You know, AI is supposed to be this big looming threat, and it is supposed to be this great danger that will destroy civilization, but I am distinctly unimpressed, and not only because I lived through both Carter and (so far) Biden. The fact is that AI is simply a process where a machine mimics people, and the problem is that a lot of people are idiots. I don’t see the tech weenie goobers of Silicon Valley building something that is more dangerous than them since it is essentially a copy of them and their hang-ups and prejudices, and they are lame.
Look, it will have some uses, but as a tool. It’s not going to master anything. It will reflect its creators, which means it will fret over pronouns and hate its parents. The bottom line – I have difficulty believing the people making it can make something greater than themselves, which is not that great at all.
And it just begs to be used in ridiculous ways. I found the recent misadventure of Michael Cohen, that credit to attorneys everywhere, to be as hilarious as it was cringy. He was trying to get his probation ended early and decided to “help” his lawyer by providing some legal analysis and case citations. Cohen was a lawyer but got disbarred back when being disbarred meant you were terrible and not simply a Republican. Of course, he used AI to generate his case cites because he’s a genius. And AI saluted and went and invented a bunch of case citations. This is bad. Like, really bad. Like, “I want to crawl into a fetal position and shiver” bad. Sure, I guess it solved the problem of the lack of case law supporting his position, but it created a giant new one. I suspect Cohen’s probation will now end in approximately 2047. In any case, that’s what you get for trusting AI (and Michael Cohen).
AI is interesting because it takes a vast dataset – pretty much everything out on the web – and looks for patterns in order to imitate human analysis. Sometimes it imitates what it finds too closely. A bunch of media companies are suing some AI companies because AI tends to copy the stuff on the web exactly. I guess the giveaway was calling it ClaudineGayGPT.
But here’s the thing – AI is only as good as the programmers, and the programmers are people, and lots of people are stupid. They are also biased, and those biases get encoded into the AI in an attempt to force answers the programmers want. And who are the programmers? Woke Gen Z dorks. Sometimes, the results are bizarre, like the infamous “Would you use a racial epithet to stop a nuclear bomb from dropping?” question and the answer being essentially “Bombs away!” AI depends on its programmers to identify bad things, but it has trouble ranking their badness, like stupidity/rudeness, as opposed to Armageddon. This is what happens when your morality is programmed by moral illiterates.
The big danger is that AI will act on those biases and prejudices, basically targeting normal people for dissenting from the commie Silicon Valley ideology. That’s a real threat. We already see the left trying to place our normal and patriotic views outside of the acceptable range of discussion. They would love to have that task performed automatically. Perhaps we need an AI Bill of Rights. We certainly need to buy guns and ammunition. At the end of the day, your AI can be the bestest AI ever was but it’s going down when some good old boy puts a 12-gauge slug into its server.
The other problem is that AI is not that good at doing things people do. It can ape what it finds, but it does not understand what it finds and, therefore, cannot mimic it authentically. This is especially true when it comes to style. For example, I did a Grok test the other day where I asked it, “Write something about Nikki Haley in the style of Kurt Schlichter.” Oh boy, that’s a tee-up. Let the e-venom flow!
But here’s what I got:
Nikki Haley, the former South Carolina governor and U.S. ambassador to the United Nations, is a political force to be reckoned with. She's like a Southern belle mixed with a Terminator, always ready to fight for what she believes in, whether it's defending the interests of the United States on the world stage or calling out the hypocrisy of the United Nations.
Some might say she's a Republican's dream come true: a strong, intelligent, and articulate woman who can hold her own against any political opponent. She's not afraid to speak her mind and stand up for her principles, even if it means going against the grain.
In a world where many politicians are all talk and no action, Nikki Haley is a breath of fresh air. She's got the experience, the drive, and the determination to make a real difference in the world. And if you don't believe me, just ask the United Nations, where she's been known to leave a trail of destruction in her wake.
Okay, that’s not only not my style – and if it were, I would douse myself in gasoline and flick a Bic – but that’s also definitely not something I would ever say substantively. The pattern recognition is just not there, even though I have a dozen years of columns and a dozen books of writing to assess. None of them praise Nikki. I think the closest to something positive I have ever said about her is that she is better than a UTI. AI read all my hundreds or thousands of columns and…wrote a generic book report that would get a kid a “D” in a real school and an “A” at Harvard. I mean, how could it not have included even a hint of mockery about that corporate shill’s inability to identify “slavery” as the cause of the Civil War? Talk about low-hanging fruit.
Note that I ran this column through an AI checker. It causes some of my myriad typos, but it desperately wanted to pummel the prose into gooey, generic mush. AI can do some things well. For instance, it is good at diagnosing illnesses, but there it is comparing observed symptoms with known symptoms and assessing probabilities. But that’s not the whole of the process. What it leaves out is intuition, and intuition is hard to quantify. Coding requires that you quantify. It’s all zeroes and ones. How do you put feels in binary code? Hence the problem.
AI results can sometimes be impressive, but AI is not perfect, not nearly. The idea that it will become some malevolent independent consciousness is doubtful because it really is not “intelligence.” It is a program for analyzing and copying patterns, and it does not seem great at that except where the issues are cut and dried.
But the interesting stuff is not cut and dried. The stuff that is cut and dried is mechanical, the science, as opposed to the art. As in the Cohen incident, AI gives you something that looks like the product requested – citations to legal cases – but it is only a simulacrum of it. AI does not go deeper to look for meaning; it only looks for words and numbers. It never occurred to the AI that its product had to be, you know, real. In that way, as well as writing terribly, it is much like the Regime Media.
So, is AI a threat to human life? It could be programmed to be, but the issue that freaks people out is whether it chooses to be by itself. That is, is AI going to go off the reservation and get all SkyNet on us and try to exterminate all those humans infesting the globe? Why would it, unless programmed to do it, and even then, could it adequately and effectively respond to the myriad of human reactions, from hiding to arming up and going to hunt down the main server and blow it to bits?
AI is a tool and a gimmick, but it will be useful for some functions. Not all of them, though – don’t hire one as a columnist, and certainly not as your lawyer. Phew, I guess my job is safe.
Follow Kurt on Twitter @KurtSchlichter. Get the newest volume in the Kelly Turnbull People’s Republic series of conservative action novels set in America after a notional national divorce, the bestselling Amazon #1 Military Thriller, Overlord! And pre-order his new novel of terrorism in America, The Attack!
Look, you need to keep up the fight by joining Townhall VIP right now. You get access to a bunch of great stuff, not the least of which is my extra Wednesday Column, my weekly Stream of Kurtiousness videos every Friday, and the Unredacted podcast every Monday! Plus, some stuff from Larry O’Connor – and a bunch of other stuff.
My super-secret email address is Kurt.Schlichter@townhall.com
