AI model

“But the last two percent will take a few decades.”

“And so look, as they start to work together, I think we'll see them, you know, completely rewrite some of these verticals, like movie production, or music production, right, or advertising, or we're seeing it now with video and creative arts, with, you know, s…”

“I don't think startups are going to be able to create the AI themselves, but they might be able to benefit from the APIs.”

“It's not fair use because in fair use, you have the ability to create derivative works on future platforms. And you are taking this person, the original content owner's ability to exploit that And you are co-opting it and you're doing it at scale. And that is …”

“I think that you'll see the emergence of these various models that are actually optimized for various ways of thinking or political leanings.”

“And I do think that AI models should be able to use the content that's available on the web. Under a fair use doctrine to train their models for the benefit of consumers.”

“I'd rather focus on is there copyright infringement in the output? Does the output mimic An original copyrighted source. And if it does, then that's where the infringement should lie. Not on the, how did you do it?”

“The truth is that the models don't actually hold the data that they're trained on. They develop a bunch of synthesized predictors that they learn what the prediction could or should be from the data.”

“It's like a race to the bottom on pricing for basic commoditizing services in the cloud, like data storage, right? And then whoever has the better, higher level services, whoever has access to better models or has a lower price on running models or on running …”

“And these are now becoming about 90% cheaper every year, which means every two years these things are going to be 99% cheaper and better.”

“And so I think what we're going to see is soon more of an obfuscation of the individual models and more of this general service type approach where the updates are happening in, in, in a more continuous fashion.”

“Well, first of all, I think AI models are predictive. I mean, at the end of the day, they're predicting the next set of texts and so forth. And when it comes to like your employee list or your customer list, you just want to have a source of truth. You don't w…”

“I think there are going to be tens of millions of models. Everybody's going to have a model.”

“Instead of having two or three models you rely on, you're going to rely on 30, or 40, or 50, and you're going to trade them off, and you're going to use effectively like an LLM router to, you know.”

“And if China is forced to engineer its way around it, yeah, Freberg, the answer to your question is they'll use these models to design chips that can be manufactured in simple ways, and they'll make simple stuff.”

“If you had asked most people in the industry a few weeks ago, how far behind is China on AI models, they would say six to 12 months. And now I think they might say something more like three to six months, right?”

“So there's only one of two stable solutions. Either you pay the copyright holders, which I actually think doesn't work, And the reason is because someone in China will crawl it, and they just dump the weights, right? So they can just crawl and dump the compres…”

“The problem is that the nature of training these models is highly centralized. They benefit from supercomputer clustered compute. So it's not clear how any decentralized model can compete.”

“I think if you trained on the open web, your model should be open source.”

“I think that just these tools are good enough where the net new incremental task-oriented role that would typically go to a new grad, a lot of that can be defrayed by these models.”

“When you allow these models to run over complicated tasks over long periods of time, The error rates compound to such a degree that the resulting output is not worthwhile.”

“In 20 years, we'll be able to know how a drug affects the body without ever actually testing it. And my prediction is actually Our AI models will be more predictive than today's clinical trials.”

“I think the cardinality of agents and models is very high. I think there will always be alpha to be achieved, you know, improved differentiation, improved outcomes by specializing to the use case. Now, it's great to start with the general models, but You will …”

“When a person reads a book or an article, you've gained great knowledge. That does not mean that you're violating copyright laws or have to make deals with every content provider”

“And it sounds crazy today, but I guarantee if you put it on Polymarket, there will be a date when this happens.”

“Mostly it's accurate, and they've It's gotten better over the last two years. I'd say substantially better.”

“We used to debate when. This is no longer a when moment. The market is very much in an if mode. Are these cash flows durable at all?”

“When I listen to what our teams are able to accomplish in a day or two weeks that would have taken them, you know, a few months or nine months time, you know, it's just amazing the speed of innovation.”

“Untying this Gordian knot, it's going to take decades to do it right.”

“I still believe models are going to become a utility layer. You'll be able to buy intelligence on the fly”

“And increasingly, it's understanding what your intent was. And if you have a chance to play with Fable or Five Six from OpenAI, increasingly, what, you don't have to get the prompt just right. You don't have to be a prompt whisperer.”

“The best thing that can happen to them is 500 different models proliferate, and they support all of them because they will make so much money at the silicon layer.”

“These models are getting commoditized much faster than anybody thought. And how do we know this? Because there is no meaningful sustained advantage once a model publishes their performance criteria. What you see is literally within weeks, other models, some op…”

“Knowledge can't be contained. It's diffuse. And so the form of copyright is very clear. The case law and copyright is very clear. I cannot lift text out of your book, reprint it, and claim it as my own. That is a violation of copyright. But my reading of your …”

“I do think fundamentally that the conversion of that data into what I would call knowledge and ultimately the ability to create new outcomes from that knowledge that are not copyrighted, that are not copies of the original material, I do think is going to end …”

“But at some point, when most of the code is generated by the model, there'll be some point in the future, say, 28 or 29 or twenty-thirty, these security holes won't exist because the errors that humans make won't be made by these models.”

“An open source ban is coming. They're not going to call it that. You know what they're going to say? They're going to say that we simply have to apply the same standards to open models that we apply to close.”

“I happen to Believe, having looked at this and thought about it, I think that it is possible. I think it is likely to happen soon.”

“The AIs are going to change soon. Right now, the AIs are trained on a corpus that privileges things like cell nature and science as the prestige journals of biology or in mathematics, it would be the Invenciones and the Al's and the Journal of the American Mat…”

“All these models For the most part of converging in capability for the 98 or 99% of tasks that matter.”