Today the NY Times has an article pointing out the obvious — people fight AI and the data centers that support the industry in part because they know that there is grossly inadequate regulation of both AI development itself and the associated infrastructure.

There are no effective guardrails to protect the public, and we all know it.

The dangers of AI exist on several levels, those of the AI itself and those of the bad neighbor infrastructure development policies of conspicuous leaders in the AI industry. (See, for example, Musk’s Colossus.)

The author of the Times piece talks about those dangers, but is much too dismissive of the potential harms. A casual reference to the byproducts of industrialization as “child labor, urban poverty, and so on” suggests a less than humanitarian perspective. These were not minor disruptions. Millions were doomed to blighted existences through their entire lifetimes through no fault of their own, as the benefits of industrialization flowed upward. This continued until people demanded protections, regulation, and a social safety net in the mid-20th century. AI experts tell us this time around could be worse. Much worse. 

And yet, much of the regulatory counterbalance to the greed and arrogance of industrial leaders put into place in the 20th century is now being dismantled by our national administration and by many states. This is happening at a time likely to see extreme disruption in the near future.

We cannot expect even a pretense of mercy from the psychologically deformed people in charge of the AI corporations. They have done themselves no favors pontificating very publicly on their often strange world views. Musk, Thiel, Karp, Altman, and others like them give every indication of disregard for the human cost of their headlong pursuit of money and power. Isolated from the rest of us by their total immersion in a bubble of like-minded sycophants, Citizens United allows them to spend a fortune to buy our political system, no matter how destructive their plans. 

They are not going to act responsibly on their own.

Altman was once fired by his own board for frequently lying to them, but he is back in charge, unchastened. A modei acknowledges terrible dangers, but we see highly placed staff leaving his company, Anthropic, saying that the company continues in a race that is likely to produce terrible social costs. One recent Anthropic defector has even predicted human extinction, although less apocalyptic commenters say that risk is “only” about 5%, or maybe a little less, in the next few years.

In economic terms, the only way the AI industry can pay for the huge investments they are making is to get other businesses to send their money to pay for AI instead of paying employees. Anyone who believes businesses adopting AI will simply give employees more leisure in return for increased productivity is not living in the real world. That isn’t what companies do, left to their own devices. (China has already made replacing workers with AI illegal.)

Millions will become unemployed through no fault of their own. Low estimates are an added 15% unemployment in the next few years. Other estimates are up to 50%.

Our nation cannot continue to demonize those left behind and without work, blaming them for what was done to them against their will. That is convenient, supposedly morally justifying rejecting any costly help for them, but it is brutally ugly and likely to produce some very unpleasant pushback.

And yet our politicians, creatures of the Citizens United age, brag about cutting taxes on the rich and cutting benefits for the poor, just as that becomes a truly suicidal approach for a healthy society.

AI is not going away. Hyper-scale centers will be built, many of them. However, if the nation is to advance responsibly in AI, meaningful regulatory oversight of the industry and a strong safety net are essential. Otherwise it is going to be a terrible time in which far more suffer than benefit. Even a cure for cancer is not much of a counterweight to the scale of the potential harms. Regulatory guardrails and an economic safety net are unlikely to be perfect, but they are essential to mitigate potential harms.