Mar. 19, 2023
by Dr. David Fialk, Editor / Publisher
Alan Turing, a code-breaker in World War Two, is credited with inventing the first computer, the Turing Machine. He proposed what became the Holy Grail of computer engineering for the last 80 years. Passing the Turing Test, as it's known, involves the creation of a computer that can fool you into believing that you are speaking with a human. With the recent unveiling of ChatGPT, in some sense the Test has been passed.
At least, a lot of people are being fooled. College students, corporate personnel and scientists are passing off ChatGPT-generated content as their own.
ChatGPT is a Large Language Model. Through overwhelming computing power, using enough electricity to power a small town, it collects (plagiarizes) relevant information from the web and synthesizes it into something resembling a human presentation. Of course, with a bit of time, this is something our human brain could do using less electricity than it takes to illuminate a 25-watt light bulb.
The tricky thing about ChatGPT is how it chats. You write asking it to do something. It writes back, producing what you asked for. You dig down, adding some parameter to the task. It revises its product according to the new condition. You ask and it answers. This dialogue fools us into believing that it's a thinking machine. But it's our programming, not its. We are programmed to attribute intelligence when there is dialogue.
A human habit that ChatGPt has picked up is lying. It makes things up. "Hallucinates" is the term used. For example, it cites scientific articles that don't exist, choosing plausible real life researchers as their authors. And it's easy to confuse it, and so get it to give strange, absurd responses.
Critics compare ChatGPT, the state of the art of Artificial Intelligence (AI) to a music box. A music box doesn't know the difference between Mozart and the Rolling Stones, and can't change its tune. Facebook's algorithm, that sends you content like the content that you have previously lingered over, is AI. The auto-complete that suggests words as you type is AI. ChatGPT is auto-complete on steroids.
An author in town messaged me the other day telling me that she had been prompting ChatGPT to write. She wondered if I would be interested in seeing what it came up with on the theme of how Americans could accommodate Mexican manners. When you edit a magazine, you don't say no to a possible article. When all you have is a hammer, everything looks like a nail.
The material on Mexican culture and manners was presented coherently enough. It was a reasonable survey of the main mistakes we gringos make here, south of the border. But it was dry, sterile, lacking all idiosyncrasy, nuance or quirks. There was no original thought, let alone poetry. One might be fooled into thinking that it was written by someone who wasn't a very good writer.
One might also be fooled into thinking that science has basically figured everything out, and that all we are doing now is filling in the gaps. But it hasn't. The closer we look, the more intricacy there is to see. Everything: biology, physics, matter, intelligence, is unimaginably complex, fine-tuned and mysterious.
I know all this from watching a lot of science videos on Youtube, featuring top scientists chatting away. On one of these, the author of Resisting AI, laments that so much effort is going into trying to teach machines to do what we humans will always do better. Getting a robot to vacuum your floor, or asking ChatGPT to compile a list of Mexican manners is making good use of a tool. But, once your house is clean, inviting your Mexican acquaintance over for dinner, and asking them their opinion is the way to go.
On another Youtube video, someone told a story about visiting Graceland, where he noticed a couple, one member of which said to the other, referencing the guided tour tablet in his hand, "If you swipe it, you can see the whole room." The observer couldn't resist butting in to point out, "If you stop looking at the tablet, you can see the whole room by turning your head."
Lokkal is the interface between the online world and the actual world. It is digital technology in service of the local community. Thank you Big Tech, but we don't need your algorithm. We'll present our own city to the planet. Lokkal, humanly-curated local internet.
Dr. David presents Lokkal, the social network, the prettiest, most-efficient way to see San Miguel online. Our Wall shows it all. Join and add your point of view.
Please contribute to Lokkal,
SMA's non-profit community publication:
Watch the two-minute video below.
Then, just below that, scroll down SMA's Community Wall.
Visit SMA's Social Network