>>13623769not at all likely within the predictable future based on the kinds of models that are likely to do it first. they won't be as impressive as you imagine.
10-20 years out is all sci-fi to me, though. no use trying to predict it.
>>13624083>Google Black Rock accidentally made it while fiddling around with their stock trading algorithmsextremely unlikely
>AGI researched ceased at that momentcompletely retarded. you haven't been paying any attention if you think this. IBM's knowledge base creation research is still pathetic but ongoing. DeepMind is more actively involved.
>self driving cars, and story writing botsget media time because people think they're cool or directly applicable to themselves.
>do any research regarding evolutionary perspectives into the creation of intelligent machinesthe problem with evolutionary approaches is that we don't have the computation to support them. even OpenAI barely has access to enough GPUs to make them worthwhile for relatively simple RL problems. they're needed for some parts of the non-differentiable problems associated with long term memory creation, but they're a bottleneck and the goal is to minimize them. they're also just a honeypot for people who haven't done much research yet; they were big in the 60s and i know i was obsessed with them right out of my CS program.
>>13624273you're going to be so disappointed with what strong AI actually is in your lifetime lol
>>13624525train=False
>>13624602this is what i wish more people would talk about
researchers follow the compute. technocrats have the compute, they have the researchers, they'll have AGI.
the only alternative is China getting it first, and that's way worse.
