When I was in Mexico this winter, I met a 30 year old man from Seattle who had just been let go from Google. His undergraduate degree was from Yale and his graduate degree was from MIT. “Surely I thought my educational background would protect me,” he said as we ate before a long hike up a volcano.

He had spent the last year working on projects related to utilizing AI to handle tasks that human workers had previously been able to complete. “I knew some of the employees that were working on the project would be let go, but I didn’t think it would happen to me.”

I expect we’ll hear that refrain over and over through the rest of the 2020’s.

In January, a few news outlets asked, “If the economy is doing so well, then why are so many tech companies laying off workers?”

Those outlets engaged in what I consider to be both lazy and irresponsible journalism. Tech was doing very well in January; they are doing even better now, eight months later. Their executives and boards found that AI could handle the jobs of a several thousand workers at a fraction of the cost. That’s what happened to the young man from Google who I met on that Mexican volcano.

This will happen more and more over the next few years.

A couple weeks later, I met a 35 year old New Yorker who worked for Amazon 25 hours a week while living in Mexico City. “There is shit I can do that AI can’t yet. But I have no illusions. I easily make Amazon about 20 million dollars a year, but eventually AI will be able to do my job. I figure I have about three years.”

We had met at an AA meeting and he reached out to me a few days later to meet for coffee. “I have to figure out what I’m going to do next.”

Twenty years ago, I was told that any job that can be moved overseas will be moved overseas. Fourteen years ago, I told my Rutgers students that any job that could be done by someone on a computer would be at risk for losing it to someone who lived in a cheaper area in the US or another country. Now the question one has to ask is, “Can my job be done by AI right now? What about in five years?”

AI will probably continue to improve at writing, but highly skilled writers should continue to keep their (sadly) underpaid jobs.

Here are the skills that are (currently) AI proof:

1. Managing people. This is different than training. This is dealing with how workers’ soft skills and personal lives affect their jobs and getting them to complete tasks. You have to have some basic people skills. The better you are at this, the safer you’ll be. If you are shy or don’t like people, you may be in a bit of trouble.
2. Public speaking. As of yet, there aren’t any AI robots that can get up in front of a crowd and deliver a speech that hits on a bunch of different emotions. This is an incredibly rare skill, but it can be developed and improved with practice.
3. Complex problem solving in a fast paced setting. AI still hasn’t figured out how to handle more advanced problems on the fly, so if you are someone who is cool under pressure and has a lot of experience coming up with solutions to sudden and random crises, you are good. For now.
4. Sharing of self to help others. AI will never know what it’s like to lose a friend to suicide or have a parent die in their arms. AI can’t connect with someone in the throes of addiction and show by example that there is a solution. AI will never know heartbreak, or financial insecurity or the fear of physical decline. A good social worker knows when to strategically share experiences like these to help a client.

One more thing to think about: if your work can be done in person, try to do it in person. For social workers who do case management or individual or group therapy, you are (a) usually more effective with your clients if you meet them in person and (b) you are less likely to be replaced by an AI therapy chat bot.