Bounded Regret

It's not a fundamental limit. Google palm had 540B parameters as dense model. But it's a practical limit because models with over 1T would be extremely slow even on newest gpus. Even now, OpenAI has limit of 25 messages. You can read more here: https://bounded-regret.ghost.io/how-fast-can-we-perform-a-forward-pass…

There is actually some basis in concept. Emergent deception ... https://bounded-regret.ghost.io/emergent-deception-optimization

> There aren't really any current advances outside of sheer scale of input in the models That isn't true. https://ai.googleblog.com/2022/11/characterizing-emergent-phenomena-in… While one can correctly argue that this is a consequense of increasing size, no one realized that increasing the size of LLMs would give them reasoning ability in 2018. Indeed, before Minerva[1] came out expert forecasters predicted[2] a 12% improvement in the ability on the MATH benchmark[3]. Minerva improved it by 50% and PaLM improved it even more. [1] https://arxiv.org/abs/2206.14858 [2] https://bounded-regret.ghost.io/ai-forecasting-one-year-in/ [3] https://github.com/hendrycks/math

I find it more illuminating to look at resolved prediction markets for concrete benchmarks; many of which have resolved faster and more positively than expected. https://bounded-regret.ghost.io/ai-forecasting-one-year-in/ I guess we will have to see what happens, but starting an anti deep learning blog after the staggering progress of the last six months seems pretty wild...

> there is a possibility that for things like AI, with extra time comes the ability to better understand and build those defenses before they're needed. Or not, and damaging wrongheaded ideas will become a self-reinforcing (because safety! humanity is at stake!) orthodoxy, leaving us completely butt-naked before actual risks once somebody makes a sudden clandestine breakthrough. https://bounded-regret.ghost.io/ai-pause-will-likely-backfire-by-nora/ > We don’t need to speculate about what would happen to AI alignment research during a pause—we can look at the historical record. Before the launch of GPT-3 in 2020, the alignment community had nothing even remotely like a general intelligence to empirically study, and spent its time doing theoretical research, engaging in philosophical arguments on LessWrong, and occasionally performing toy experiments in reinforcement learning. > The Machine Intelligence Research Institute (MIRI), which was at the forefront of theoretical AI safety research during this period, has since admitted that its efforts have utterly failed. Other agendas, such as “assistance games”, are still being actively pursued but have not been significantly integrated into modern deep learning systems— see Rohin Shah’s review here, as well as Alex Turner’s comments here. Finally, Nick Bostrom’s argument in Superintelligence, that value specification is the fundamental challenge to safety, seems dubious in light of LLM's ability to perform commonsense…