The scaleable, general meta-learner (AGI) is humanity’s final invention. Once you get something that can reason like humans, you just clone it, give it a bunch of resources, and have it invent everything else. It defines a clear step function in society. It is hard, nearly impossible, but if you figure it out, everything changes. This constitutes a worthy life endeavor.

Infinite intelligence frees us up from being productive. We can connect with one another and form rich communities. We can spend more time understanding ourselves. We will be at the top of Maslow’s hierarchy.

When is this happening?

Consensus estimates are 2030. But it is very possible that we are in a local maximum. We may be too bullish on LLMs. Maybe we need a completely different paradigm to build this. Maybe we need a different substrate (organic matter?).

But it is also a function of how much we do as a society. It is up to us. If we are too pessimistic and deem it impossible, we will not do anything, so it will never get done. If we are too optimistic and deem it inevitable, there is nothing to be done, so it will never get done. Extreme pessimism and extreme optimism converge on laziness.

Right now there is healthy optimism. More resources are being devoted to building AGI, which yields some progress (scaling laws), which means even more resources are devoted to it, which yields even more progress! It is a snowball effect. I think this means it’s a really exciting time to work on it.

Why am I not working on this?

I fear being victim to the hype cycle. One of the hype cycles I fell victim to was programming. I wanted to build things as a young kid. Invent things that gave us bigger “levers”. This is, broadly speaking, technology.

But from the 1960s onwards, the word “technology” was equated with “information technology”. Anything in the physical world didn’t count! Society had a very narrow view of where invention was going to happen, and it was at the software layer. This kickstarted a massive “Learn to Code” movement that coincided with my childhood. So I spent a lot of time on these skills. Now my programming skills feel useless because I can just prompt Claude Code.

Why didn’t I spend more time studying the world of atoms? Hardware? What about biology? Physics? Why was the scope of invention so small?

I fear being the person who tried to build AR/VR in the 2010s. Or who tried to do crypto in 2018.

So is AGI just hype?

I do not believe the quest for AGI is part of the hype cycle. The demand for it, if you can produce it, is promised. I do not think that in 50 years, if it is not built, it will cease to be an interesting field of study. It is perhaps the most robust science that exists, because all other sciences are threatened by it.

It is the most exciting open question in the world, for as long as it remains an open question.