denialism and its counterfeits

Freddie de Boer noted that Yascha Mounck strives to explain The Peculiar Persistence of the AI Denialists — and I want to note what has happened to Mounck’s key term, “denialism.” It originated of course in the debate over climate change: it was and is used to describe people who deny that the climate is changing, and instead insist that everything is what it has always been and that any apparent warming is merely ordinary variation in weather. The point of the phrase is that we have masses and masses of data demonstrating a long-term trend of increasing temperatures, data that can’t be argued out of existence — so if you don’t like that data you can’t refute it, all you can do is deny. And if you deny all the time you become a “denialist,” and your intellectual strategy becomes “denialism.”

But this is not the situation we’re in with regard to machine learning. Nobody knows what’s going to happen, though we can make some reasonable guesses. We don’t know how much better the LLMs will get; it’s possible that their rate of improvement will slow, and that some problems will prove insoluble without serious methodological change. And if that latter is the case, we don’t know whether new methodological strategies will be tried, and if they are tried whether they will succeed. We don’t know whether hallucinations will become less common. We don’t know whether our comatose legislative branch will arise from its torpor and do something: it’s not at all likely — but legislation could well happen in Europe, legislation that offers a template for U.S. legislation. I wouldn’t bet on it, but we might experience a low-grade Butlerian jihad. And one thing I would bet on is, in the not-too-distant future, some serious and widespread black-hat hacking that the big AI companies would be at least as vulnerable to as the rest of the tech sector. (In this matter we’ve been too lucky for too long.) And it’s impossible to guess what the run-on effects of such an exploit would be. 

So what Mounck is doing here is dismissing anyone who disagrees with his predictions of the future as “denialists” — as though his predictions have already come true. Which of course they haven’t; that’s what makes them predictions. It’s not “denialism” to doubt that some extraordinarily dramatic thing will eventually happen — even if your doubts turn out to be unfounded. People only use that word with regard to the future when they think their predictive powers are infallible — which Mounck apparently does. 

Thus he concludes: 

But if there is one thing I have learned in my writing career so far, it is that it eventually becomes untenable to bury your head in the sand. For an astonishingly long period of time, you can pretend that democracy in countries like the United States is safe from far-right demagogues or that wokeness is a coherent political philosophy or that financial bubbles are just a figment of pessimists’ imagination; but at some point the edifice comes crashing down. And the sooner we all muster the courage to grapple with the inevitable, the higher our chances of being prepared when the clock strikes midnight.

Ah, the old “bury your head in the sand” trope — the last refuge of the truly thoughtless. And then the claim that, since some events in the past have turned out to be worse than some people expected, therefore whatever Mounck is most worried about is “inevitable.” Because no one has ever expected things to be worse than they turned out to be, right? Nobody in 1963 ever said “Anyone who thinks that we can avoid nuclear war is burying his head in the sand.” Nobody in 1983 ever said “Anyone who thinks the Soviet Union will just go away is burying his head in the sand.” Nobody! We’ve never been wrong in anticipating the most dramatic outcome … have we?  

I don’t know what machine learning will bring, because, contra Mounck, nothing in this crazy old world is inevitable, and if his writing career lasts as long as mine has, he’ll eventually learn that. But as we move into uncharted territory, I will keep three maxims in mind: 

Proceed With Caution “We must cultivate our garden” For every Nostradamus there are a hundred Nostradumbasses