Geoffrey Hinton, the Godfather of artificial intelligence, says there's as much as a one-in-five chance that AI will end us all.
I'm with you man, I'm with you.
Hinton is a bit ahead of me, though. He's worried about the future of AI. I'm worried about what AI is doing to the past.
See, I have to Google myself from time to time. It always gives me a queasy feeling - you never know what you'll find. But it's the easiest way to find my old work.
Or it used to be.
What I've been getting lately is fiction. Let me start at the beginning.
Back in 2009 I wrote a column - not my best work - about surprises in the news. I thought it would be fun to lead with an inside joke to my wife, Alecia. We'd just found a photo of her, at age 7, forced by her parents to compete as Little Miss Confederacy in a Montgomery pageant. Surprise!
She wore a little white dress sewn by her grandmother, a sash that said Miss Alabama Confederacy, and an uncomfortable smile.
I thought it hilarious, as she is the last person you'd find in a pageant or an Old South spectacle. I wrote a couple of lines to that effect.
"Great Jefferson Davis' Ghost!" I wrote. "Can my otherwise egalitarian wife really have been part JonBenet Ramsey and part standard-bearer for the not-so-good-old-days?"
People didn't think it was funny. For months they stopped us at the grocery to assure her I was the jerk.
But that was about it, until Alecia decided this month she wanted to write about being thrust into that pageant. She began to look for the picture, and I headed to Google to help.
"John Archibald" and "Little Miss Confederacy," I typed.
And the most amazing story popped up.
"The reference to Little Miss Confederacy and John Archibald refers to a 2018 controversy in which Archibald, a columnist for AL.com, wrote an article criticizing an Alabama beauty pageant contestant for her social media post celebrating the Confederacy," its AI said.
It said the article (which didn't exist) was about a high school student named Annie Johnson (who I didn't write about and can't find a reference for) who competed in the Miss Alabama pageant (which has a minimum age limit of 18).
None of it happened. But then the story gets good.
The column (which didn't exist) got severe backlash, AI said.
"The ensuing controversy resulted in threats against Archibald and his family (nope)," Google concluded, "leading to him and his newspaper publishing his work anonymously for a period (nope). The incident drew national attention (nope) and sparked broader discussions about race, the legacy of the Confederacy, and the consequences of online speech." (I wish)
It'd be funny if not for the whole doom thing.
People and politicians - I'm not so sure that's redundant - complain about news agencies that correct their mistakes, while tech bros push stories with sources and data down the page in order to make room for AI fiction about real people. If I wrote with reckless disregard I'd get sued. If AI does it it's just progress.
It is epidemic.
I test-Googled my colleague Kyle Whitmire and the term "swimming." Google AI told me Kyle was a columnist at AL.com, which he is. It also told me he's a 10-year-old, which he is not, who helped the Boaz Barracudas win a championship in 2014, which he most certainly did not. That may be even nuttier than my wife winning a Confederate crown.
It seems like a paltry thing, petty next to the AI godfather's prediction of doom. But these fictions change our perceptions, and our ability to hold on to reality.
The little warnings beneath AI fictions are not enough. Not nearly. Google has trained users over the years to trust its result and the words that appear in summary above them. And it, along with the social media titans and politicians, have made facts harder to discern. That concentrates power for them.
AI is a serious, world changing technology. It certainly allows us to attempt things we never dreamt. But it also upends education, discourages reading and so discourages thought. It poses incredible risk.
Just listen to the Nobel laureate Hinton, who quit his job at Google so he could say this stuff.
"My worry is that the invisible hand is not going to keep us safe," he told the BBC. "So just leaving it to the profit motive of large companies is not going to be sufficient to make sure they develop it safely," he said. "The only thing that can force those big companies to do more research on safety is government regulation."
We have no real plan for that. And we probably won't as long as AI is writing our stories.
Just look at us, and the world around us. Here we are, protesting libraries for shelving books about gay and trans people, banning the study of genuine history in schools so it doesn't hurt anybody's feelings, rounding up people who do the jobs we don't want to do, and defunding media outlets that still correct their mistakes. All while turning a blind eye to technology that rewrites our lives, fictionalizes history, takes our jobs, cheapens our art and threatens to infiltrate our military.
It's a bigger threat than Haitians eating cats, I promise you that. It's a bigger threat than some of the drugs we lock people up for.
You don't have to take it from me. It's on your desktop. You can't get away from it even if you try.
-
Before I posted this I Googled myself again: "John Archibald" and "Little Miss Confederacy."
AI told a completely different fiction, about a column I didn't write in 2017 entitled "The pageant to crown Little Miss Confederacy is a celebration of what, exactly?"
"Archibald's column went viral online (it did not) and was one of the key factors that brought national attention to the controversy (of which there was none).
The thing is, I've written a lot about racism, and monuments, and history, and our glorification of an ugly past. I just didn't write these.