No. of Recommendations: 3
In a recent interview, Dario Amodei, CEO of Anthropic, a leader in artificial intelligence, grabbed headlines. Amodei argued that the next generation of AI systems could replace half of entry-level jobs and drive up the unemployment rate to 20%. All of this could occur in the next five years, he said.
Recent data seem to support these glum predictions. Mark Zuckerberg said AI will be as capable as a mid-level programmer by the end of this year. Microsoft announced thousands of jobs cut this spring, with programmers disproportionately affected. In an announcement last week, Amazon’s CEO wrote, “It’s hard to know exactly where this nets out over time, but in the next few years, we expect artificial intelligence to reduce our total corporate workforce…”
As a business owner, productivity was paramount in many of my decisions. This was juxtaposed against the restrictions of a trade union in one of my businesses whose contract agreements forbade the use of a number of common sense "labor saving devices". On at least one occasion, I replaced an employee with a gadget.
There has been a constant drumbeat for businesses to take the least expensive route in all things by most businesses - causing the drop-in customer service that we have begun to take for granted.
There are many obvious quantum changes in technology which have created the rise and fall of associated businesses, but few have affected entire populations of workers. The challenge is that we are still thinking of AI as "apps", rather than as people. Adopting AI is not a gradual transition, but the inescapable draw to businesses is its vast productivity advantages over using high-cost employees.
To become a "professional", most people spend 16-20 years in school, with the final 4-8 of them at great cost (at least in the US) and effort. if proficient, they tend to be appropriately rewarded by their employers. They tend to be in specialties requiring processing comparatively large amounts of previously learned information supported by "on the job" training. In contrast, an AI has typically absorbed an even larger body of information and can continue to integrate information, pretty much real-time, as it is developed. After proper training, they can generate better output faster and cheaper than the humans they compete with.
While, for now at least, jobs requiring mobility and manual dexterity such as various types of surgeons (but not all), plumbers, electricians and barbers are likely safe (for a while), professions such as doctors, accountants, lawyers and engineers are very vulnerable.
I had a discussion with a nephew who was a freshly-minted doctor who disagreed. I politely described a day, not far into the future, when you could get a full-body scan, a battery of blood tests, readouts of various body functions and a sifting of your genome compared by an AI to tens of millions of others instead of the annual exam by a doctor poking fingers into your various orifices, etc. Similarly, if supported by AI, a programmer can do the work of ten, what happens to the other nine.
The last time we had a labor upheaval of this scale was the advent of the industrial revolution which caused mass migration from farms to cities - not necessarily with beneficial results to the workers. There are those today who claim AIs will benefit by allowing a four-day work week. As a former employee, I can be certain that if I was using an AI to cut employee hours, I'm not going to pay for "phantom hours" that they are no longer working.
Until now, being unemployed had a stigma attached to it. Assuming AIs will significantly reduce the human work hours in a society, we have a choice of directions. Our society and government are supported by payroll tax revenues (simplistically) and our domestic commerce by employees spending their earned money. If AIs cause a massive reduction in the employment of professionals, the only way I can think of for society to "make up the difference" is by some sort of "digital tax". Whether this is based on number of CPU's, number of AI connections, bandwidth or some other metric, we can be sure that the major suppliers of these services (currently including Amazon, Meta, Google, Apple, Microsoft Nvidia, OpenAI and Anthropic) will use substantial resources to fight against this (legally, buying politicians, societal pushback by classes not (yet) affected, etc.). In short, the job tenure of a class which has defined our society for over a century (in some cases for millennia) is about to be dramatically curtailed and we have nothing prepared to deal with it.
We are also unprepared for all sorts of impacts that AI will make on society. I happen to play bridge (card game). Cheating has always been a factor in card games. It is particularly easily done in on-line games (where a pair of partners could be sitting next to each other and looking at each other's hands for all their opponents know. Last week my partner and I (he's in California and I'm in NYC, so little chance of peaking :-) played a team who were quite accomplished (based on metrics shown on their profiles). One of them took an inordinately long time to make each bid or play each card. It dawned on me afterwards that he might have been using an AI and it simply took time to enter the data each time he was to make a decision.
Anyhow, our society is very poorly prepared to deal with the impending impact of AI and those who are spending billions to develop it will fight any incremental costs we, as a society, demand.
Jeff
