Kolter: AI scaling limits are financial and practical, not physical
Zico Kolter · Zico Kolter: OpenAI's Newest Board Member on The Biggest Questions and Concerns in AI Safety | E1197 · 20VC with Harry Stebbings · Sep 4, 2024 · at 18:14
Zico Kolter, CMU machine learning professor and OpenAI board member, responds to claims that AI compute scaling has hit a wall of diminishing returns.
0:00 / 0:55exact quote · 55.0s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“I'm not really sure what the rationale is for saying that we've, we plateaued in the compute sense. Most scaling laws that I've seen certainly suggest it can keep going. It's more expensive. You could argue that it's by far just scaling is, may not be the most efficient way to achieve better results. And I actually think that's very likely true. There are other better ways you could argue to Achieve the same level of improvement than compute, but compute still does seem to be both A major factor, and B, still seems to improve things. So I'm not quite sure. It's more a calculus about also, you know, the monetary trade-offs of how much models will cost at inference time and how much they cost a train, all this kind of stuff. These are much more, I would say, kind of becoming more practical concerns than a concern about the actual limits of scaling.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Zico Kolter
Kolter: LLMs are demonstrably intelligent despite being next-word predictors
“Oftentimes I know people say, oh, well, AI is, it's just predicting words. That's all it's doing. Therefore it can't be intelligent. It can't be. And I think that's just. Demonstrably wrong. What I think is amazing, though, is the scientific fact that when you…”
Kolter: AI model performance is not plateauing in coding tasks
“The domain I use models for most probably is coding, and also doing things like transcribing lectures and stuff like this. On those tasks, I am absolutely not seeing plateauing gains. The latest models, they are notably better than the previous iteration, and …”
Kolter: AI models contain an unpatchable buffer overflow vulnerability
“This is sort of like these models have a buffer overflow in all of them that we know about, and most importantly, that we don't know how to patch and fix. We don't know how to fix this yet with models”
Kolter: AI safety should focus on practical risks, not sci-fi doom
“The first is that I think the vast majority of AI safety should not be about these topics. The vast majority should be about quite practical concerns we have on making systems safer, like the kind that I've talked with you about so far.”
Kolter: AI model architectures and Transformers no longer matter
“I, for a lot of my career, was thinking that model architectures really mattered, and by having clever, complex architectures and sub-modules inside architectures, you would have, that was the route to sort of better AI systems. For the most part, I don't beli…”
Kolter: Word prediction intelligence is the top scientific discovery in decades
“You can train word predictors, and they produce intelligent, coherent, long-form responses. This is one of the most notable, if not the most notable, scientific discovery of the past 1020 years. Maybe much longer than that, right?”