I had kind of a dark realization as I was reflecting on the state of the AI debate. Note, a lot of this some from fast-moving online social network sites like Twitter and Bluesky, which many of you do not use because you are sane and well-adjusted people, so the conversations may not be familiar to you.
But basically I’ve recently noticed interesting things about the pro/anti AI polarized debate.
First, it’s the most intensly polarized debate I’ve ever seen. Love it or hate it, there seems to be essentially no in-between. (fair disclosure, I’m employed in the AI industry, I have an interest that translates to being mildly pro-AI with some big caveats. I often feel like I’m the world’s only AI centrist).
I trust we are all familiar with the rhetorical excesses of the pro-AI crowd, what it is, what it can do, what it means for society. No need to delve into that. Much of it is BS, some of it is true in ways that aren’t yet fully emerged.
As for the anti-AI crowd, a remarkable theme I notice is their insistence that AI simply is not useful or effective. These folks don’t even engage in the conversation about the existential risks of AI because doing so would be an indirect admission that it’s not just a “stochastic parrot”, that it has powerful capabilities with real-world consequences. If you admit it’s a sword, you also have to admit it can be a plowshare.
This is why they lean so hard on the datacenter debate. Datacenters have been with us since the 1980’s. They’ve always been energy hogs with special cooling needs. They’re just big boxes of computers. So why care, why now? Well, see the previous paragraph. Claiming that AI is smart enough to do dangerous things is to admit that it’s smart enough to take your knowledge job or creative job. Your liberal arts degree might be worthless. This admission seems to benefit the thing that frightens you, so your only story can be that it’s cooking the environment for nothing. (never mind the fact that energy is more costly than it’s ever been, if nobody truly felt it was valuable enough to pay for, then nobody could pay for the juice to run it).
So here’s the thing that fucking terrifies me. The syllogism goes something like:
- AI is stupid and useless
- No it’s not, it’s very powerful
- Yes it is. Here, look at this example of dangerous things it’s doing. We need safeguards and a standards-setting body.
- scoff any email virus can do stuff like that. We don’t need any new guardrails for anything but datacenters.
- Oh yeah? I’ll show you, I’ll build something that shows exactly why guardrails are so critical
- Oops
So this cycle of escalation is now built into the conversation of whether AI is good or bad, and the warm water under this hurricane is the capitalist profit motive that says I don’t give a fuck, I’m throwing billions into these toasters, so I had better see some return on my investment.
Side note, I know there’s a lot of sensitivity over people like me saying things like “it’s a done deal, it can’t be stopped.” This isn’t because me or my employer have decided to steamroll our preferences onto society. It’s because there really is demand for this stuff. The datacenters you see aren’t there for demand generation, it’s to provision capacity to satisfy contracts that were closed before some of you ever heard of a datacenter.
The hurricane analogy is real. You can’t nuke the hurricane. You cannot lie down in front of the bulldozer and stop anything. The deals are done, the resources are committed, an entire industry you can’t see is already birthed. AI is not happening. It has already happened. Your only choice is whether you stick your head in the sand or raise your voice and engage with it before the biggest dickbags in the universe consolidate more power over the economy, our wolrld, and your life.
