“These language programs do write some “new” things—they’re called “hallucinations,” but they could also be described as lies. Similar to how autocorrect is ducking terrible at getting single letters right, these models mess up entire sentences and paragraphs.”
“Lies” assumes an intelligence; subterfuge. It is anthropomorphizing software and its bugs.
The proper term is bug. When I write code and it delivers the incorrect response, that is a bug.
We need to stop with incorrect and factually flawed terms — ChatGPT doesn’t have hallucinations, it has incorrect and erroneous responses.
Bard for example, couldn’t nail how many indictments there were under the Clinton Administration. How many were convicted? Throw of the dice seems to be Bards way of solving it.
These aren’t lies, they are bugs. They are fundamental to the FAILURE of the system as a whole.
These bugs would be inexcusable in any other REAL software industry.
The “AI” is just code. It is an amalgam of weighted information. But if I had released code for a traffic light that performed this poorly, I would be out of business.
There is a LOT of money being made from the hype and a LOT of people pretending the hype is real and IGNORING the real problems in the code bases.
I’ve been writing code for over four decades and it doesn’t take me but three false propositions to get @Alfred to agree that “tireless” can mean one is without tires.
These systems are inherently flawed, fundamentally untrustworthy and the bugs are ignored in favor of “oh, look, it’s like a human being” — oh, no it is not.
We need to stop pretending that this is “AI”. It is not.
Perhaps “AI” is suffering from economic anxiety and that is why its performance is so poor? 🤣
@feloneouscat @DavidSalo @Alfred It was a pretty compelling argument…
@feloneouscat we want to humanize it’s errors. Makes us feel warm and squishy while we redo what we asked it to. LOL
I see the nature of “AI” (LLM as I don’t really feel there is intelligence associated with them — hah! I kid!) as mostly error with little to no Q/A.
If the results look similar to what they think it should be, it’s a win.
Testing for success feels great, but it doesn’t really work in the real world.
Real engineers test for failure.
@feloneouscat "Throwing the dice" is actually not an entirely inaccurate high level depiction of how generative transformers work, as they're statistical language models.
FWIW, and this isn't at all me excusing the wildly overstated claims of LLM accuracy and capability, "hallucination" is a term of art in this field referring to the model creating assertions that are both false and not in the original data set. They're a big problem! LLMs lacking a concept of truth is inherent in their design.
@lenaoflune "’Throwing the dice’ is actually not an entirely inaccurate high level depiction of how generative transformers work”
I know. That’s why I said it.
“Hallucination” is a term that is wildly inaccurate, highly anthropomorphic, and really doesn’t describe the nature of the problem or indicate how the issue may be resolved.
A bug by any other name still goes on the punch list.
@feloneouscat
This is just historical and political misinformation. What happens when people go to AI for recipes and get food poisoning? What happens when they ask it for health information and make themselves sick? What happens when they treat it as a counselor and become suicidal?
Nobody's taking any responsibility for what the machines spit out. Yet the AI itself cannot be held responsible for anything.