A Mug's Game
AI unlikely to define our future
Gordon Wood, 92, the Pulitzer-prize winning historian on the creation of the United States, probably didn’t anticipate the cause of his death before he was struck by a car in a supermarket parking lot last weekend. But we readers may see the incident as a hint confirming his take on the meaning of history,
“I don’t think history teaches a lot of little lessons, frankly,” he said. “I think it teaches one big lesson, which is that nothing really ever works out the way the perpetrators intend. I can’t think of any major event in the history of the world that ever turned out the way the participants who launched it expected.”
That’s helpful to those of us trying to make sense of our current conflict with Iran, to say nothing of the wars that preceded it, and perhaps to Senate Candidate Graham Platner’s wife, who attempted to smooth his political career by confiding some potentially damaging information to a campaign aide, who betrayed this effort to anticipate a problem.
I can’t find any record of Wood’s take on artificial intelligence, but suspect he would have found his general rule, that big plans often fail to have their anticipated impact, as anyone old enough to recall the great Y2K challenge has seen repeatedly. And even any fair-minded critic of President Trump must concede that there’s many a slip twixt the cup and the lip. It would be imprudent of the January 6 rioters to spend much time anticipating how to invest their apology checks from the Justice Department.
Analysts point out that every war game or exercise involving conflict with Iran predicted they’d act quickly to control the Strait of Hormuz. Yet here we are with those in control saying, “Who knew?” My guess is that predicting the outcome should the People’s Republic of China invade Taiwan would be similarly problematic.
All of which colors my extremely unsophisticated view of artificial intelligence, which appears to have a heavy reliance on precedent. I think the most interesting parts of our future are dependent on events that defy accurate prediction – climate and demographic changes, unstable and unpredicted political change nations irrespective of their form of government.
A lot of futurists can make a reputation or get a lot of clicks by predicting what would happen if Donald Trump, Vladimir Putin – or both – died within the next two years. Some around them are likely thinking about probable next steps. More than a few gamblers are probably already betting on it. But the reality is that no one knows and, AI notwithstanding, that’s ultimately unpredictable.
It is probably safe to assume that the chances Russia will merge with Canada and Brazil are no greater than the United States taking the title to Greenland. Maybe both the US and Russia will become more religiously observant, but I’d not bet on that either. But things get fuzzy quickly once such extreme options are eliminated.
I personally rely on movies for perspective. Ultimately my 2001 bias is that humans must retain the right to make the important decisions and reject Dr. Strangelove’s Doomsday Machine that has a computer decide when nuclear annihilation is appropriate. AI can analyze the options and suggest preferences, but as they say on cable news, we ultimately get to decide —or certainly should.
As Wood pointed out, many of our plans will go awry, but at least we’ll know who’s responsible and retains the power of mid-course corrections. The past may be prologue, but it isn’t necessarily predictive.

A much better lede than the hackneyed God laughs.
brilliant bounce off a historian;s sad demise.