The Singularity is Coming

And it's good for [my bags]

Sam Altman was on the All-In Podcast recently. Pundits are mostly criticizing him for talking for an hour without saying anything, but as usual the pundits are missing the point. I will quote some of the things the pundits didn’t notice, and explore some ideas about the future that they spark.

Data Quality

I don't think data quality is going to be the thing that differentiates these models. I think it’s going to be much more about new architectures, new training techniques, and fundamentally new ideas.

I wrote recently about how the big tech giants are going to be the big winners of AI. I still believe that, but let’s think about how I could be wrong.

What if LLMs are close to capping out, and to get farther along there is the need for a fundamentally new idea? To be clear, that isn’t what Sam is saying here because he goes on to say that he believes LLMs can be pushed much farther, but let’s just pretend that it’s true.

In that case, a lot of what I wrote before would be invalid, and assuming the “new idea” is harder to replicate than an LLM, there could be a very dramatic winner-take-all scenario for whoever gets there first.

The problem is that unless you are an ultra-high-net-worth individual who is very connected in private capital markets, you will not be able to allocate to that company.

So I think it’s interesting to be aware of Sam’s view here, but I don’t think it’s actionable for us at the moment.

Universal Basic Compute

I wonder if the future looks something more like universal basic compute than universal basic income and everybody gets a slice of GPT-7’s compute and they can use it, they can resell it, they can donate it to somebody to use for cancer research

Sam is talking about a very cypherpunk future in which the single most valuable thing in the world is AI compute time. The government might handle this by distributing a subsistence level of AI compute to every citizen. You can imagine in this scenario that the US becomes like Norway, but instead of having a sovereign wealth fund handling its oil profits, we have a government body managing our new most valuable resource—AI compute—and redistributing some of it to the people.

This is a fascinating idea, and the truth is, I don’t think it’s absurd at all. In this future, the US is the first country to develop AGI (artificial general intelligence) and thus becomes a global superpower on steroids. Other countries may develop AGI after us, but because of the nature of exponential improvement, they will always be behind.

A truly exponentially improving AGI would be the 'omega invention'—the final invention. Once developed, this AGI would rapidly surpass human capabilities in every field, effectively revolutionizing and dominating every industry.

Coin Meet Computer

You could easily see a world where global AI compute gets tokenized and becomes the currency of the world. Imagine a future where compute resources are as integral to the economy as oil and gold are today. In this scenario, AI compute units, represented by tokens, could be exchanged, traded, and utilized as a universal currency.

Blockchain technology provides the infrastructure to ensure these transactions are secure, transparent, and immutable. By leveraging smart contracts, these tokens could represent not just a portion of compute power but also the rights to use specific AI models or access particular datasets.

It’s not difficult to see how crypto technology could interact with this idea of “universal compute” in an AGI-enabled future.

Conclusion

This is all speculative, but that’s what I took away from Sam’s podcast appearance.

Ray Kurzweil was on the Joe Rogan Experience a couple of months ago. The podcast sucked (Ray is getting old unfortunately), and really Ray’s whole appearance can be boiled down to one point he made over and over: exponential growth is more powerful than you can imagine.

Well the podcast sucked but he is right. Take a look at this chart below:

We are somewhere near the red star. We are just starting to say “wow, this stuff is getting real.” If you haven’t spent the last 20 years of your life geeking out on sci-fi (like I largely have) then you are in for some big surprises.

Best to start thinking about this stuff. Now.

If you missed my current thoughts on how LLMs will affect the markets, check them out here.

Good luck out there.