What to believe about AI

These days the news is filled with stories about artificial intelligence (AI). Much like reporting of past technology breakthroughs, what’s being reported is confusing, contradictary and often alarming. The truth is most of the writers (myself included) have little sense of what it all means, even those designated as experts. That doesn’t stop the major networks and newspapers from reporting about it with great certainty and authority.

I’m always amazed at how suddenly experts appear within days of a new trending technology. I’m not referring to true experts that have spent their careers working in the field. But I get several pitches a week from PR firms asking if I want to speak to one person or another on the subject, usually someone that’s trying to latch on to the next hot topic. You can bet that some of them will be appearing on cable news.

But even true experts can’t predict how the technology will be used. This past week a number of industry experts warned us that AI was putting mankind on the edge of extinction, saying that AI needs to be regulated (while they’re each trying to move faster than the other to get their products out). That was alarming and certainly may even be true. But none of them explained their fear with more specifics and none addressed the issue of how it’s even possible to regulate something in a world where some countries would never participate. It’s as if an arsonist was screaming about the dangers of fire.

When we first began using the Internet, we saw it primarily as a way to speed communications with email and share information instantly. Companies could promote their business on a website instead of mailing printed brochures. No one envisioned all that we do with the Internet today. It’s the same with AI.

AI today is being used for some interesting time saving chores normally done manually. That immediately leads to stories that predict millions of jobs are going to disappear. That’s a huge leap and is misleading if you look at history.

Just four years ago Andrew Yang, running for the the Democratic nomination for president predicted truck drivers’ jobs will disappear because of self-driving trucks, yet today there are not enough drivers to meet demand. Both exist side by side. The fact is that no one knows exactly how technology will be used, although it certainly will be used for both good and bad. The experts don’t know and those that say they know are not experts. Everyone is guessing.

The job market has been able to keep pace through the centuries when new inventions come along, in spite of dire predictions. What is predictable is we’ll be doing many things differently and more efficiently. While some occupations will be affected (think travel agents), new types of jobs will be created (website developers).

Take ChatGPT, which in its first weeks was described as putting lawyers, writers, doctors, poets, and journalists out of business. It would cause a huge problem for college professors who wouldn’t be able to tell the difference between a student-written paper or one generated from this new miracle product. A few months later we’re seeing its flaws, as how badly it failed a lawyer in his trial. Just because ChatGPT was able to pass a bar exam, doesn’t mean it has a lawyer’s reasoning abilities. And have you tried using it to write a poem? I tried using it write a few poems for my wife’s birthday and all it got was a lot of laughs.

The truth is experts don’t know and those that say they know are not experts.

Automated image recognition was going to put radiologists out of business, but that’s not happened. Instead machine language has aided radiologists in their work and made them more effective. AI lets doctors be more effective by being able to scan medical literature to research a patient’s symptoms and being able to compare the results of different treatments.

Remember, most of those who report the news don’t have a scientific or engineering background and can’t see the nuances. Nor do marketing and PR people that want to promote their products and services. Seven years ago I ridiculed the idea that Uber would be moving to self driving cars, yet that was the prevailing thinking back then becacuse Uber said so in widely promoted announcements.

Beware of predictions, especially the timelines given. Ultimately some of these things may happen, but rarely as fast as predicted, and more often not at all.

Be skeptical, read different opinions, and look for the journalists that display a healthy skepticism and that know their subjects well. In the field of AI I’ve come to rely on Casey Newton, a very smart technology journalist that writes a newsletter, The Platformer. Others are Peter Kafka, a journalist,and Gary Marcus, an AI scientist with a blog and podcast.

The truth is experts don’t know and those that say they know are not experts.

by Phil Baker