AI Predictions from a Non-Expert

 

How do you feel today?

I have seen a prediction that significant job loss due to AI will be evident by 2028.

This prediction is of course incompatible with the drive by various governments for mass levels of immigration. If the AI prediction is correct, we are only importing mass levels of unemployment. Wouldn’t it be wise to at least wait and see? Or, to be safe, to bring in only temporary guest workers?

Experts are usually wrong, of course. They are likely to be wrong about the need for mass immigration, and they are likely to be wrong about the AI impact on jobs. Making it a wash. Whenever government intervenes, the results are likely to make things worse.

This is not just a casual statement, nor an expression of cynicism. It is demonstrable. Experts are less likely to be correct in their predictions than either a coin toss or a man in the street interview. Studies show this, repeatedly. And there are good reasons for this. It is always in the interest of the experts to predict some dramatic change. This gets them attention and publication; and convinces the public that their expertise is needed. And there is no down side. Just as with fortune-tellers, nobody remembers predictions that turn out to be wrong. But should you get lucky, you become immortal.

However, while experts will always predict dramatic change, dramatic change is actually rare. In most cases, the best prediction is the boring one that things will continue on more or less as they are. And that gets you no attention. And gives governments no excuse to extend their power and influence.

Remember that next time you are warned about climate change.

If AI is going to start replacing jobs, I, being no expert, predict that the one profession that should be most concerned is medical doctors. Their expertise is diagnosis of illnesses, and writing matching prescriptions. The process is mechanical. It seems obvious that AI can diagnose and prescribe more reliably than a human. And because MDs earn so much, there is huge business case for replacing them. Suddenly health care costs are manageable again—a huge public and private expense. I suspect they would have been replaced already were it not for their political power as a lobby; but they must be living on borrowed time.

Clinical psychologists ought to go next. What they do is exactly what AI chatbots are already perfectly capable of: to agree with and echo back whatever the patient feels and thinks. An ersatz non-judgmental friend.

Is there any need for supervision of these chatbots? Not beyond that of the user. And, indeed, for the sake of privacy, a machine is preferable to a human. Moreover, you can make the chatbot exceptionally physically attractive, improving the experience and making the friendship seem more valuable; you can even let the patient select the physical features he feels most comfortable with. A human therapist is rather hit and miss. Nor will a chatbot take advantage of the patient, the way therapists are often discovered to do.

Nor can we tell all the unemployed therapists and doctors to learn to code. Actual coding is probably best done, again, by AI.