It’s so strange to see people accusing tech companies of using AI to concentrate power and wealth when thus far, AI has almost entirely been all consumer surplus. You have crazily high competition in the industry that allows you the consumer to use SOTA models for free, or even run them yourself.

My prediction is that this will keep going all the way to the AGI stage. Someone will release (or leak) an AGI capable model that’s able to design AI chips, as well as the Fabs needed to build them, as well as robots to build and operate the Fabs and robot factories and raw material mines and refineries.

OpenAI and Microsoft have defined AGI as a revenue number so yeah maybe using that definition.

I believe AGI will require the ability to self tune its own Neutral network coefficients which the current tech cannot do because I can’t deduce it’s own errors. Oh sorry “hallucinations”. Developing brains learn from both pain and verbal feedback (no, not food!) etc.

It’s an interesting problem where just telling a LLM model it’s wrong is not enough to adjust Billions of parameters with.