Problems with the technologies lumped together as artificial intelligence are worsened by the lumping – defying the thinking that the best will be so good that the worst will be overlooked. That’s not a terrible idea in theory; when asked to think about the Kennedys, it’s still likely that more people’s minds will go to RFK than RFK Jr.
It’s not working very well for “artificial intelligence,” which has no generational goodwill and for which there are few wholly positive narratives. Never mind the underlying threat of environmental devastation as the real cost of the industry, and that we are paying higher electricity bills to subsidize a tech that is taking away our jobs and ability to pay them; for every story of how it aids in the rapid creation of potentially lifesaving drugs, there are many more about how it just sucks. (“AI agents get office tasks wrong around 70 percent of the time, and a lot of them aren’t AI at all,” was a June 29 headline on the news site The Register, with the subhead “More fiction than science.”)
A desperation to make multiple trillions of dollars in speculation seem worth it is pushing the idea that are AI narratives at all. The technology should really be in a testing phase of development that most of us don’t know about, followed by a rollout phase of development that most of us don’t know about, followed by widespread implementation that most of us don’t think about. Instead, a bunch of ideally boring technologies that are supposed to help us have been shoved in front of us as a single exciting thing that is exciting mainly because it will be terrible for us. If this were an “Alien” movie, we’re at the part where we shouldn’t have allowed the monster in … but we did. And now we’re going to die one by one for the pleasure of a few Weyland-Yutani Corp. executives.
Fans of AI want us to distinguish between the vapid, maddening and dangerously hallucinatory large language models – the chatbots and agents that give us imaginary legal precedents, encourage the mentally ill and make child porn – and more functional technologies that work in the background and do things humans cannot. “Artificial intelligence” is in the software that lightens and color-balances photos. It transcribes (imperfectly, to say the least, but quickly) large audio files. That the average person doesn’t distinguish between the two is the fault of the tech industry, not its often unwilling customers and test subjects. Its branding of AI is like taking what should be a needed but obscure chemical and making us excited about having it in our food. It’s like if you were supposed to now buy your yogurt just because it includes ethyl maltol, or be more excited about sausage knowing it’s packed with sodium erythorbate. Less esoterically, the industry is like a deli where the owners bring over a sandwich and tell you to be really excited about it because “the bread is coated with something we call mayonnaise. I mean, it always was, but now we’re telling you about it.” And charging you more for the hype.
A better approach would have avoided bad blending and dumb branding by just having tech companies do what they do better. The sadness of Siri made it less of a surprise when Apple failed to pull off an AI upgrade to its system software, and now it plans to turn to Google for an artificial intelligence boost. Yet it took Google until the past few months – long after it introduced its Gemini artificial intelligence effort – to allow its basic search to understand the role of the possessive apostrophe. A search for “Martin Scorsese’s ‘Taxi Driver,’” for instance, would lead the search to clarify if the search was for the same term without the apostrophe. This was Google messing up a PB&J while vowing to make a six-course meal with a soufflé, beef Wellington and baked Alaska.
Artificial intelligence is coming for us, clearly. Not in the smartest way.
About The Author
