🎇 What if super AI happens during Trump's presidency?

Both the White House and the rest of Washington should be thinking hard about the possibility

Even though President Trump mentioned the Stargate Project — the (as much as) $500 billion AI data-center initiative from OpenAI, SoftBank, and Oracle — during his congressional address last night, he didn’t talk about what the infrastructure project is for or how advanced AI might affect American society in coming years. Well, at least he mentioned Mars.

The AI absence reminded me that the recent emergence of highly-capable artificial intelligence wasn’t a big issue in the 2024 election — or really an issue at all. (Zero mentions in the two presidential debates, for instance.)

Still, it’s my best guess that a) the current American president will be talking about AI quite a bit at some point during his second term, and b) AI will be a big issue in the 2028 presidential election. How big? Perhaps, at minimum, at least this big (via Manifold Markets):

Of course, the salience of AI in our political conversation will depend on how fast the technology progresses, and how deeply it penetrates our lives as workers, consumers, investors, and people who want to be protected from military threats. On that first point, it’s worth noting this conversation chunk between New York Times columnist Ezra Klein and Ben Buchanan, former special adviser for AI in the Biden White House, during a recent episode of Klein’s podcast:

Ezra Klein: After the end of the Biden administration, I got calls from a lot of people who wanted to tell me about all the great work they did. But when you called me, you wanted to warn people about what you thought was coming.

What’s coming?

Ben Buchanan: I think we are going to see extraordinarily capable A.I. systems. I don’t love the term artificial general intelligence, but I think that will fit in the next couple of years, quite likely during Donald Trump’s presidency. There’s a view that A.G.I. has always been something of corporate hype or speculation. And one of the things I saw in the White House, when I was decidedly not in a corporate position, were trend lines that looked very clear. What we tried to do under President Biden’s leadership was get the U.S. government and our society ready for these systems.

Before we get into what it would mean to get ready, what do you mean when you say “extraordinarily capable A.I. systems”?

A canonical definition of A.G.I. is a system capable of doing almost any cognitive task a human can do. I don’t know that we’ll quite see that in the next four years or so, but I do think we’ll see something like that, where the breadth of the system is remarkable but also its depth, its capacity to, in some cases, exceed human capabilities, regardless of the cognitive discipline —

Systems that can replace human beings in cognitively demanding jobs.

Yes, or key parts of cognitive jobs. Yes.

I will say I am also pretty convinced we’re on the cusp of this. So I’m not coming at this as a skeptic. But I still find it hard to mentally live in the world of it.

So do I.

Depending on your definition of AGI, Buchanan’s forecast is pretty mainstream among AI technologists and those who run AI companies. This is especially the case if your AGI definition includes what one might call “weak” AGI (a narrowly capable but advanced system beyond current models)) rather than just “strong AGI (at minimum, fully versatile, human-like intelligence).

Another way to think about AGI: Focus less on defining its capabilities and more on impact. AI that causes this impact is quite likely to meet any reasonable AGI definition:

Preparing for an AI leap

Let’s say that Biden adviser Buchanan, as well as all those Silicon Valley folks, are correct, and systems that can be considered AGI are likely to emerge during Trump’s second term. What sorts of things should the Trump White House and the rest of official Washington be thinking and doing right now?