From time to time I write something about artificial intelligence (AI), such as this and this. I’m not a tech nerd, even though once in a while someone decades my junior will compliment me on my grasp of such things, adding “for someone your age.” What interests me is not so much how the AI sausage is made but the consequences of eating it. Today, I’ve got a full plate: three courses of AI-related experience.
FIRST COURSE: LAUGHTER
CBS dropped a new series this year, “Cupertino.” The main characters are two attorneys fired by a Silicon Valley start-up (one of whom was a founder of the company). They start their own practice dedicated to helping people who’ve also been cheated or exploited by tech companies, particularly by the kind of huge deep-pockets companies such as those being called “The Magnificent 7,” a term you may know but which I learned from watching the series. (Look it up and learn a new definition of hubris.)
You can stream it for free on Pluto TV or on quite a few subscription and purchase platforms. It offers accessible entry points to just about every social, economic, and political crime of the tech lords. I find it pretty satisfying to watch the protagonists win most of the time. You’ll learn about high tech, especially AI, but I think its greatest value is to explain how two trains running across our body politic—capitalism and the legal system—work together to render ordinary people collateral damage while treating wealth and privilege as God-given virtues. Although all the particulars of time, place, and story couldn’t be more different, it kind of reminds me of the original “Star Trek” in its approach to drawing ethical lessons from unlikely juxtapositions and strange new worlds. Highly recommended.
SECOND COURSE: TEARS
My personal use of AI has been scant. When I search for answers on Google, the AI response that comes up first is often quite useful, and because it comes with links, I can check it. I haven’t involved AI in my writing in any way, but I have made use of it to prep drawings that I transfer to painting panels, uploading images of elements of my composition and getting its help in integrating them. That’s how I learned firsthand how difficult it is to avoid anthropomorphizing AI. I keep reflexively thanking it when things look right and when they go very wrong, I find myself engaging it as I would a collaborator.
But AI has also been coming at me unbidden. Since this summer, I’ve received multiple messages each week from purported humans who love one of my books (including those published long ago) and want to help me promote it. The first time this happened, I was fooled. The author wrote an elegant analysis of one of my books, full of subtle flattery describing how much the world needed to read it and why my prior promotional efforts had fallen short. I was offered a new publishing deal by what appeared to be an enthusiastic editor at a legitimate publishing house who said I needed to work with an agent she recommended, also at a well-known agency. I replied with questions, attracting more well-written and well-calibrated prose. It was only when I received a menu of miraculous benefits, each with its price tag attached, that I researched further. The tells were small things: email addresses that were just one letter off from the addresses of real publishers and agents, photos that when searched turned out to be stock images used in all kinds of scams, the fog of sycophancy that rose as compliments multiplied.
These scams appear to be very common. Some legitimate publishers and book sites now have warning signs on their home pages alerting visitors to the fact that they have been targeted by such scams and would never send anyone a letter soliciting submission of a previously published title. Sites where writers gather for conversations are full of such tales. I have taken to moving the scam letters into a folder to keep track of them: right now, it contains 123 messages.
How does this connect to AI? Because the individuals operating these scams would never be capable of the graceful and articulate writing that sparks some sort of hope or belief in the recipient. I never received anything like this before the advent of AI, and now I’m inundated.
Go ahead and think me soft if you want, but all this affects me emotionally because of its cruelty. It’s my habit to check my junk folder most days to be sure an important message hasn’t gotten waylaid. I always find tons of emails saying I’m about to receive a large sum of money; a long-lost relative has remembered me with a bequest; my email account is about to be shut down unless I click a link to pay for something I never bought; or the sender has captured my computer and unless ransom is forthcoming will share obscene images with the world. They prey on fear and greed. Some of these scams are clearly intended for elders, making threats about Social Security or endangered retirement accounts. I can’t help but marvel at the heartlessness of preying on the weak. I know scams are nothing new; I’m sure there were shell games before there was writing, let alone internet. But the way AI has spread, polished, and multiplied them is horrifying. Sadly, we are by now so used to this that it is assumed to be a bug—bushels of bad apples spilling into a system. But clearly, it is a feature, attracting users by enabling people with abundant greed and no care about the harm they do to effortlessly elevate their own capabilities at others’ expense.
THIRD COURSE: FEARS
“Machine Gods” is an excellent podcast about high tech created by two journalists who used to helm the New York Times podcast “Hard Fork,” which I’ve often quoted. I listened yesterday to a recent episode that started with the Jensen Huang/Nvidia scandal, and went on to share a recent report by METR about how the AI company Anthropic, often seen among the frontier AI companies as most concerned with ethics and protections, has achieved the capability of accelerating AI development with a rapidity and danger far beyond prior predictions—the report says “at twice their historical rate.” Anthropic redacted a good deal of the report before it was allowed to be shared publicly, so the predictions could be even more dire than this podcast episode suggests. The podcast cohosts, Kevin Roose and Casey Newton, have been judicious and cautiously optimistic about AI, believing its power to do good is likely to surpass its dangers. By the end of this episode, they’re not so sure. I encourage you to listen.
“People Hatin’” by The Red Clay Strays.