Silicon Valley, AI, and the Government

I woke up this morning with a blog post all planned out, but before I sat down to begin writing, I changed my mind. I was inspired by a tweet I read on X that got me thinking. I will quote that tweet in full because I think it nails something that hasn't gotten enough attention: "The government are going to spend billions contracting an AI that doesn't work, fire a bunch of people, and then hire a bunch more people whose job is to make it look like the AI works." This was written by someone named Tom Whyman of whom I have previously never heard. I believe that he is writing about the United Kingdom (he is Northumbrian), but the tweet very much fits our likely future here in the United States.

As is often the case, I feel the need to start with some caveats. I am just a guy who owns and operates a parenting website. I am not a philosopher, a technologist, or anyone with any special insight. I have thoughts and I have a platform on which to publish them. Readers are free to take them or leave them. With that said, it is impossible not to notice that we are currently at a critical point for artificial intelligence. Billions of dollars have been plowed into the technology and expectations raised, with no exaggeration, as high as Mars. But, AI has not been delivering. Sure, it can undertake a few remedial tasks. Maybe write a poem in the style of a famous poet or provide an error-filled response to a question that someone has asked on DCUM. But the technology has not yet reached world-changing functionality. Based on what I read in the press, the main use of AI these days is by college undergrads cheating on their assignments and college professors catching the students cheating, with the technology making errors in both cases. Welcome to our mistake-filled future.

Before I go any further, I want to extol the virtues of humans in government. Today, a large number of federal employees are going to find out that they are victims of a large-scale reduction in force. My heart goes out to these individuals who are losing work through no fault of their own. I, for one, value the contributions you have made and know that we will be worse off without you. It is not beyond the realm of possibility that some of those being dismissed today were selected by AI. We know that the United States DOGE Service (generally referred to as DOGE) has widely used AI for a variety of tasks, and the group has a long list of errors to prove it.

Shadow President Elon Musk is a big fan of AI and has his own AI project, GrokAI. Moreover, Musk associates with a number of other big investors in AI. The tech industry's fixation on AI has driven the share price of Invidia's stock sky high, led to the construction of new data centers throughout the nation, and even caused renewed interest in nuclear power due to the electricity requirements of AI servers. In the eyes of many, AI is now too big to fail. Yet, the technology is not delivering on its promises. When the going gets tough, the hard-nosed libertarians of Silicon Valley inevitably turn to the government.

Silicon Valley technologists, many of whom like Musk only got to where they are due to government subsidies and investment, one day decided that government is bad. "Wasteful and inefficient", they said. But they also had the solution. Surprise, surprise, the solution was exactly the white elephant into which the technologists had sunk much of their wealth. Cast as a solution to government inefficiency, the push to replace federal employees with AI is actually nothing more than a gigantic bailout scheme. The country's most vocal opponents of government spending are big fans of government spending on their own products. But, will it work? Probably not.

As noted above, DOGE has widely employed AI tools. Musk has suggested that the "What Did You Do Last Week" email responses that federal employees are required to provide will be evaluated by AI. AI tools are said to be searching for waste, fraud, and abuse within the government. But the deficiencies of these technologies are clear. The DOGE claims of savings are error-filled and exaggerated. Despite claims of sophisticated AI tools being employed, many DOGE-related tasks are being conducted by primitive word searches. For instance, the Department of Defense recently removed photographs of the Enola Gay bomber that dropped the atomic bomb on Hiroshima as a result of searches for the apparently banned word, "gay". This is not cutting-edge technology, but rather a capability that was available in 1980s versions of Microsoft Word.

Axios has reported that U.S. Secretary of State Marco Rubio is launching an AI effort "to cancel the visas of foreign nationals who appear to support Hamas or other designated terror groups". This will include AI reviews of students' social media accounts. Shortly after this report, the State Department ordered the detainment of Mahmoud Khalil, a student protester at Columbia University. It is unknown whether Rubio's AI tool had anything to do with Khalil's arrest, but there are a number of errors in the State Department's action that make it seem likely. For instance, the officers that arrested Khalil believed that he was in the U.S. on a student visa. In fact, Khalil is a legal permanent resident holding a green card. Moreover, while U.S. officials have repeatedly accused Khalil of supporting Hamas and making antisemitic remarks, no evidence supporting such claims have been produced. In fact, Khalil appears to have been about as moderate of a protester as possible.

As the New York Times reports, Helyeh Doutaghi — a deputy director of a project at Yale Law School — was abruptly barred from campus and placed on administrative leave after an AI tool identified her as a member of a terrorist group. In fact, Dr. Doutaghi is not a member of the group to which she was accused of being associated. She is now being forced to prove her innocence after being found guilty by a computer program. Yale Law School has apparently devolved to the point that it now values AI programs more than human intelligence. If this is occurring at one of the nation's top law schools, imagine what will happen in the government.

Even Apple Computer has egg on its face after being forced to remove claims on its website about features of Apple Intelligence, Apple's version of AI. The company had previously promised "personalized Siri" features would be available in an upcoming operating system release. Now that technology has been pushed back to "the coming year". This comes after Apple Intelligence news summaries were so comically error-filled that Apple had to disable the feature. Apple, famous for its computers that "just work", cannot make AI work.

I don't claim to be Nostradamus, but I think our future is pretty clear. Musk and DOGE have convinced cult leader, convicted felon, and failed President Donald Trump that hundreds of thousands of federal employees can be removed and replaced with AI. Just by coincidence, that AI technology is owned by Musk and his buddies. But what will happen when the AI tools fail? One indication might be what Musk has been doing in his own companies. Last October, Musk sponsored a special Tesla event at which he announced the Cybercab. During this event, Tesla's Optimus robots served drinks to guests. Many in attendance were impressed by the robots' capabilities. However, it later emerged that, behind the scenes, humans were controlling the robots and even doing the talking for them. So, Musk's solution to lagging AI capabilities: use humans to hide the technology’s deficiencies. We are right back to the tweet that inspired this post. When Musk's government AI tools fail, humans will be employed to make it look like the tools are working. Those humans, of course, will not be federal workers, but rather employees of the private companies supplying the AI technology.

What is happening today is a huge giveaway to Silicon Valley. The government is expected to bail the technologists out of the immense hole in which they have dug themselves with their AI investments. If the AI technology is successful — doubtful as it is — it will be a gigantic financial boon to the industry. If it fails, Silicon Valley will simply become the workforce provider of the U.S. government. Heads they win, tails we lose.