you should interact with your audience, sometimes check the comments
2026-08-13 16:53:41
7
K :
It’s all AI
2026-08-17 09:28:00
1
Jay Wren :
Next question after that which will succeed is what is success look like? Folks always compare this LMAI revolution to the Internet in the late 90s. I think that’s a good example. At the time folks thought that they would make money selling web browsers, and Web servers things that we take entirely for granted today but there was some money made for a few years selling web servers IBM web sphere Sunhead net sweet I think it was called. There was a bunch of Java startups that got bought and are owned by mostly IBM and Oracle these days so I think we can predict what such an important utility will do. The giant closed players like Anthropic and OpenAI will Flegel for the next five years or so and fail to ever become highly profitable if they are profitable at all, the margins will just be too low because of the open weight and open source model competition, and the models will be a lot like Web servers back then they might even still exist Microsoft IIS being the famous proprietary Web server that still exists. But no one will be paying directly for it. It’ll be part of some bundle because the cost just won’t be worth it as a standalone. It will be entirely commoditized. No one will really talk about it even like no one cares if you run nginx or Apache or lighttpd now. That’s just my prediction.
2026-08-17 12:17:05
1
d4vi35 :
Perhaps this is missing the key element: open source would have to include the training data that was used. We’re seeing improved LLM performance over the last 42 months due to technical developments, but also capability improvements due to larger amounts of training data. I’d argue that there have been no real improvements in cognitive capacity, only ‘intelligence’ in the 19th century ‘knowledge’ sense of the word.
Corollary is that the ‘bad behaviour’ that has been observed with recent models is also acquired from the training corpus. So if you want to put LLM -based software into critical situations, you’re going to have to vet the training data.
This is going to be fun…
2026-08-14 21:28:57
2
jeremy._ng :
We should be advocating for more open source releases when it comes to increased access and accountability in the sector that’s becoming increasingly obscure and opaque
2026-08-13 15:05:07
4
Buddyo Vlogs :
What is the difference between openSource, openWeights, and openAI?
2026-08-13 18:55:35
1
ned :
People you can ask the models anything. Now lets focus on that beautiful woman.
2026-08-18 11:28:46
1
MartinDarko :
that's cool
2026-08-16 13:50:56
1
Timothy McGuire :
So open weight is a gated foundation to help to hit the ground running
2026-08-14 00:42:57
1
user7808870931453 :
Open source?
2026-08-13 20:59:18
1
globalbeautyfinder :
I don’t think either one of them are going to dominate, but if one does dominate, it will probably be open source of open weight, but the big legacy models will continue because they will be able to do things that the smaller local models open source open weight won’t.
2026-08-13 15:08:48
3
LJDio :
You can play around with abliterated models, allowing you to use an edited version of those LLMs with those weights essentially blown away…pretty much making it uncensored.
2026-08-13 17:56:50
0
therealmapzs :
For some reason I can’t use picture-in-picture with your videos
2026-08-13 15:33:03
1
🇺🇸Kirbiie | Puts People On :
Interesting
2026-08-13 21:28:01
1
9Bapa :
sup?
2026-08-13 23:43:47
1
ClassicMiddleton :
I can't hear "proprietary" without now thinking of the home made meat girl💀
2026-08-13 23:15:07
1
Dildo Baggins :
tbh i just come by for the titties. no interaction needed.
2026-08-14 08:41:23
1
user6563919061635 :
I'm just trying to look at your cleavage
2026-08-15 02:03:17
2
John Roa :
😘
2026-08-13 18:28:57
1
Anthony Conklin :
👏
2026-08-16 04:04:29
1
To see more videos from user @cirquedusunny, please go to the Tikwm
homepage.