If AI is so smart, why does it get so much about me wrong?

Content about my demise has been greatly exaggerated.A recent Chat GPT post summarizing one of my articles referred to me as the “late” writer Steve Cuozzo, before restoring me to life the next day.If artificial intelligence is so smart, how come it’s so stupid? In fact, “AI” might better stand for “Appalling Ignorance.”We’re told AI will soon control our governments, minds and  bodies.

The pope warned us to beware AI’s dehumanizing, “Tower of Babel” effect.Economists fear it could wipe out jobs and reduce urban downtowns to uninhabited Pompeii-like ruins.But, those who attribute life- and humanity-altering power to AI ought to first query the bots about themselves to see whether the platforms mangle their identities and histories.

The apocalyptic prognosticators rarely consider whether AI is any good at one of its core missions: to collect, distill and synthesize information into a form that transcends what ordinary search engines can do.Sure, ChatGPT can readily make an elephant dance with a monkey or create entirely made-up people like the “actress” on the cover of the Sunday New York Times Magazine.But it’s all thumbs when it comes to knowledge of real people.Buzzy Claude is no better.

I never wrote a cookbook.I never lived in Greenwich Village.

I have no children.But Claude — the marquee product of Anthropic PBC — said I had done all of those things when I asked it about myself.

Claude knows I’m a longtime writer on commercial real estate.Way to go! But when I asked whether I’ve written about Herald Towers on West 34th Street, it found “nothing specific.” It even suggested I search the New York Post website! Isn’t that what AI is for? Earlier this week, California-based Anthropic, which is valued at nearly $1 trillion, filed with the Securities and Exchange Commission for an initial public offering.

I won’t be investing.Of five AI platforms I tested, Elon Musk’s Grok did the best job of summing up my...