --> 00:53:58,260
that well we have to look at why is the data be what are you doing what are you posting let's say

692
00:53:58,260 --> 00:54:04,980
threads what are your threads you post a thread for engagement so that's one way one type of thread

693
00:54:04,980 --> 00:54:09,680
so you post a thread you ask a question that's for engagement you're looking for an answer

694
00:54:09,680 --> 00:54:17,020
you want people to engage you post a short it's for engagement okay you want people viewing you

695
00:54:17,020 --> 00:54:24,740
want people commenting whatever but then there's also data for the database this could be news

696
00:54:24,740 --> 00:54:30,740
now you want people to engage yeah that's great you know you post a news article people engage

697
00:54:30,740 --> 00:54:36,160
oh yeah that that that trump running his mouth again or that camilla running her mouth again or

698
00:54:36,160 --> 00:54:44,600
whatever so you want engagement but it also is data that you put in the database but there's

699
00:54:44,600 --> 00:54:46,740
also two type two other types of data

700
00:54:46,740 --> 00:54:53,260
or it breaks down into two waves in this realm you have one human and two synthetic

701
00:54:53,260 --> 00:55:02,440
human data is obviously stuff we post synthetic data is stuff that's created by the computer

702
00:55:02,820 --> 00:55:10,500
and synthetic data is becoming a big thing with these chatbots so you go to ChatGPT you say

703
00:55:10,500 --> 00:55:15,680
or any of them Grok with the q or Grok with the k or

704
00:55:15,680 --> 00:55:16,420
you know

705
00:55:16,740 --> 00:55:24,700
Gemini or Meta or ultimately LeoAI, and you'll be able to prompt in, give me a recipe for

706
00:55:24,700 --> 00:55:30,280
blueberry pie, and it'll spit out a recipe for blueberry pie.

707
00:55:30,280 --> 00:55:33,400
Now you better check it out to make sure it's worth a damn.

708
00:55:33,400 --> 00:55:39,720
But that data is synthetic, it was created by the computer.

709
00:55:39,720 --> 00:55:43,720
Now that could be posted on Hive, it could be posted through threads.

710
00:55:43,720 --> 00:55:53,280
And this is why I've said, when you do a prompt with a chatbot, copy it and post it in threads.

711
00:55:53,280 --> 00:56:01,140
In fact, Ahmad, if you scroll to the top and you see his icon up there, 1, 2, 3, 4, 5,

712
00:56:01,140 --> 00:56:11,520
the fifth one over from the left, and you click on it, he is setting up a daily Threadcast

713
00:56:11,520 --> 00:56:13,700
for his conversations with ChatGPT.

714
00:56:13,700 --> 00:56:21,300
He has a Threadcast where you can go in and post your conversations with these chatbots,

715
00:56:21,300 --> 00:56:25,660
Claude, whatever, just copy it, post it in there.

716
00:56:25,660 --> 00:56:30,480
This gets the synthetic data on chain.

717
00:56:30,480 --> 00:56:37,940
Because when you're in ChatGPT, that synthetic data is owned by OpenAI.

718
00:56:37,940 --> 00:56:41,260
When you do it on Gemini, what the result is Gemini, is Google.

719
00:56:41,260 --> 00:56:42,260
That's what you get.

720
00:56:42,260 --> 00:56:43,260
That's what you get.

721
00:56:43,260 --> 00:56:43,300
That's what you get.

722
00:56:43,700 --> 00:56:52,460
And the reason why we get it on Hive is that anybody can use it, number one.

723
00:56:52,460 --> 00:56:59,500
Number two, how about the hybrid of synthetic and human data?

724
00:56:59,500 --> 00:57:05,360
How about the posting of the recipe for blueberry pie?

725
00:57:05,360 --> 00:57:06,360
Somebody posts that.

726
00:57:06,360 --> 00:57:09,500
It came from ChatGPT or whatever.

727
00:57:09,500 --> 00:57:13,260
And I go in and I say, that's all well and good.

728
00:57:13,260 --> 00:57:20,260
But I found instead of a quarter cup of flour, you really need half a cup of flour.

729
00:57:20,260 --> 00:57:25,340
A quarter cup just makes the crust too thin.

730
00:57:25,340 --> 00:57:26,340
It doesn't work well.

731
00:57:26,340 --> 00:57:28,820
You need half a cup.

732
00:57:28,820 --> 00:57:38,220
Now I provided feedback to the synthetic data, simply by answering the thread.

733
00:57:38,220 --> 00:57:39,260
Simply by adding to.

734
00:57:39,260 --> 00:57:40,260
Or I could say, add to.

735
00:57:40,260 --> 00:57:41,260
Add to.

736
00:57:41,260 --> 00:57:42,260
Add to.

737
00:57:42,260 --> 00:57:43,260
Add to.

738
00:57:43,260 --> 00:57:44,260
Add to.

739
00:57:44,260 --> 00:57:45,260
Add to.

740
00:57:45,260 --> 00:57:46,260
Add to.

741
00:57:46,260 --> 00:57:57,620
As an option, you also may want to add a quarter teaspoon of cinnamon, or whatever.

742
00:57:57,620 --> 00:58:01,300
This is stuff that enhances the value of that synthetic data.

743
00:58:01,300 --> 00:58:06,640
Now there's great dispute among people involved in the AI world, whether synthetic data just

744
00:58:06,640 --> 00:58:12,260
has a degradation, because as you keep using it and you get, you get further derivatives

745
00:58:12,260 --> 00:58:13,260
of it.

746
00:58:13,260 --> 00:58:19,820
generation, you basically end up with it just creating its own feedback loop and it doesn't

747
00:58:19,820 --> 00:58:22,020
create anything new.

748
00:58:22,020 --> 00:58:23,820
Some people dispute that.

749
00:58:23,820 --> 00:58:26,020
Some people say that's the case.

750
00:58:26,020 --> 00:58:29,120
They're all knowledgeable AI experts.

751
00:58:29,120 --> 00:58:34,960
So I'm not going to say who's right or who's wrong, but we certainly can enhance it by

752
00:58:34,960 --> 00:58:39,320
simply engaging with what is posted on the chain.

753
00:58:39,320 --> 00:58:43,800
Whether it's human generated or whether it's synthetic generated, especially if it's synthetic

754
00:58:43,800 --> 00:58:48,440
generated, we can go in and say, I agree with this.

755
00:58:48,440 --> 00:58:55,760
I think this is right, but this is also where I think it's off base.

756
00:58:55,760 --> 00:58:58,780
And it brings that into play.

757
00:58:58,780 --> 00:59:06,260
Another thing that we can do is if you have PDFs, PDFs are not picked up by crawlers.

758
00:59:06,260 --> 00:59:08,920
They can't read them because they're images.

759
00:59:08,920 --> 00:59:09,120
So...

760
00:59:09,320 --> 00:59:15,980
So what we can do is if you have a PDF, there's PDF extractors that you can turn the PDF,

761
00:59:15,980 --> 00:59:18,860
what's on the PDF into text.

762
00:59:18,860 --> 00:59:24,640
Then that can be copied and put onto Leo.

763
00:59:24,640 --> 00:59:34,420
Again, if data is the new oil, where are we putting the oil?

764
00:59:34,420 --> 00:59:38,200
It may seem like a senseless thing to do.

765
00:59:38,200 --> 00:59:39,200
But as I've threaded...

766
00:59:39,200 --> 00:59:48,680
I've done it on a number of occasions, ChatGPT with an estimated value of $150 trillion started

767
00:59:48,680 --> 00:59:52,100
the process by collecting data.

768
00:59:52,100 --> 00:59:55,580
Without data, ChatGPT was going nowhere.

769
00:59:55,580 --> 00:59:58,640
It didn't matter how much money Elon Musk put up.

770
00:59:58,640 --> 01:00:03,300
It didn't matter how smart Sam Altman and Ilya were.

771
01:00:03,300 --> 01:00:05,680
It didn't matter the algorithms they wrote.

772
01:00:05,680 --> 01:00:09,080
It didn't matter how much compute they got.

773
01:00:09,080 --> 01:00:13,640
They were done without data.

774
01:00:13,640 --> 01:00:20,860
So they had to go and start collecting data, which, as was pointed out, probably meant...

775
01:00:20,860 --> 01:00:25,540
Which obviously meant going and scraping, sending crawlers out and scraping whatever

776
01:00:25,540 --> 01:00:29,340
data they could.

777
01:00:29,340 --> 01:00:31,400
This is the same way.

778
01:00:31,400 --> 01:00:38,080
If we want to build a platform, it is important to keep feeding data of all types.

779
01:00:38,080 --> 01:00:39,080
Okay?

780
01:00:39,080 --> 01:00:41,280
Now, and I couldn't find the thread.

781
01:00:41,280 --> 01:00:42,920
I was looking for Khal's thread.

782
01:00:42,920 --> 01:00:44,960
This is why we need LeoAI.

783
01:00:44,960 --> 01:00:46,660
But I couldn't find Khal's thread.

784
01:00:46,660 --> 01:00:51,920
I posted a thread, and with mine, it was just buried, so I couldn't find it in mine either.

785
01:00:51,920 --> 01:00:55,920
I posted a thread about something to this effect about training LeoAI and the different

786
01:00:55,920 --> 01:01:01,340
ways of inputting data and how LeoShorts and why I'm putting transcripts on them and

787
01:01:01,340 --> 01:01:02,720
blah, blah, blah.

788
01:01:02,720 --> 01:01:08,080
And Khal replied, and he said, I already see a difference on the back end.

789
01:01:08,080 --> 01:01:15,160
And from what task is posted.

790
01:01:15,160 --> 01:01:16,220
I don't know.

791
01:01:16,220 --> 01:01:18,680
He didn't elaborate any further.

792
01:01:18,680 --> 01:01:21,600
And I don't know exactly how that means.

793
01:01:21,600 --> 01:01:27,520
But I think what it is, is I'm posting a lot of AI technology stuff, and I'm just bombarding

794
01:01:27,520 --> 01:01:30,400
with it.

795
01:01:30,400 --> 01:01:37,940
And now LeoAI is being trained on this and being trained on stuff that meta is not necessarily

796
01:01:37,940 --> 01:01:39,500
picking up.

797
01:01:39,500 --> 01:01:44,800
Because don't forget, threads are not indexed.

798
01:01:44,800 --> 01:01:47,700
So it's nowhere.

799
01:01:47,700 --> 01:01:53,800
So what's on threads is on the Hive Blockchain.

800
01:01:53,800 --> 01:01:58,840
But somebody has to go through and consistently scrape the Hive Blockchain.

801
01:01:58,840 --> 01:02:01,640
Well, LeoAI does that.

802
01:02:01,640 --> 01:02:02,640
Will these other bots?

803
01:02:02,640 --> 01:02:06,880
No, because these other bots don't update in real time unless it's their platform.

804
01:02:06,880 --> 01:02:07,880
So what's on threads?

805
01:02:07,880 --> 01:02:15,640
That's the advantage we have with LeoAI.

806
01:02:15,640 --> 01:02:19,820
Now ultimately, a lot of this stuff could get picked up.

807
01:02:19,820 --> 01:02:27,100
Because quite frankly, many of these platforms are scraping Hive.

808
01:02:27,100 --> 01:02:32,700
I mean, I've done enough tests on different ones asking about what is Resource Credits

809
01:02:32,700 --> 01:02:33,780
on Hive?

810
01:02:33,780 --> 01:02:35,280
What is Hive Power?

811
01:02:35,280 --> 01:02:36,440
What is the Hive Blockchain?

812
01:02:36,440 --> 01:02:37,440
And it comes up.

813
01:02:37,440 --> 01:02:40,420
And it comes up with the answers.

814
01:02:40,420 --> 01:02:45,020
So they're scraping our blockchain.

815
01:02:45,020 --> 01:02:46,020
Which it's open.

816
01:02:46,020 --> 01:02:48,320
They can do.

817
01:02:48,320 --> 01:02:52,280
But it's not necessarily up to the minute stuff.

818
01:02:52,280 --> 01:03:01,500
So this is where I would say we have to focus our attention.

819
01:03:01,500 --> 01:03:05,200
So what is it that feeds LeoAI?

820
01:03:05,200 --> 01:03:06,280
What is it?

821
01:03:06,280 --> 01:03:07,280
What is it?

822
01:03:07,280 --> 01:03:12,620
See, I see people posting videos.

823
01:03:12,620 --> 01:03:20,140
If you don't post a transcript, the video doesn't do anything.

824
01:03:20,140 --> 01:03:22,820
Now does a video do something for engagement?

825
01:03:22,820 --> 01:03:24,320
Maybe.

826
01:03:24,320 --> 01:03:25,960
Maybe a short off YouTube does.

827
01:03:25,960 --> 01:03:28,760
People watch shorts.

828
01:03:28,760 --> 01:03:31,700
Maybe a video of somebody on Hive.

829
01:03:31,700 --> 01:03:36,260
But for the most part, I see people posting hour long videos.

830
01:03:36,260 --> 01:03:37,300
With pump or somebody like that.

831
01:03:37,300 --> 01:03:49,200
Somebody on Thread is going to sit and watch an hour long video, a YouTube video?

832
01:03:49,200 --> 01:03:54,220
Maybe but not really.

833
01:03:54,220 --> 01:03:56,340
Somebody going to watch a YouTube short?

834
01:03:56,340 --> 01:03:58,080
Yeah, if it's something interesting.

835
01:03:58,080 --> 01:04:02,400
If it's something funny, yeah, maybe.

836
01:04:02,400 --> 01:04:03,600
But outside of that, what value does it have?

837
01:04:03,600 --> 01:04:04,600
I don't know.

838
01:04:04,600 --> 01:04:05,600
I don't know.

839
01:04:05,600 --> 01:04:06,840
I don't know.

840
01:04:06,840 --> 01:04:13,180
Now what I've done with a lot of YouTube videos, I'll post a video, especially in like the

841
01:04:13,180 --> 01:04:16,120
technology Threadcast, I'll post a technology video.

842
01:04:16,120 --> 01:04:19,420
You can only do this with like a 10 or 12 minute video.

843
01:04:19,420 --> 01:04:22,700
Then I go and take the transcript from YouTube.

844
01:04:22,700 --> 01:04:28,700
I copy the transcript, run it through chat through Claude and say, give me a summary

845
01:04:28,700 --> 01:04:30,820
of this transcript.

846
01:04:30,820 --> 01:04:32,300
And it'll give me a summary.

847
01:04:32,300 --> 01:04:35,600
And then I take the summary and I post it under the video.

848
01:04:35,600 --> 01:04:38,440
I don't post the whole transcript.

849
01:04:38,440 --> 01:04:46,280
I just post a summary of the transcript that I have Claude or Mehta or one of them create.

850
01:04:46,280 --> 01:04:53,120
Now the transcript, at least a summary of the transcript is online, is on the blockchain.

851
01:04:53,120 --> 01:04:56,360
The video isn't.

852
01:04:56,360 --> 01:05:00,060
That's why also with shorts, you'll notice I have a transcript under each short that

853
01:05:00,060 --> 01:05:01,600
I do.

854
01:05:01,600 --> 01:05:02,600
Why?

855
01:05:02,600 --> 01:05:04,140
Because that's on the blockchain.

856
01:05:04,140 --> 01:05:05,480
The short isn't.

857
01:05:05,600 --> 01:05:08,660
Short's on IPFS.

858
01:05:08,660 --> 01:05:17,500
Now I imagine at some point, multimodal will be able to read the IPFS shorts.

859
01:05:17,500 --> 01:05:25,580
The multimodal will not legally be able to read YouTube videos because that's against

860
01:05:25,580 --> 01:05:29,160
Google's term of service.

861
01:05:29,160 --> 01:05:31,220
Now does anybody care?

862
01:05:31,220 --> 01:05:32,220
Will anybody care?

863
01:05:32,220 --> 01:05:33,220
Who knows?

864
01:05:33,220 --> 01:05:34,220
All right.

865
01:05:34,220 --> 01:05:35,220
Thank you.

866
01:05:35,220 --> 01:05:43,920
It'll be a waste of ChatGPT Gemini Lama didn't make use of public database like I.

867
01:05:43,920 --> 01:05:45,000
They'll be stupid not to do it.

868
01:05:45,000 --> 01:05:48,260
Of course they're going to do it.

869
01:05:48,260 --> 01:05:52,900
That's why LeoAI must be fine-tuned in a special way to be more valuable for us than

870
01:05:52,900 --> 01:05:53,900
any of them.

871
01:05:53,900 --> 01:05:57,680
I need LeoAI to be as good as promise or better, otherwise I'll be worthless compared

872
01:05:57,680 --> 01:05:58,680
to the competition.

873
01:05:58,680 --> 01:06:04,280
Well, it can't be worthless compared to the competition because first off, it's trained

874
01:06:04,280 --> 01:06:06,280
on Lama.

875
01:06:06,280 --> 01:06:08,480
So it's a Lama-based model.

876
01:06:08,480 --> 01:06:12,180
Anything that's in Lama, when it goes through its training repetition, it's going to have

877
01:06:12,180 --> 01:06:14,180
Lama in there.

878
01:06:14,180 --> 01:06:18,500
Now, it's doubtful any of these have the updates.

879
01:06:18,500 --> 01:06:20,000
They will go through and scrape.

880
01:06:20,000 --> 01:06:24,340
When they do a major update, they'll get the information and update it.

881
01:06:24,340 --> 01:06:29,280
But LeoAI is updating in real time.

882
01:06:29,280 --> 01:06:32,960
And LeoAI is also pulling different information from the blockchain.

883
01:06:32,960 --> 01:06:34,060
Then it's not going to be the same.

884
01:06:34,060 --> 01:06:39,740
And what these other LLMs are doing.

885
01:06:39,740 --> 01:06:42,760
See this is the difference between the SLM and the LLM.

886
01:06:42,760 --> 01:06:45,660
Khal's talk about the copilot.

887
01:06:45,660 --> 01:06:52,000
And what they're doing essentially is they are building a way for you to pull data that

888
01:06:52,000 --> 01:06:56,360
is important to a Hive user, to a Leo user.

889
01:06:56,360 --> 01:07:03,260
So for example, what posts of mine have gotten the most upvotes?

890
01:07:03,260 --> 01:07:09,380
Will you get that from meta?

891
01:07:09,380 --> 01:07:11,580
Will meta care about that?

892
01:07:11,580 --> 01:07:15,540
Meta is probably just scraping what they can scrape.

893
01:07:15,540 --> 01:07:18,320
And they're not worried about configuring any of it.

894
01:07:18,320 --> 01:07:22,520
They just want the general baseline data to train their LLM.

895
01:07:22,520 --> 01:07:29,100
So basically, they're just adding into the language model.

896
01:07:29,100 --> 01:07:33,020
This is where SLMs are different than LLMs, because they can get very specific.

897
01:07:33,260 --> 01:07:44,160
So Khal can write the Copilot or the developers can write the Copilot type feature and say, OK, integrate this in based upon username, based upon all this other stuff.

898
01:07:44,780 --> 01:07:55,880
Now, you may be able to go to Meta and say, you know, give me, you know, who is Bradley Arrow and it'll be based upon what he did on Hive.

899
01:07:55,880 --> 01:08:00,260
There will be some type of general profile, but it won't be as specific.

900
01:08:00,260 --> 01:08:02,200
It won't be as nuanced.

901
01:08:05,620 --> 01:08:09,540
And so all of this is part of LeoAI.

902
01:08:09,780 --> 01:08:11,440
LeoAI is not just a Chapa.

903
01:08:12,380 --> 01:08:17,580
LeoAI is also going to be a Copilot type application.

904
01:08:18,100 --> 01:08:20,200
I hope LeoAI is an image generator.

905
01:08:20,200 --> 01:08:25,100
I hope LeoAI at some point is a video generator.

906
01:08:25,920 --> 01:08:30,240
LeoAI can incorporate and can train on different open source models.

907
01:08:30,260 --> 01:08:31,120
Other than Meta.

908
01:08:32,620 --> 01:08:42,820
So they can go and create on an image generator based on some other open source image generating software and incorporate that in if they want to.

909
01:08:43,820 --> 01:08:46,960
They can get very specific about what they do.

910
01:08:47,340 --> 01:08:52,960
LeoAI can also incorporate AI generation tools for content creators.

911
01:08:55,520 --> 01:09:00,080
LeoAI can also incorporate AI generated or AI.

912
01:09:00,260 --> 01:09:05,260
Editing tools for content creators, if they want.

913
01:09:07,260 --> 01:09:16,180
Anything you can do on Canva, technically, Leo could make that a part of LeoAI.

914
01:09:17,080 --> 01:09:18,060
I mean, it has to be built.

915
01:09:18,060 --> 01:09:27,820
It has to be incorporated in, but I hope these tools would be built in so that people could do stuff with their images.

916
01:09:27,820 --> 01:09:29,660
People could do stuff with their video.

917
01:09:29,660 --> 01:09:33,260
People could do stuff with their text.

918
01:09:34,500 --> 01:09:41,260
People can create a whole assortment of things and edit and expand and alter and change.

919
01:09:43,340 --> 01:09:49,840
Remix all that good stuff that you do on different platforms and see this is the key right now.

920
01:09:49,840 --> 01:09:51,600
You got to go to all these different things.

921
01:09:51,900 --> 01:09:58,540
So if you want to try this one thing on, you want to generate an image, you go to, you know, in ideogram or you go to this one.

922
01:09:58,540 --> 01:09:59,060
Or you go to this one.

923
01:09:59,060 --> 01:09:59,560
Or you go to this one.

924
01:09:59,560 --> 01:10:00,180
Or you go to that one.

925
01:10:00,520 --> 01:10:06,800
Now, you could go to Meta, but their image generation may not be as good as ideogram.

926
01:10:08,080 --> 01:10:13,640
You could go to this other platform and maybe they're not as good as that, but then you want to create video.

927
01:10:13,640 --> 01:10:15,640
Well, you got to go to a totally different platform.

928
01:10:16,680 --> 01:10:18,440
Then you want to edit an image.

929
01:10:18,460 --> 01:10:22,320
Well, they don't have any tools built on this one.

930
01:10:22,660 --> 01:10:28,200
So you got to go to Canva and you got to download it and go into Canva and do this, that or the other thing.

931
01:10:29,560 --> 01:10:42,680
And it's all over the place and see, this could be incorporated in, I mean, just look at the transcript right now.

932
01:10:42,680 --> 01:10:56,680
If you want, you can go and take your shorts and feed them through some type of audio transcript software and get a transcript.

933
01:10:56,680 --> 01:10:58,680
How's that?

934
01:10:59,560 --> 01:11:00,680
How do I do it?

935
01:11:01,360 --> 01:11:16,360
My possibly created, I don't know if he created for me, he shared with me a script that he created using ChatGPT.

936
01:11:17,360 --> 01:11:28,680
And so what I have to simply do, and I had to download some things on my laptop, but basically what happened was now I have the capability to just take

937
01:11:28,680 --> 01:11:38,400
a short, run it through this thing, and it automatically pulls the transcript, and then I manually post the transcript.

938
01:11:41,360 --> 01:11:58,480
He's also created scripts, which automatically post the transcript for certain videos, because he's taken a lot of my three-speak videos, fed them through, and then it just automatically posted the transcript.

939
01:11:58,680 --> 01:12:08,800
Because, again, the video is not in the database.

940
01:12:08,800 --> 01:12:12,360
See, this is part of the idea.

941
01:12:12,360 --> 01:12:15,520
Not only is it the Leo team building it, but it's what Ahmad did.

942
01:12:15,520 --> 01:12:17,360
It's what Mightpossibly is doing.

943
01:12:17,360 --> 01:12:20,360
It's what other people are doing.

944
01:12:20,360 --> 01:12:23,360
Look, there's the video.

945
01:12:23,360 --> 01:12:24,360
It's not in the database.

946
01:12:24,360 --> 01:12:25,360
It's not in the database.

947
01:12:25,360 --> 01:12:26,360
It's not in the database.

948
01:12:26,360 --> 01:12:27,360
It's not in the database.

949
01:12:27,360 --> 01:12:28,360
It's not in the database.

950
01:12:28,360 --> 01:12:33,760
There's Mightpossibly saying it, because Ahmad asked for it.

951
01:12:33,760 --> 01:12:42,360
I'm planning to post the code of the transcription script open source with some instructions on how to use it.

952
01:12:42,360 --> 01:12:44,700
There you go.

953
01:12:44,700 --> 01:12:47,380
This is the multiplication of efforts.

954
01:12:47,380 --> 01:12:48,380
He created it.

955
01:12:48,380 --> 01:12:51,020
He created it out of ChatGPT.

956
01:12:51,020 --> 01:12:52,020
Got me working it.

957
01:12:52,020 --> 01:12:53,400
I've tested it.

958
01:12:53,400 --> 01:12:58,040
I've done, I don't know, hundreds of shorts in the last month.

959
01:12:58,040 --> 01:13:05,040
And it's working like a champ.

960
01:13:05,040 --> 01:13:07,980
So the alpha stage went very well.

961
01:13:07,980 --> 01:13:13,000
Now if he posts the open source, somebody like Ahmad may be able to do something.

962
01:13:13,000 --> 01:13:15,440
Other people may be able to do something.

963
01:13:15,440 --> 01:13:19,040
It's a great idea.

964
01:13:19,040 --> 01:13:22,860
Ahmad, studying about AI generated tools.

965
01:13:22,860 --> 01:13:24,360
I'm interested in Hug & Face.

966
01:13:24,360 --> 01:13:27,040
There are many image generation tools available to use for public.

967
01:13:27,040 --> 01:13:28,040
The only downside?

968
01:13:28,040 --> 01:13:31,040
Is they're really slow, except when you pay.

969
01:13:31,040 --> 01:13:32,040
Yeah.

970
01:13:32,040 --> 01:13:38,420
And this is where the trade-off comes in.

971
01:13:38,420 --> 01:13:43,280
It's the same way with all of this AI stuff.

972
01:13:43,280 --> 01:13:44,280
You know, ideogram.

973
01:13:44,280 --> 01:13:53,540
When I started using ideogram, used to get 25 prompts a day, which was 100 images.

974
01:13:53,540 --> 01:13:57,980
I think they're down to 10 prompts a day or five prompts a day.

975
01:13:57,980 --> 01:14:00,040
I mean, they just keep declining.

976
01:14:00,040 --> 01:14:02,660
And if you want it, you got to pay for it.

977
01:14:02,660 --> 01:14:04,680
And they've gotten better with their images.

978
01:14:04,680 --> 01:14:08,220
They're a lot better with version two than version one.

979
01:14:08,220 --> 01:14:09,480
So you get limited.

980
01:14:09,480 --> 01:14:10,480
But guess what?

981
01:14:10,480 --> 01:14:14,860
There's other image generation tools out there.

982
01:14:14,860 --> 01:14:17,360
And there's more showing up.

983
01:14:17,360 --> 01:14:22,720
And this is why LeoAI, I think, needs to look at, incorporate all this stuff in.

984
01:14:22,720 --> 01:14:25,980
And this is why we're having a conversation here.

985
01:14:25,980 --> 01:14:26,980
And we're talking AI.

986
01:14:26,980 --> 01:14:28,480
We're talking AI tools.

987
01:14:28,480 --> 01:14:30,420
We're not talking social media.

988
01:14:30,420 --> 01:14:31,420
Yeah.

989
01:14:31,420 --> 01:14:32,800
These are used for social media.

990
01:14:32,800 --> 01:14:36,900
You're going to create an image and you're going to post it on Leo.

991
01:14:36,900 --> 01:14:38,540
Or you're going to post it wherever.

992
01:14:38,540 --> 01:14:45,880
I mean, for the past, I don't know how many months, most of the images in my blog posts

993
01:14:45,880 --> 01:14:47,780
are created in an ideogram.

994
01:14:47,780 --> 01:14:52,980
They're AI images.

995
01:14:52,980 --> 01:14:56,680
I mean, it's just, I mean, I don't think anybody goes to my posts for any of the pictures.

996
01:14:56,680 --> 01:14:57,800
They're the images.

997
01:14:57,800 --> 01:14:58,800
So who cares?

998
01:14:58,800 --> 01:15:04,240
It could be an image, AI-generated image, or it's an image I pull off Google Images

999
01:15:04,240 --> 01:15:07,000
or whatever, or pull off my search engine.

1000
01:15:07,000 --> 01:15:10,380
Just go to images and there, pull something off.

1001
01:15:10,380 --> 01:15:12,920
What's the difference?

1002
01:15:12,920 --> 01:15:19,420
Kind of just a filler, a spacer, and an atmosphere type thing.

1003
01:15:19,420 --> 01:15:22,800
So what's the difference if it's AI-generated?

1004
01:15:22,800 --> 01:15:25,120
But having these tools allows us to create more.

1005
01:15:25,120 --> 01:15:26,120
You know, I, you know, I don't know.

1006
01:15:26,120 --> 01:15:29,680
You know, I sometimes get an image and I'd like to lighten it up.

1007
01:15:29,680 --> 01:15:30,680
It's a perfect image.

1008
01:15:30,680 --> 01:15:33,880
I had one yesterday where, or a couple of days ago, it was a perfect image, but it was

1009
01:15:33,880 --> 01:15:35,480
in black and white for some reason.

1010
01:15:35,480 --> 01:15:36,920
And I said, I went and remixed it.

1011
01:15:36,920 --> 01:15:41,260
I said, add some color, add red, didn't come out right.

1012
01:15:41,260 --> 01:15:44,240
I would have liked to have been able to edit it, but I'm not going to send it to editing

1013
01:15:44,240 --> 01:15:45,240
software.

1014
01:15:45,240 --> 01:15:46,240
So it got close enough.

1015
01:15:46,240 --> 01:15:47,240
It was so-so.

1016
01:15:47,240 --> 01:15:48,240
So I said, okay, what the hell.

1017
01:15:48,240 --> 01:15:49,580
For what I'm doing, it's fine.

1018
01:15:49,580 --> 01:15:54,600
But if this was LeoAI and it was generated, I could have just added the right, adjusted

1019
01:15:54,600 --> 01:15:55,920
the color slightly.

1020
01:15:56,120 --> 01:15:59,080
That would have been great.

1021
01:15:59,080 --> 01:16:02,880
That would have been great.

1022
01:16:02,880 --> 01:16:07,880
Yeah, Marty didn't create it.

1023
01:16:07,880 --> 01:16:09,280
It was Mightpossibly.

1024
01:16:09,280 --> 01:16:12,120
Let me go down so I didn't miss anybody.

1025
01:16:12,120 --> 01:16:21,280
Um, you guys were active while I went on my tirade.

1026
01:16:21,280 --> 01:16:22,280
Ahmad's back.

1027
01:16:22,280 --> 01:16:23,280
Yeah.

1028
01:16:23,280 --> 01:16:25,280
Don't forget the dirty data.

1029
01:16:25,280 --> 01:16:26,120
Oh.

1030
01:16:26,120 --> 01:16:27,120
Dirty fish.

1031
01:16:27,120 --> 01:16:28,120
Yeah.

1032
01:16:28,120 --> 01:16:33,400
Well, we, we, we, it, with the subscription, maybe we could, uh, give it some adult content

1033
01:16:33,400 --> 01:16:41,320
adult, uh, topics.

1034
01:16:41,320 --> 01:16:46,680
We don't talk about that data.

1035
01:16:46,680 --> 01:16:48,880
Closing up Twitter for scraping was the right move.

1036
01:16:48,880 --> 01:16:55,080
Well, Ben, the reason why Twitter and, and Reddit, well already ready, closed it up or,

1037
01:16:55,080 --> 01:16:56,920
or, or reduced the ability to scrape.

1038
01:16:56,920 --> 01:17:00,520
You can still scrape it, but you can't do very much because they limit what you can

1039
01:17:00,520 --> 01:17:01,520
pull out.

1040
01:17:01,520 --> 01:17:07,240
Uh, and you'd have to use other, other protocols, uh, because they're blocking the robot text,

1041
01:17:07,240 --> 01:17:16,800
but Reddit turns around and sells their data to Google for 60 million bucks.

1042
01:17:16,800 --> 01:17:20,760
These companies know these companies knows there's a fortune in it.

1043
01:17:20,760 --> 01:17:24,080
And the thing is not only does adding data.

1044
01:17:24,080 --> 01:17:33,400
Build the value of Leo, uh, of Hive, but it democratizes the data because not only can

1045
01:17:33,400 --> 01:17:40,540
Facebook scrape it or Meta scrape it or Google scrape it, but ABC startup can.

1046
01:17:40,540 --> 01:17:45,920
That's important in my opinion.

1047
01:17:45,920 --> 01:17:49,920
I mean, Meta Facebook has it in its terms of service can use your data in any way they

1048
01:17:49,920 --> 01:17:50,920
want.

1049
01:17:50,920 --> 01:17:51,920
Yeah.

1050
01:17:51,920 --> 01:17:52,920
Well, I'm in.

1051
01:17:52,920 --> 01:17:53,920
Yeah.

1052
01:17:53,920 --> 01:17:54,760
It's not your data.

1053
01:17:54,760 --> 01:17:55,760
Your data.

1054
01:17:55,760 --> 01:17:58,040
It's not your data.

1055
01:17:58,040 --> 01:18:06,000
If you came into my house and gave me a cake and put a cake in my fridge, the fridge that

1056
01:18:06,000 --> 01:18:14,620
I paid for, I pay for the electricity and you gave me the cake and you leave and four

1057
01:18:14,620 --> 01:18:17,740
days later you come back and say, that's my cake.

1058
01:18:17,740 --> 01:18:19,240
I'll say you're full of shit.

1059
01:18:19,240 --> 01:18:21,440
It's in my fridge that I pay for that.

1060
01:18:21,440 --> 01:18:22,920
You gave, you gave me the cake.

1061
01:18:22,920 --> 01:18:23,920
Okay.

1062
01:18:23,920 --> 01:18:24,920
Not your cake.

1063
01:18:24,920 --> 01:18:25,920
It's not your data.

1064
01:18:25,920 --> 01:18:33,880
And by the way, the other thing with this saying, well, you use Hive, then you own your

1065
01:18:33,880 --> 01:18:34,880
data.

1066
01:18:34,880 --> 01:18:36,060
No, you don't.

1067
01:18:36,060 --> 01:18:40,280
If you own your data, take it back.

1068
01:18:40,280 --> 01:18:42,740
If it's yours, take it back.

1069
01:18:42,740 --> 01:18:43,740
You can't.

1070
01:18:43,740 --> 01:18:46,700
It's the public's data.

1071
01:18:46,700 --> 01:18:49,040
It's nobody's data.

1072
01:18:49,040 --> 01:18:50,200
Nobody controls it.

1073
01:18:50,200 --> 01:18:51,920
Nobody owns it.

1074
01:18:51,920 --> 01:18:52,920
It's public.

1075
01:18:52,920 --> 01:18:55,160
Everybody owns it.

1076
01:18:55,160 --> 01:18:56,160
It's shared.

1077
01:18:56,160 --> 01:18:57,080
You own your data.

1078
01:18:57,080 --> 01:18:58,160
You know it.

1079
01:18:58,160 --> 01:18:59,660
We don't own our data.

1080
01:18:59,660 --> 01:19:00,860
Nobody can control it.

1081
01:19:00,860 --> 01:19:03,420
You keep feeding it and impaling statistics.

1082
01:19:03,420 --> 01:19:07,960
That's why it's called That change the data.

1083
01:19:07,960 --> 01:19:09,300
It's not your data.

1084
01:19:09,300 --> 01:19:11,320
You can't change it.

1085
01:19:11,320 --> 01:19:12,700
Nobody owns it.

1086
01:19:12,700 --> 01:19:16,980
It's not in your data.

1087
01:19:16,980 --> 01:19:17,980
Nobody owns it.

1088
01:19:17,980 --> 01:19:18,880
You're living in the existence of your data.

1089
01:19:18,880 --> 01:19:19,880
You don't have information in your data.

1090
01:19:19,880 --> 01:19:20,880
Whatever you do in your data, you're not going to ever to onto private data.

1091
01:19:20,880 --> 01:19:30,720
I mean, I don't know if there's a line of thinking that says search engines, as we traditionally

1092
01:19:30,720 --> 01:19:32,500
know them, are going away.

1093
01:19:33,420 --> 01:19:37,420
I think my opinion is a lot of stuff's going away.

1094
01:19:38,660 --> 01:19:42,280
We are going to see generative AI disrupt so much stuff.

1095
01:19:43,680 --> 01:19:48,420
Generative AI is changing knowledge and information totally.

1096
01:19:48,420 --> 01:19:54,380
How we find it, how we relate to it, how we interact with it, completely.

1097
01:19:56,860 --> 01:20:00,680
Now, I don't feel too bad for Google, because guess what?

1098
01:20:01,100 --> 01:20:05,100
Google's right on top of everything else with Gemini, just like anyone else.

1099
01:20:05,680 --> 01:20:12,460
And I guarantee you Gemini 3 or whatever the hell the next Gemini is going to be, is going

1100
01:20:12,460 --> 01:20:18,180
to be light years ahead of this version of Gemini, especially with Google, because I

1101
01:20:18,180 --> 01:20:18,400
know that.

1102
01:20:18,400 --> 01:20:20,140
They have plenty of processing.

1103
01:20:23,760 --> 01:20:27,420
If you put it on chain, anybody can use an API and scrape it, which is a double-edged

1104
01:20:27,420 --> 01:20:27,960
sword, by the way.

1105
01:20:28,020 --> 01:20:29,140
No, it's not a double-edged sword.

1106
01:20:30,520 --> 01:20:31,300
It's out there.

1107
01:20:31,520 --> 01:20:32,080
It's the public.

1108
01:20:36,740 --> 01:20:38,100
Or you can thread threads.

1109
01:20:38,320 --> 01:20:38,520
Yeah.

1110
01:20:40,520 --> 01:20:47,520
Well, although, Fish, I think if the AIs are trained on Twitter and Facebook, the

1111
01:20:48,300 --> 01:20:48,320
way...

1112
01:20:48,400 --> 01:20:51,860
The way people behave on there, there's no peacefulness.

1113
01:20:51,980 --> 01:20:53,160
Those are cesspools.

1114
01:20:58,420 --> 01:21:01,120
Human data and synthetic.

1115
01:21:01,900 --> 01:21:05,500
Human-generated, created, and AI or chatbot-generated.

1116
01:21:05,620 --> 01:21:08,200
Yeah, that's what it is, and then you engage with that.

1117
01:21:09,100 --> 01:21:12,260
I saw the Watcher's question author, who that is, this journalist.

1118
01:21:12,260 --> 01:21:15,520
She shared her own article and published it in a news website.

1119
01:21:18,400 --> 01:21:22,640
Don't they have a community on Hive that was destroyed by Hive Watchers?

1120
01:21:26,220 --> 01:21:27,960
I don't know.

1121
01:21:28,160 --> 01:21:29,720
I mean, Biltcoin is...

1122
01:21:29,720 --> 01:21:31,160
That's going back to Biltcoin.

1123
01:21:33,280 --> 01:21:35,180
Are you baking now, Taskmaster?

1124
01:21:35,300 --> 01:21:40,640
No, I'm just making crap up in examples, and I don't know if you add more flour if it does

1125
01:21:40,640 --> 01:21:41,580
a damn thing or not.

1126
01:21:43,180 --> 01:21:45,080
That's not true for all PDFs.

1127
01:21:45,080 --> 01:21:46,080
Well...

1128
01:21:48,400 --> 01:21:53,760
There's a lot of data that is in PDF form that they want to try to get a hold of that

1129
01:21:53,760 --> 01:21:58,840
they claim is not available, so you tell me, Ben, how that is wrong.

1130
01:22:02,140 --> 01:22:06,500
Your lines, then, AMA is what convinced me to do the conversions with AI forecast.

1131
01:22:06,720 --> 01:22:10,340
I wasn't keeping up on putting machine-made data on blockchain under my name, but now

1132
01:22:10,340 --> 01:22:11,240
I understand the value.

1133
01:22:12,300 --> 01:22:17,840
No, your name or Bradley Arrow's name, put it under whoever...

1134
01:22:18,400 --> 01:22:27,560
Well, I mean, the PDF is...

1135
01:22:27,560 --> 01:22:33,620
Yeah, most of them are encrypted, and you can't copy and paste very easily, but, you

1136
01:22:33,620 --> 01:22:34,760
know, it depends on the PDF.

1137
01:22:42,760 --> 01:22:44,400
Food, talk, dry cash.

1138
01:22:44,460 --> 01:22:46,800
Yep, that's certainly there.

1139
01:22:47,860 --> 01:22:48,380
My phone.

1140
01:22:48,380 --> 01:22:52,660
I possibly said, yep, I spoke briefly with Aman, and he mentioned that LeoAI is indeed

1141
01:22:52,660 --> 01:22:56,080
picking up and integrating short transcripts in its outputs already.

1142
01:22:58,440 --> 01:23:05,460
I guess that's in relation to Khal commenting on that my stuff's already having an impact.

1143
01:23:06,640 --> 01:23:15,240
So, I guess Khal was specifically referencing the transcripts that I place in my shorts.

1144
01:23:15,240 --> 01:23:22,900
Since that is not showing up, since it's a video.

1145
01:23:23,460 --> 01:23:28,880
And again, I do this with a couple YouTube videos a day that I post.

1146
01:23:28,980 --> 01:23:31,080
Most of them are in the technology Threadcast.

1147
01:23:31,420 --> 01:23:33,580
But again, it's the same thing in the food Threadcast.

1148
01:23:33,680 --> 01:23:35,400
It's the same thing in the news Threadcast.

1149
01:23:36,400 --> 01:23:45,220
If you take a food video, and again, I'm not talking a half-hour food video.

1150
01:23:45,320 --> 01:23:47,480
I'm talking an eight or ten minute video.

1151
01:23:47,900 --> 01:23:51,900
And you put it up there, and you go grab the YouTube transcript, and you put it through

1152
01:23:51,900 --> 01:23:52,880
the chatbot.

1153
01:23:54,260 --> 01:23:56,680
And Claude seems to be the best that I've tried.

1154
01:23:57,780 --> 01:24:04,520
But Claude, of course, limits you, so, you know, you can only do so many in a day or

1155
01:24:04,520 --> 01:24:04,820
whatever.

1156
01:24:05,320 --> 01:24:10,120
But then you take it, and you put it under the video, in the comments under the video.

1157
01:24:11,420 --> 01:24:12,760
Now you have a summary.

1158
01:24:13,260 --> 01:24:15,020
It's not the transcript itself.

1159
01:24:15,020 --> 01:24:16,440
It's a summarized transcript.

1160
01:24:17,800 --> 01:24:20,580
It adds to it.

1161
01:24:25,400 --> 01:24:30,800
Ahmad, Leo could also be fine-tuned to do things that Lama can't do by itself.

1162
01:24:31,220 --> 01:24:34,600
It's not that complex, but it's a hard type of simple.

1163
01:24:35,780 --> 01:24:40,320
Yeah, Ahmad is absolutely correct.

1164
01:24:40,320 --> 01:24:43,240
What he's referring to, for those who don't know,

1165
01:24:43,560 --> 01:24:45,000
large language model.

1166
01:24:45,020 --> 01:24:52,960
Large language models are just that large, and they cover a wide range of things, and

1167
01:24:52,960 --> 01:24:54,700
they're not very focused.

1168
01:24:55,640 --> 01:25:04,120
They're trying to establish baselines and trying to reduce hallucinations, and they're

1169
01:25:04,120 --> 01:25:07,900
not very specific, and they don't deal with the specifics.

1170
01:25:07,900 --> 01:25:13,920
And what he said there is the type of hard stuff.

1171
01:25:13,920 --> 01:25:21,600
There's a paradox that stuff that's very hard for humans to do is very easy for machines

1172
01:25:21,600 --> 01:25:22,800
and vice versa.

1173
01:25:22,800 --> 01:25:27,440
For example, why are calculations easy for machines?

1174
01:25:27,440 --> 01:25:34,100
Machines can take rows and rows and rows of numbers and add them up with a simple formula

1175
01:25:34,100 --> 01:25:36,740
and do it in a nanosecond.

1176
01:25:36,740 --> 01:25:40,540
Take human forever to do that.

1177
01:25:40,540 --> 01:25:43,460
But then there's a lot of things that humans can do that machines.

1178
01:25:43,460 --> 01:25:46,260
It's just impossible for them to figure out.

1179
01:25:46,260 --> 01:25:49,220
And it's very, very difficult.

1180
01:25:49,220 --> 01:25:52,640
And so this is where Ahmad's absolutely correct.

1181
01:25:52,640 --> 01:25:58,600
You got a large language model, which the algorithms are trained to get general shifts

1182
01:25:58,600 --> 01:26:00,700
and general improvements.

1183
01:26:00,700 --> 01:26:06,400
But they don't worry about the specifics, the fine-tuned stuff that could be very, very

1184
01:26:06,400 --> 01:26:07,400
hard.

1185
01:26:07,400 --> 01:26:13,460
But this is where Leo could come in and take something and get very specific with what

1186
01:26:13,460 --> 01:26:21,280
it's trying to do to answer something and pull up specific information in specific

1187
01:26:21,280 --> 01:26:22,280
ways.

1188
01:26:22,280 --> 01:26:31,940
So for example, like Mightpossibly wrote a script and created a site to find out who

1189
01:26:31,940 --> 01:26:34,860
voted for you.

1190
01:26:34,860 --> 01:26:40,320
That's looking for very, very specific information and trying to produce a very specific result.

1191
01:26:40,320 --> 01:26:43,300
And if you try to go and do something like that with a meta.

1192
01:26:43,300 --> 01:26:50,900
A, the information is going to be up to date and B, you'll probably get a hodgepodge of

1193
01:26:50,900 --> 01:26:56,820
information because I don't even know if meta is worried, focusing upon that.

1194
01:26:56,820 --> 01:27:01,980
They're not constructing the data in such a way.

1195
01:27:01,980 --> 01:27:04,120
They have the raw data.

1196
01:27:04,120 --> 01:27:08,480
So the raw data of having everybody's votes doesn't mean anything.

1197
01:27:08,480 --> 01:27:13,000
What Mightpossibly had to do was he had to structure the data in a certain way.

1198
01:27:13,000 --> 01:27:20,740
Through the algorithm that he wrote, through the coding that he wrote to say, configure

1199
01:27:20,740 --> 01:27:26,800
the data this way so that when you say, okay, who voted for me on these dates or whatever

1200
01:27:26,800 --> 01:27:31,120
that the heck you're looking for, it can compile.

1201
01:27:31,120 --> 01:27:42,600
And that is what LeoAI could create.

1202
01:27:42,600 --> 01:27:47,080
Time is limited to five or 10 per day, depending on how complex it is.

1203
01:27:47,080 --> 01:27:59,580
You get four images per prompt, but they keep reducing the prompt.

1204
01:27:59,580 --> 01:28:02,600
For easy stuff now, you don't need a graphic designer, Ben.

1205
01:28:02,600 --> 01:28:03,600
Yeah.

1206
01:28:03,600 --> 01:28:06,940
I mean, think of that.

1207
01:28:06,940 --> 01:28:12,600
I did this with some of the movie stuff I'm working on.

1208
01:28:12,600 --> 01:28:17,100
And I went and I said, I picked a film and I forget what film it was.

1209
01:28:17,100 --> 01:28:23,220
And I said, I went to, I believe it was Ideogram.

1210
01:28:23,220 --> 01:28:30,000
And I don't remember if it was before Ideogram 2 or after it, but they've really improved

1211
01:28:30,000 --> 01:28:31,480
their text on Ideogram 2.

1212
01:28:31,480 --> 01:28:38,000
But I went in there and I said, create me a movie poster for, and I listed the film

1213
01:28:38,000 --> 01:28:40,960
in the year, so I made sure it had the right film.

1214
01:28:40,960 --> 01:28:42,600
Create me a movie poster for it.

1215
01:28:42,600 --> 01:28:47,960
That was the prompt, and it created me a movie poster.

1216
01:28:47,960 --> 01:28:56,560
And I posted it in threads and I said, how much did the studio spend to get this poster,

1217
01:28:56,560 --> 01:29:00,420
to get the poster created?

1218
01:29:00,420 --> 01:29:08,400
And maybe my poster, the Ideogram poster wasn't the greatest of posters, but it took like

1219
01:29:08,400 --> 01:29:12,100
not even a second and it was a prompt and it cost zero.

1220
01:29:12,100 --> 01:29:18,940
And I actually got four posters, but I picked the best one and there's a good chance two

1221
01:29:18,940 --> 01:29:20,640
or three of them, two of them suck.

1222
01:29:20,640 --> 01:29:24,320
But anyway, the point being, and we know this stuff's only going to get better, but what

1223
01:29:24,320 --> 01:29:26,660
Ben's saying is absolutely correct.

1224
01:29:26,660 --> 01:29:31,220
When I first got into marketing, I played around with a program, any of you may know

1225
01:29:31,220 --> 01:29:36,600
this, CorelDRAW, where you actually had to vector everything on your graphics.

1226
01:29:36,600 --> 01:29:42,100
It was computer-based, but you were vectoring and designing logos and everything.

1227
01:29:42,100 --> 01:29:43,700
I mean, it was a nightmare.

1228
01:29:43,700 --> 01:29:48,800
It took hours to do anything.

1229
01:29:48,800 --> 01:29:53,060
And that's how far we've come.

1230
01:29:53,060 --> 01:29:55,820
ChatGPT also have image generation now.

1231
01:29:55,820 --> 01:30:01,160
It's based on DALI, but for free users, you're limited three or so generations per day.

1232
01:30:01,160 --> 01:30:02,160
Yeah.

1233
01:30:02,160 --> 01:30:06,100
I mean, they're all going to try to get you to sign up.

1234
01:30:06,100 --> 01:30:07,420
That's what it comes down to.

1235
01:30:07,420 --> 01:30:11,940
And that's why I'm hoping a lot of this is incorporated InLeo AI.

1236
01:30:11,940 --> 01:30:17,720
See, this is getting back to Leo as an AI platform and the tokenomics.

1237
01:30:17,720 --> 01:30:20,980
We know how the tokenomics work, so I'm not going to go through that.

1238
01:30:20,980 --> 01:30:25,040
But a lot of it is based, or much of it's based on Leo Premium.

1239
01:30:25,040 --> 01:30:31,180
Well build Leo Premium like Amazon Prime built, huge amount of value InLeo Premium.

1240
01:30:31,180 --> 01:30:32,180
What is huge value?

1241
01:30:32,180 --> 01:30:33,180
AI tools.

1242
01:30:33,180 --> 01:30:36,680
Ahmad's not the only one paying attention to this stuff.

1243
01:30:36,680 --> 01:30:39,540
Mike Possibly's not the only one paying attention to this stuff.

1244
01:30:39,540 --> 01:30:41,540
So the more tools you build in and put it in one place...

1245
01:30:41,540 --> 01:30:46,280
The more value Leo Premium instantly becomes.

1246
01:30:46,280 --> 01:30:48,120
Instantly becomes.

1247
01:30:48,120 --> 01:30:55,140
So if we have LeoAI continually expanding and building more tools in, not only with

1248
01:30:55,140 --> 01:31:03,080
what Meta offers, because anything that Meta offers in its base model is going to be available

1249
01:31:03,080 --> 01:31:04,900
in terms of the training.

1250
01:31:04,900 --> 01:31:09,740
Now there may be agents and craft like that that end up tied to the Meta ecosystem since

1251
01:31:09,740 --> 01:31:10,740
it's open source.

1252
01:31:10,740 --> 01:31:11,340
Yeah.

1253
01:31:11,540 --> 01:31:13,440
They may not be able to be incorporated.

1254
01:31:13,440 --> 01:31:17,980
But then again, there may be agents that are open source that Khal and the team can say,

1255
01:31:17,980 --> 01:31:19,360
hey, look at what we can do.

1256
01:31:19,360 --> 01:31:25,420
We can pull this agent in and simply integrate it and provide this as part of LeoAI.

1257
01:31:25,420 --> 01:31:26,880
So that's the other event.

1258
01:31:26,880 --> 01:31:31,620
But all of this stuff feeds in and is in one place.

1259
01:31:31,620 --> 01:31:38,820
And if we're paying 10 HBD a month or whatever for it, there it is.

1260
01:31:38,820 --> 01:31:41,420
And you know, maybe ultimately there has to be...

1261
01:31:41,420 --> 01:31:47,500
There has to be a second layer of billing where, okay, you are part of Leo Premium level

1262
01:31:47,500 --> 01:31:52,280
A for 10 HBD a month, and I'm just making crap up, I'm making the numbers up.

1263
01:31:52,280 --> 01:31:55,480
But for 10 HBD, and this is what you get.

1264
01:31:55,480 --> 01:31:59,660
And you get your image generation, you get this tool, you get that tool, you get this,

1265
01:31:59,660 --> 01:32:02,600
you get that, you get the copilot, you get all this.

1266
01:32:02,600 --> 01:32:07,660
But then for another five HBD a month or another 10 HBD a month or whatever on top of it, you

1267
01:32:07,660 --> 01:32:11,220
get image generation or you get this or that or the other thing.

1268
01:32:11,220 --> 01:32:13,900
Or you get this, all these editing tools.

1269
01:32:13,900 --> 01:32:18,080
You can start to build different layers based upon what people need.

1270
01:32:18,080 --> 01:32:22,120
And people sit there and say, well, I don't give a crap about all that image, those editing

1271
01:32:22,120 --> 01:32:23,120
tools.

1272
01:32:23,120 --> 01:32:29,200
And I don't give a crap about video generation or I don't give a crap about 30-minute image

1273
01:32:29,200 --> 01:32:30,280
video generation.

1274
01:32:30,280 --> 01:32:35,620
Give me a 60-second video, I'm happy, I'm going to post it on Shorts, whatever.

1275
01:32:35,620 --> 01:32:41,220
And so you can end up expanding your offerings and expanding the game.

1276
01:32:41,220 --> 01:32:44,960
And cater your offerings to your audience.

1277
01:32:44,960 --> 01:32:51,900
And sit there and start to promote and market and say, we have this all here.

1278
01:32:51,900 --> 01:32:57,600
And don't think for a second that the other social media entities aren't going to do it.

1279
01:32:57,600 --> 01:33:01,040
Don't think for a second, Canva's trying to do something like this.

1280
01:33:01,040 --> 01:33:05,500
Canva's nickel and diamond everybody and they're trying to charge for everything up to scale,

1281
01:33:05,500 --> 01:33:06,500
which is fine.

1282
01:33:06,500 --> 01:33:08,400
I mean, if they get it, great.

1283
01:33:08,400 --> 01:33:10,200
Wonderful for them.

1284
01:33:10,200 --> 01:33:11,200
The problem is.

1285
01:33:11,200 --> 01:33:13,460
They are very isolated.

1286
01:33:13,460 --> 01:33:20,040
Why the hell do I want to take my image that I generate and go over to Canva and download

1287
01:33:20,040 --> 01:33:27,960
it, save it, move it, upload it, have it altered, then save it, download it.

1288
01:33:27,960 --> 01:33:30,780
I'm giving it to Canva again.

1289
01:33:30,780 --> 01:33:35,160
Then I'm giving them the image, helping them train their stuff, which why do I want to

1290
01:33:35,160 --> 01:33:38,940
do that when I can just do it all InLeo?

1291
01:33:38,940 --> 01:33:39,940
Now.

1292
01:33:39,940 --> 01:33:40,940
Is that clear?

1293
01:33:40,940 --> 01:33:43,000
Is that going to be very difficult to build?

1294
01:33:43,000 --> 01:33:46,200
I'm sure there's a lot of things that are going to be difficult to build, but we start

1295
01:33:46,200 --> 01:33:51,020
to see AI tools all over the place.

1296
01:33:51,020 --> 01:34:04,240
So did that?

1297
01:34:04,240 --> 01:34:06,940
Ahmad uses search engines 80% less now.

1298
01:34:06,940 --> 01:34:07,940
There you go.

1299
01:34:07,940 --> 01:34:08,940
He's a living example.

1300
01:34:08,940 --> 01:34:09,940
One of the reasons.

1301
01:34:09,940 --> 01:34:10,940
Yeah.

1302
01:34:10,940 --> 01:34:13,000
One of the reasons is bad first page results.

1303
01:34:13,000 --> 01:34:14,000
Yeah.

1304
01:34:14,000 --> 01:34:26,760
I mean, especially, yeah, they give you the advertising first.

1305
01:34:26,760 --> 01:34:31,700
You said you did over hundreds of videos with transcript so far.

1306
01:34:31,700 --> 01:34:37,300
That's about $1.50 in compute token so far, I just checked.

1307
01:34:37,300 --> 01:34:38,260
So there you go.

1308
01:34:38,260 --> 01:34:40,000
I mean, I've done...

1309
01:34:40,000 --> 01:34:52,140
I've done hundreds of videos and cost $1.50 in compute tokens.

1310
01:34:52,140 --> 01:34:55,060
I'll send you HBD, Mike, possibly.

1311
01:34:55,060 --> 01:35:03,080
Oh, yeah, okay, I see what you're saying, Ben, okay.

1312
01:35:03,080 --> 01:35:06,060
Many PDFs are image scans of books.

1313
01:35:06,060 --> 01:35:07,060
Yeah.

1314
01:35:07,060 --> 01:35:08,060
Scans are very difficult.

1315
01:35:08,060 --> 01:35:09,060
Yeah.

1316
01:35:09,060 --> 01:35:11,120
I mean, yeah.

1317
01:35:11,120 --> 01:35:13,780
Khal Marvel was late.

1318
01:35:13,780 --> 01:35:21,320
So is Bradley Arrow, and he said, what did I miss?

1319
01:35:21,320 --> 01:35:25,620
You can copy any text out of PDF document, image, whatever, via OCR.

1320
01:35:25,620 --> 01:35:28,880
Yeah, you have to OCR it, MoreTea.

1321
01:35:28,880 --> 01:35:33,180
And that's the problem with the crawlers is they're not OCRing.

1322
01:35:33,180 --> 01:35:36,300
They're not an OCR deal.

1323
01:35:36,300 --> 01:35:37,200
They just go straight.

1324
01:35:37,200 --> 01:35:38,200
Yeah.

1325
01:35:38,200 --> 01:35:40,960
So yeah, absolutely.

1326
01:35:40,960 --> 01:35:46,080
You go OCR it, and you'll be able to do it.

1327
01:35:46,080 --> 01:35:51,120
Another thing is, and I've talked about this before.

1328
01:35:51,120 --> 01:35:59,360
If you post a meme, and it has the text, well, again, LeoAI is not going to read it because

1329
01:35:59,360 --> 01:36:01,360
it's an image.

1330
01:36:01,360 --> 01:36:06,920
Brave, and I presume many of the browsers have it, but Brave on desktop, I can go right-click,