'AI Revolution' Will Be Disappointing: Picture More Realistic But Still Unhelpful Customer Service

A few weeks ago, the latest version of ChatGPT, GPT-4, came out. Because we have social media now, and 24-hour news, we always need something new to get all worked up about. So, everyone has been freaking out about it.

One of my colleagues gave the thing a go, asking it to write the first draft of a letter to a client explaining a fairly basic legal concept. The letter turned out, you know, OK. ChatGPT got the law wrong, but only minimally wrong, and it sounded pretty close to a marginally competent human. It was about what you'd expect to get from a summer law clerk who just finished 1L year with a 2.0 GPA.

GPT-4's underwhelming performance aside, Elon Musk, along with a lot of other influential people, are issuing some very dire warnings about artificial intelligence. They think AI will become frighteningly advanced in the very near future.

Musk has obviously had a lot of success. Keep in mind though that this is the same guy who has been promising us that fully self-driving technology is only a year away since 2014.

I suppose it's nice that the people working on AI are excited about it. Still, let's tone it down a little, ladies and gentlemen.

For one thing, a lot of the AI overhypers base their dire predictions on the flawed premise that exponential growth can (and will) go on forever. It can't, and won't. Sure, bacteria can reproduce exponentially, right up until they hit the edge of the Petri dish (or run out of food). Just because AI technology very well might be two or four times more advanced than it was just a few years ago does not mean it will be twice as good as it is now by next year.

Also, there is a really big, obvious problem with AI taking over anything that a lot of tech bros seem to overlook: AI cannot make decisions that we don't allow it to make. Perhaps one can type out a conversation with ChatGPT and be fooled into thinking it's a human being. Yet, being able to talk in circles with people who've been made desperately lonely by modern society isn't all that impressive these days. ChatGPT can decide what to say next; it can't exercise any discretion on what actually happens in the real world. As soon as ChatGPT can get my goddamn Priority Mail insurance claim for $13.84 processed at the post office, then we'll talk about whether it's passed the Turing test.

Don't get me wrong, AI is definitely going to become more and more intertwined with our daily lives. A recent Goldman Sachs report found that AI is set to impact some 300 million jobs around the world. That is probably not far off. Even so, with the world population approaching eight billion, 300 million is not that many jobs in the grand scheme of things, and this same report found that the use of AI technology could actually boost labor productivity and grow global GDP by as much as seven percent.

There are a lot of potential applications for more and more advanced AI on the horizon -- as a tool, that is. Doctors will totally be able to diagnose patients more accurately with the help of AI. Computer programmers will totally get to outsource many of the mundane coding exercises to the machines. Lawyers can continue to get crappy first drafts of letters written.

It's not going to wholesale replace people though. Don't believe me? Well, the last time you called into customer service with some one-off issue, did you get what you wanted from the automated system they run you through first to try to get rid of you, or to get anything accomplished did you have to mash "0" repeatedly until a human being picked up?

We should prepare for the possible threat a super-advanced AI could pose, even if that threat is very far off and very unlikely to materialize. We should absolutely regulate AI (there are a lot of things we should be regulating but aren't).

What we shouldn't do is sit around wringing our hands about little boxes of chips and circuits taking over the world when we don't even yet understand what allows human beings to have consciousness and free will. A computer program that can spit semi-cogent prose back at you when prompted is not the huge cause for concern it has been made out to be.

Jonathan Wolf is a civil litigator and author of Your Debt-Free JD (affiliate link). He has taught legal writing, written for a wide variety of publications, and made it both his business and his pleasure to be financially and scientifically literate. Any views he expresses are probably pure gold, but are nonetheless solely his own and should not be attributed to any organization with which he is affiliated. He wouldn't want to share the credit anyway. He can be reached at jon_wolf@hotmail.com.