Has AI already achieved meaningful recursive self-improvement beyond what human-led model development alone provides?

“And because it, you use it over and over again, some of it is skills, some of it is going to be a memory, we're going to improve the memory. Okay, so that next time you use it, it's even better than last. Recursive self-improvement. You could also decide that you take all of this skills, all of this memory,”

“I think that that's already here.”

“I think that that's already here. So let's agree that these agents are dangerous. Imagine that you have non-agentic AI, that you have an air gap, you've got a, a computer that can answer and, math problems.”

“the first level is what we call the delegation. You can run recursive self-improvement loop, but it's not strictly better than human doing R&D. And, we believe that all the previous public results are sitting on this level, level zero. And, the level one is what we called net positive, where you run the self-improvement loop, and you found that”

“To us, we care about the outcome of, it actually bending the curve of, recursive self-impro, like, of R&D efforts. We see this is the first evidence that the curve actually starting, get bent, for a fully autonomous system that can improve itself consistently.”

“We're already past that point. We're already past the point, probably past several months ago, if not earlier, where the highest revenue, or at least future discounted revenue per token application of current frontier models in the frontier labs is to develop via recursive self-improvement even stronger models. We're past the event horizon at this point for that.”

“I think we're already there. I think recursive self-improvement, we're already in the RSI era. Both Anthropic and OpenAI have made a number of pretty detailed and quantitative public statements to that effect. So we're already in an era when RSI is driving the vast majority of research at this point.”

“I personally think that we're further away from that and that the things that the AIs keep getting better at, it's kind of, like, still verifiable stuff and the generalization has been less than, than people, like, expected or, or, or hoped for to, to, like, actually do the last mile of, of fully automating R&D.”

“although recursive self-improvement per se ... we're not there. But the dynamic has clearly kicked in of each generation of AI making a greater contribution to the, to the, to the next generation, and we're seeing the acceleration, you'd, you'd expect.”

“Ali Ghodsi from, Databricks had a great formulation of, well, okay, well, what would be the preconditions for that? And he had four preconditions. One is, look, the next model has to require fewer GPUs than the last. ... Zero of those four have been met.”

“To us, we care about the outcome of, it actually bending the curve of, recursive self-impro, like, of R&D efforts. We see this is the first evidence that the curve actually starting, get bent, for a fully autonomous system that can improve itself consistently.”

“Ali Ghodsi from, Databricks had a great formulation of, well, okay, well, what would be the preconditions for that? And he had four preconditions. One is, look, the next model has to require fewer GPUs than the last. ... Zero of those four have been met.”