Fuck AI

I am unsure myself. I am fairly certain that if all other current AI metrics were the same, but logic was significantly better, we would have the start of AGI now. Logic however seems to have been the one metric that did not show much improvement when training sets and compute scaled up. The AI industry is clearly avoiding talking about this massive issue in a meaningful way (some are of course), but you better believe this is the focus area right now as the first to crack will have an instant advantage. Will it be cracked and if so when? No one knows. Could be any day now, could be 10 years + or maybe we discover there is a hard cognitive valley that will be extremely hard to cross.

Given how many companies, small teams and even individuals are trying to crack logic, I suspect we will have some progress in the years to come. If not, AI, and even robots will never really rise to the potential we are expecting.

If logic is not cracked, a lot of people are going to loose a lot of money as the ROI is just not there with the current state of AI. Sure, AI is not going anywhere as it has its use cases in its current form, but the errors inherent in the current tech will hold it back in a big way.