FOUND IT
-
@resuna @lokeloski good old Knoll's Law/Gell-Mann Amnesia Effect, so striking it was named twice.
I first noticed this effect without having an eponym for it in the late '80s early '90s when reporters started reporting on the nascent internet, which is something that I knew quite a bit about, and they always made out that they knew what they were talking about but what they came up with was such utter authoritative twaddle that I decided that the main skill set for journalists and reporters was sounding confident.
-
FOUND IT
@lokeloski Also, i feel like a lot of the coding examples seem to focus on “boilerplate” stuff which makes sense… as that is the stuff that has tons of examples online that is probably part of the training set.
-
@SteveClough @lokeloski well, *someone* doesn't work in software
@Nymnympseudonymm @lokeloski Someone may not, but I do.
-
FOUND IT
@lokeloski@mastodon.social Thinking that a slop answer generated by an LLM is "right" requires sufficient ignorance of the subject that you don't understand why the LLM's answer is wrong. If you actually understood the subject, you wouldn't have asked an LLM in the first place.
-
FOUND IT
@lokeloski And this explains why the place where you just can’t get away from AI enthusiasts is LinkedIn.
-
I first noticed this effect without having an eponym for it in the late '80s early '90s when reporters started reporting on the nascent internet, which is something that I knew quite a bit about, and they always made out that they knew what they were talking about but what they came up with was such utter authoritative twaddle that I decided that the main skill set for journalists and reporters was sounding confident.
@resuna This is not fair; imagine being a journalist who writes about tons of things and is expected by its outlet to produce 3 to 4 stories every week. By default journalists are allrounders; not -experts. Their main task is to find the real experts and translate their knowledge for a large audience. @LilFluff @lokeloski
-
FOUND IT
@lokeloski Oddly I feel programming has the opposite effect where programmers think only their own field can be automated. What's up with that?
-
@lokeloski And like I said the last time I saw it: no one considers asking the script writer or the concept artist or whatever... because when they do it's always boring stuff like "oh I just need to know what scene this character last appeared in, and I can do that with Ctrl+F"
@FrankHghTwr @lokeloski Back in the 90s the Writer's Guild of America (West) had a campaign in the entertainment magazines with famous movie quotes like "I'll have what she's having," and the tagline "Somebody wrote that."
-
@resuna This is not fair; imagine being a journalist who writes about tons of things and is expected by its outlet to produce 3 to 4 stories every week. By default journalists are allrounders; not -experts. Their main task is to find the real experts and translate their knowledge for a large audience. @LilFluff @lokeloski
> Their main task is to find the real experts and translate their knowledge for a large audience.
The point is that they were too often utterly failing at that while pretending to actually know what they were talking about, and being good enough at pretending to pass their nonsense off as authoritative truth.
-
FOUND IT
@lokeloski The good old Knoll's Law/Gell-Mann Amnesia Effect finds a new niche.
-
@geeeero @lokeloski important to note the Gell-Mann effect is made up trash. It's literally something Crichton said once. So imagine how cognitive psychologists feel about it.
@davidgerard @geeeero @lokeloski
The story may be made up but the effect is real. I started noticing it in journalism in the '90s... journalists often seemed authoritatively good at stuff I didn't know anything about, but as soon as they started writing about the Internet, or anything else that was at the time a bit esoteric but I know a lot about, their stuff was obvious twaddle.
See also Knoll's Law:
-
@mathew @resuna @geeeero @lokeloski "ultracrepidarianism" has long been a word
-
@lokeloski Oddly I feel programming has the opposite effect where programmers think only their own field can be automated. What's up with that?
@robinsyl
I suspect that with programming, bad programmers love AI because it “levels them up” (or appears to!), but the best programmers despise AI because they can write much better code much faster by the conventional route. -
@mathew @resuna @davidgerard @lokeloski In the arena of science, physicists are similar:
-
> Their main task is to find the real experts and translate their knowledge for a large audience.
The point is that they were too often utterly failing at that while pretending to actually know what they were talking about, and being good enough at pretending to pass their nonsense off as authoritative truth.
@resuna true and for that you can hold them to account. But not that they are not expert in every field and don't know as much as some people who have been sinking their teeth in a topic for ten years straight.
-
@lokeloski whenever I do end up trying to use gen AI for anything, it's precisely because I know I'm not good at whatever I ask it to do, knowing it'll spit out something mediocre to either use as a placeholder so I can at least move on with the stuff I am good at or as something that's "good enough" if it's of lower importance but still needs to be present. If it was as quick and... dirt cheap to commission, that might be good for me but let's be real, commissioning stuff from people who actually know what they're doing shouldn't be paid bargain-bin style.
Some of my D&D characters have faces now because this one time I messed around with Stable Diffusion a little. It's not something I'd have been able to justify spending money on to get commissioned, and the results from the gen AI were mediocre at best but they serve their purpose of visualising my characters at least more than just words on paper. I don't have an amazingly vivid imagination and I definitely don't have that of anyone else so sometimes it's just convenient to have a visual representation that can be shared with others in this case.
And tbf a lot of the time when I want AI to fill in the gaps, it's not actually good enough to do so. The sad reality is that even with the more ethical approach of using AI as a tool to help with productivity rather than replacing people, it's usually not nearly good enough to actually help. Oftentimes, just struggling to do it yourself anyway and producing something crappy is still better than an AI generated hallucination that looks alright at first but progressively gets worse with every passing second of looking at it. I'll sometimes use it for stuff that it's sufficiently okay at and when it makes sense, and will certainly never claim credit for what pops out of it. But besides that it's just often not even useful in the first place.@al1ce_rabb1t @lokeloski
There are perhaps parallels with desktop publishing. Amateur results are good enough for many uses, but if you really want a brochure or a book fit for publication, you will get a significantly better result if you trust a professional. -
@resuna true and for that you can hold them to account. But not that they are not expert in every field and don't know as much as some people who have been sinking their teeth in a topic for ten years straight.
I don't think you're even arguing with my actual point.
-
@mathew @resuna @davidgerard @lokeloski In the arena of science, physicists are similar:
@geeeero @mathew @resuna @davidgerard @lokeloski Very educational! https://xkcd.com/1732/
-
@geeeero @mathew @resuna @davidgerard @lokeloski Very educational! https://xkcd.com/1732/
@NSalwen @mathew @resuna @davidgerard @lokeloski XKCD by Randall Munroe is really a treasure.
-
I don't think you're even arguing with my actual point.
@resuna could be, wisdom signal is low on these channels. Apologies for that.
To me it seems that in these days of fake-news and fascist hatred to journo's, one should not lightly lash out to them...
-
R relay@relay.infosec.exchange shared this topic
