In a world where half the voting public in the US can still not just take this visible-from-space con man seriously,
but even elevate him to messianic status,
I am the last person on earth to underestimate the impact on the gullible that AI will have in the hands of skilled liars:
At the same time, the notion that all AI is some kind of master technology that no human mind can possibly distinguish from reality is, well, overrated:
Lots of AI is transparently dumb and is wielded by dumb people (like me) in ways that are far more hilarious in their results than menacing harbingers of the apocalypse. I don’t just mean stuff like my attempt to get AI to render pics of “Mark Shea, Catholic apologist, on a dinosaur”:
I mean magical thinking like this too:
Here’s the thing: My name is Mark. It is derived from Mars the Roman god of war and one of the false gods of the nations, according to to the Tanakh. There is no Hebrew equivalent for Mark because people who speak Hebrew are rather famously noted for not worshipping Mars. In antiquity, if Mars was mentioned in conversation with a Hebrew speaker, the person mentioning Mars was likely a Roman speaking Latin, not Hebrew (though, of course, as the example of St. Mark makes clear, in the polyglot world of the ancient Mediterranean, it was common for Jews to have, so to speak, indigenous names and also names common among Gentiles).
Claiming every name is a Hebrew name or has a Hebrew equivalent reminds me of that moment in MY BIG FAT GREEK WEDDING when the Papa is proudly declaring that all words come from Greek roots and challenges somebody to name any word and he will give them the Greek root.
They toss out a word like “kimono”.
Whenever a new technology becomes ascendant, pop culture gloms on to it for a while and does what science fiction loves to do: speculate on the question, “What if this current trendy thing becomes the only thing forever and ever?” 30 years ago, it was cloning and people were all into Jurassic Park. Before that it was space travel. Before that nuclear energy was the magic pixie dust that would shrink us, make us giants, confer superpowers or create giant insects. Magnetism was going to magically cure everything a century ago.; Before that, it was electricity, or hypnotism, or petroleum.
Now AI is the New Hotness that will rule all. Is it a huge danger? Yes. Is it omnipotent? No. It can do a lot. But it can’t do things that AI, by its nature, cannot do.
Advertisement
8 Responses
You’re right, but if I might offer a minor linguistic quibble.
You are right in that Hebrew and Latin are entirely separate languages and cultures. Any common origin they might have shared is lost in time. Hebrew is entirely a Semitic language of the Near East, and Latin has its roots in Proto-Indo-European. To find their common connections you’d probably have to go back to Stone Age Africa. They are no more linquistically related than a wolf and a tiger are genetically related. 100% correct.
That said …
“Marcus” probably did have a precise meaning in Latin. What that might have been, I don’t know, but for the sake of argument let’s say it meant “strong arm.” You could then go over to Hebrew and find out how to say “strong arm” in Hebrew. So in that sense, you could find a Hebrew equivalent to Marcus.
It’s basically the same methods nerds use for “finding” their name in Quenya or Sindarin.
I should have invited Trump to my daughter’s wedding. He could have saved me money on the bar bill by changing water into wine.
Regarding the name Mark:
https://en.wikipedia.org/wiki/Marcus_(name)
Imagine a hospital spends millions to buy a machine that’s supposed to perform open-heart surgery so they can fire their surgeons and cardiologists, and you look at the machine and it’s clearly just a wood chipper. That’s AI.
That is, aside from the image-generating diffusion AIs, most of what’s marketed as “generative AI” right now is actually a Large Language Model, or LLM. They’re fairly sophisticated statistical models of which words in a language are likely to be found next to other words. They are “trained” on vast amounts of text, mostly scraped from the Internet without the consent of the people who wrote it. (“Conditioned” is probably a better word than “trained” though; training implies learning, which implies at least some kind of understanding or some concept of truth/correctness.) What they’re doing is basically a more complicated version of your phone keyboard’s auto-suggest feature, and much more effective at imitating the structure of coherent writing.
That’s it. That’s all they do. Because that’s all they were ever designed to do. The problem (well, one of them) is that the people marketing it keep trying to act like it is a reliable source of information, and the people buying it keep believing it (or pretending to because they can’t resist the prospect of free labor that never complains or unionizes). Which is how you get AI telling people to eat poison mushrooms, advising people with eating disorders to go on a diet, spouting nonsense science citing fake papers by imaginary researchers, or even (in at least one case) falsely accusing people of sexual assault. But no, see, it’s totally okay, because they slapped an “AI sometimes makes mistakes” disclaimer on it. (That’s also a lie. The LLM is doing exactly what it’s supposed to do; the techbro hucksters are just lying about what that is. The only mistake is trusting them.)
And that’s entirely aside from the problem of the image and voice AIs being used to create deepfakes. And the massive environmental impact of the huge server farms required to run these things once they get big enough to be somewhat effective. And the fact that those giant data sets are mostly stolen.
Bravo!
Aw, thanks!
Good point about which form of science is the most “magical” in the current zeitgeist. Circa 2002 I noticed that, whereas the original Hulk was caused to mutate by radiation, and Peter Parker was bitten by a radioactive spider, both of the film reboots changed that to deliberate genetic modification.
Never forget Kryptonite!!