While I appreciate the meme, I'm not sure I like what it says about this subs direction...

I fervently hope he's wrong, but one look at our current socio economic setup spits out all sorts of alarm bells. The first mover benefits in AI are so extreme even in pure dollar terms that every tech company with acsess to a server farm (which with cloud services today means anyone with a few millions to spend) is going to be hurtling towards AGI like a shot out a cannon, aligment be damned. Its pretty much an 'I win' button for capitalism.

Even if we lived in an almost utopic and unified world goverment, the danger posed by rogue research teams skipping safety in favor of speed and releasing a botched up AI would be enormous and very difficult to manage or police.

As it stands? I've been lately grasping at straws about how this all wont end badly for the human race, possibly in less than 10 years. Given I'm not an AI researcher, I'm pretty much reduced to not thinking about it, and naivily thinking that we'll probably be okay if most of the teams at the vanguard of AI research are not themselves panicking yet.

Thanks, great read. Carmack's an interesting guy; he's humble, but also not shy about his qualities and what he brings to the game. Its good we have people exploring alternative pathways to AI that are not in the same billion dollar tech giant wheelhouse. Seems almost cyberpunk.

Common use AI's like ChatGPT are setting themselves up to be the next consumer-tech revolution, and you can understand a lot of the attitudes behind the creators by evaluating the fallout from the last revolution, Social Media.

Soceity got burned hard by social media, and whether it has been a net good or a net wrong is still, IMHO, an open question. It stands to reason that devs are wary of becoming the next Facebook but worse, polarizing already strained soceities past the breaking point, letting spammed disinfo wreck public discourse, turn kids into functional addicts or who knows what else that we cant foresee.

Having said that, I cant help but be wary of how theyre taking this. Too much hem hawing could mire us deeper in a 'boring distopia' where the big tech AI are completely sanitized 'for your own good', a 'good' that most ofteb coincides with what is good for the company's PR, image, and the Company itself. As always, we'll have to hope in open source projects to save the day if this gets too dire.

This is a topic Im really interested and you seem pretty well informed. Would you mind expanding on examples of censorship degrading AI preformance?