Insert thoughts here

Artificial Intelligence offers great benefits; we'll figure out the cost later

The comedian Sarah Silverman once said she knew an AI company was stealing her material when she asked ChatGPT for an example of a Sarah Silverman-type joke and the service spat out a humorous observation she had shared only once before – in an unpublished book.

To be fair, that’s probably an apocryphal story, or at least I’m trying to make it one. But the fact is, Silverman and two other authors did file class action lawsuits in federal court in July, 2023, alleging that the contents of their copyrighted books were used without their consent to help train Large Language Models so AI will know how to talk good. (No, I didn’t consult the Internet for help on that sentence.)

I was thinking about Silverman’s lawsuit Saturday morning right before leaving town, while scouring the news rather than doing what I should’ve been doing – writing this Substack. Luckily for me, Saturday’s New York Times was full of distracting stuff, including a report that the AI startup Anthropic had agreed to pay a $1.5 billion settlement in a similar but different lawsuit. This one had a class of 500,000 authors whose work it allegedly pirated to train its AI model, Claude. For the math-challenged, that works out to about $3,000 for each class member.

Many writers might consider that a nice little sum, seeing as how it comes on top of whatever they’d already been paid for their work in the first place. The settlement, if given final approval by the judge, would be kind of like an additional royalty payment, except it would really be, in Legal-Speak, “a retroactive licensing fee” or more simply, “damages.”

As for the total of $1.5 billion to be paid by Anthropic, regular people might think of that as “a shit-ton of money,” although a thriving AI company might just call it “the cost of doing business.” Anthropic, according to its own statements, has already earned more than $5 billion this year and has reportedly told investors they’re shooting for annual earnings of $34.5 billion in 2027, which in the tech world might be called “the distant future.”

So now that we’ve got our math and English lessons covered, let’s spend a minute on history and social studies. AI has crept into our lives gradually – through Internet search engines like Google and then with recommendation systems like the ones used by advertisers and social media to show each of us what they want us to see. The new industry powered fearlessly ahead with the introduction of virtual assistants and navigation, and apps for ride-sharing, finance, healthcare, etc.

Then came the chatbots. Three years ago, no one had ever heard of ChatGPT; now about a quarter of U.S. adults use it daily. (And that doesn’t count its competition: Claude, Gemini, et al.)

Who knows what's next, but I may have gotten a glimpse of it while plowing through the NYT before my trip last weekend. I came across a review of a new documentary, Riefenstahl, about the eponymous filmmaker (first name Leni) who rose to prominence in 1930s Germany as a chronicler and promoter of Hitler’s vision. Riefenstahl wanted to be considered an artist, though many of her films, including her best known, Triumph of the Will (1935), were funded by the Nazi party for propaganda purposes.

Today, governments still practice Riefenstahl-type tactics, as does, for instance, the highest rated television network in the U.S. Almost all nations, media outlets and political candidates try to influence us; it’s just that some are more subtle about it than others. Many of us think we’re resistant to such manipulation, but probably that’s just by the side we don’t like

.

Still in its relatively early stages, AI plays a big role in these efforts to shape opinion. In time, we won’t need to ask questions on desktops, laptops or iPhones. We’ll be wearing AI in our eyeglasses, jewelry and clothes; it may even be implanted in our bodies, guiding us through our daily lives, telling us when and what to eat; how to dress appropriately for the occasion and comfortably for the weather; what work needs to be done and how to do it; when to vote and for whom.

Of course, we’ll be able to override these constant, automated suggestions — at least at first — but why bother? That would just create stress.

And where will all this helpful information originate? From whoever provides the AI.

Until then, most of us are still reading and writing on our own. Some are even publishing Substacks… when we’re not too distracted. I hope you'll want to keep reading mine, which, Full Disclosure: AI may help shape in the research stages but never in the writing itself.

If you haven’t yet subscribed or shared with friends, please do! No payment is required, but voluntary subscriptions are appreciated, and for a lot less than $34.5 billion a year.