[mark_l_watson]

Cool article.

While I have been using GPT-3 via OpenAI’s APIs for about a half a year and I very much also appreciate using GitHub’s CoPilot because it saves me time, I wish for much more research into hybrid AI systems that are multi paradigm: deep learning, symbolic AI, new types of RL learning, breakthroughs in scaling conventional search, etc., etc.

There is so much work to get to the point where AI systems can effectively do counter factual reasoning, autonomously develop better models of the world, etc.

Symbolic AI as I learned it in the 1980s and deep learning in the last ten years are all great first steps, but we have a long way to go. Assuming parallel work in AI ethics, I don’t think there are any real limits on how much this technology can improve our lives.