Does Professor Acemoglu advocate for the Ministry of Truth?

In his new article, “Why we must stop talking about artificial general intelligence — and instead build ‘pro-worker’ AI,” Professor Daron Acemoglu proudly declares that he does not understand what progress is. In the subtitle, he put in bold: “Rather than racing to replace people, we should build AI tools that amplify human expertise and expand opportunity.” He does not get that progress means exactly the opposite, freeing people from the toil of hard work.

About ten thousand years ago, our ancestors domesticated horses, and horses’ work replaced the hard work previously performed by humans. At that time, although it took a few millennia to see the results, it was a technological shift as important as AI is today. Weaving machines at the beginning of the Industrial Revolution replaced millions of workers. As a result, clothing became affordable for everyone. The same industrial progress resulted in a shocking shift in how and where people worked. Civilization did not collapse. Quite the opposite, it lifted a standard of living for everyone.

It means that Prof. Acemoglu is plainly wrong when he writes that “Technology does not have a predetermined social purpose.”  It is the contrary – be it taming a horse, using weaving machines, or AI – we have an advancing civilization because people perpetually strive to find ways to achieve more by working less. It is – using Professor Acemoglu’s words – the predetermined social purpose of technology.

Furthermore, Professor Acemoglu is concerned that the “current focus is on automation.” He sees a problem in the United States tax code because it has a built-in bias that favors automation even when a worker can perform a task at a similar cost. Recently, at a family gathering, I spoke with construction workers in their mid-fifties. As they age, they would not mind automation. However, they would like to keep their union salaries. Ergo, Professor Acemoglu’s reasoning holds only under an assumption that automation replaces union jobs. He ignores that automation is not the problem, but rather the privileged position of unions and the artificially low earnings of non-union workers, which are lamely compensated by earned income tax credits.

Professor Acemoglu is also concerned that AI could replace teachers. People dealt with a similar problem after Gutenberg invented a printing press. Before that, only a very select few had access to written knowledge. They were the teachers. Martin Luther challenged that, claiming that everyone should be able to read the Bible, as well as other books, and make their own conclusions. Civilization did not collapse.

With that in mind, I agree with Professor Acemoglu that we need a legal framework for data markets. The public should have as broad access as reasonably possible to data that AI can aggregate and to the data AI can produce from that collection. Resolving this conundrum will be one of the biggest challenges of our times.

Professor Acemoglu advises us that to address that challenge, “we need a dedicated AI agency at the federal level in the United States.” The only experts on AI are its creators, so whatever the government does will be in collaboration with the largest AI corporations. It smells like a collusion between big business and government power. Capitalism has a bad reputation, largely because of those backroom deals between business and government. In the case of AI, it will produce the Ministry of Truth as we know it from George Orwell’s “1984.” Or as Christian Andersen would say, the “emperor has no clothes.”