How would anyone trust the products these people put worth? They aren’t working on making LLMs more accurate (spoiler alert: they can’t, by design), they’re working to make them more appealing to companies targeting unsuspecting consumers. By any means necessary.
There’s an LLM called Grok? What the fuck. Seriously. Tech bros understand the absolute most obvious thing from your favorite piece of science fiction, please.
The meaning of grok is in the plain text! We’re not even talking about obvious metaphor, it’s just a made-up word for truly deep understanding and you’re going to name your idiot, fabulist, word generator after it?! #grok#LLMs#TormentNexus#StochasticParrot#Tech
With more and more output from chatbots tainting all of the web, the quality of search results will go down, especially for queries where there are few actual results (for example because the thing searched for doesn't exist): Chatbot transcripts or things the chatbots integrated into some search products will fill the void with garbage. Thanks "AI".
Was talking to @Beantin about the “amazing” feat where ChatGPT passed the bar exam. We agreed that if you feed all the relevant content for the bar exam into ChatGPT there really should be no big surprise about it being able to spew out statistically relevant content.
The fact that it still got such a relatively low score should be a cause for worry, not celebration(!) It’s evidence that the tool has no understanding of what it is doing. It has the answers, it’s just not able to use them in the right way. It’s like a student sitting with a textbook with all the answers to the test and not being able to understand which answer fits where.
As Paris Marx wrote in March:
"it’s so funny to me that the AI people think it’s impressive when their programs pass a test after being trained on all the answers”
In my latest Locus Magazine column, "Plausible Sentence Generators," I describe how I unwittingly came to use - and even be impressed by - an AI chatbot - and what this means for a specialized, highly salient form of writing, namely, "bullshit":
If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:
Want to generate 2,000 words of nonsense about "the first time I ate an egg," to run overtop of an omelet recipe you're hoping to make the number one Google result? ChatGPT has you covered. Want to generate fake complaints or fake positive reviews? The #StochasticParrot will produce 'em all day long.
It wasn't just Ottawa: #MicrosoftTravel published a whole bushel of absurd articles, including the notorious Ottawa guide recommending that tourists dine at the #Ottawa#FoodBank ("go on an empty stomach"):
If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:
These were very likely written by a #StochasticParrot, and they comprised training data for a human intelligence, the poor schmucks who are supposed to remain vigilant for the "hallucinations" (that is, the habitual, confidently told lies that are the hallmark of AI) in the torrent of "content" that scrolled past their screens:
I was supposed to be on vacation, and while I didn't do any blogging for a month, that didn't mean that I stopped looking at my distraction rectangle and making a list of things I wanted to write about. Consequentially, the link backlog is massive, so it's time to declare bankruptcy with another #linkdump:
But every now and again someone gets a #StochasticParrot to do something genuinely delightful, like the #JollyRogerTelephoneCompany, who sell chatbots that pretend to be tantalyzingly confused marks in order to tie up telemarketers and waste their time:
Jolly Roger sells different personas: "Whitebeard" is a confused senior who keeps asking the caller's name, drops nonsequiturs into the conversation, and can't remember how many credit-cards he has.
"Bots, Rhymes & Life: Ethics of Automation as If Humans Matter"
Sets a faux rap battle ChatGPT4 v A Tribe Called Quest shows absurdity of ChatGPT4 as capable of thinking. It is as Drs. Bender & Gebru argue a #StochasticParrot
Joseph Weizenbaum, the author of the famous #Eliza chatbot was somewhat startled and dissapointed by the world's reaction to his software.
This prompted him to write the 1976 book "Computer Power and Human Reason: From Judgment to Calculation"
I found a PDF last night and started reading it. A lot of it is applicable if not a remarkable mirror for the #StochasticParrot mania we find ourselves in.
@bkrupp @ACM Except when studying Language Models, there implications, etc. using ChatGPT et al. in a publication (to generate the publication itself) is a bad idea, no matter your subject. A #StochasticParrot will only generate a lot of verbose text.
“LLMs not only fail to properly generate correct Python code when default function names are swapped, but some of them even become more confident in their incorrect predictions as the model size increases, an instance of the recently discovered phenomenon of Inverse Scaling, which runs contrary to the commonly observed trend of increasing prediction quality with increasing model size” https://arxiv.org/abs/2305.15507
"Folks totally don't understand much about this ridiculously serious AI tech that is literally blowing up our world with its insane ability to think like a person. This shit is incredible—believe it!"