I think the content of this post is self-evident. I write it because the core concept underlying it is what bothers me most about the point in time that we are in. It is helpful to me to write things out sometimes, and I suspect it’ll also be helpful to have something to point to in the future.
I don’t believe any of the big research labs can be aligned with the best interests of humanity. Sometimes they appear to be aligned, but this is through the force of will of their leadership. It is not the natural propensity of these organizations.
Here is my argument:
- Each of the big labs (Deepmind and MSL included!) has taken on a tremendous debt load to fund their supercomputer build out.
- This debt is a direct result of the labs desire to race towards expanding the capabilities of their AIs well ahead of the revenue those AIs can currently generate.
- To default on these debts means bankruptcy.
- Even PBCs with a mission statement to build “for the betterment of humanity” must prioritize solvency over their mission.
- Thus, when push comes to shove, the labs will at times be forced to make choices which are counter to their mission to stay solvent.
I think the principal mistake we made was to compete on AI development. There are certain types of technologies where competition in a capitalist environment simply does not make sense - and AI is definitely one of them.
When we decided to compete, we started to take the knock-on effects of RSI seriously. A small lead of a few months to a year for any one lab could compound to a monopoly very quickly. So we needed to race. This incentive led to us borrowing to build faster - either directly or through taking investments.
And this all led us to wind up in the situation we are in today: the existing AI companies must continue to grow revenues. The ship still has rudders: they can choose how to grow (for example - by replacing software engineer labor), but they must grow. And if/when we get into a situation where growing means doing something that otherwise would have caused some hesitation - I have very little faith that the labs will hesitate.
The best case scenario here is that we find a place to channel our development efforts which is truly beneficial to humanity. As an example: medicine seems like a great candidate: if we truly automated drug discovery the benefits to humanity would be undeniable. If that caused some job displacement amongst health researchers, I think it would be worth the price.