What got me thinking about this is how whenever there's a big pro-AI / AI skeptics are deluded thread, it's always ugly short poor losers like cowgod (who's cuck dad died of diabetes!) hanging onto their last strands of hair with daily dutasteride leading the charge, probably because IRL no one would ever follow them.

So I support this rigorous testing so I don't have to hear another 5'8" faggot who makes $172K/yr. and whose wife cheats on him tell me that actually AI writes great briefs.

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49920440)

And we instituted a bortwide barefoot height-measurement standard in 2023 when desantis was moving from 4-in. to 7-in. lifting footwear in his public appearances.

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49920999)

it wouldn't even be close actually

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49920940)

(If it had the "proper training data", as if that's some trivial detail and not a major limiting factor for AI's IRL usefulness.)

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49921004)

But you see it in legal practice: dudes who used to write 2-page, numbered-paragraph briefs that began with COMES NOW and cited only "US. Consti. amendment IV" are now putting together 8-page bangers that cite the seminal case law in the field. Moving their work-product quality from an 8/100 to a 43 on a normalized distribution. There's just no way that a competent brief-writer can possibly extract that level of improvement over baseline from AI.

... Or are you literally just making the point that "a smart person will get better results, in an absolute sense, than a dumb person," vs. measuring value over non-AI baseline? Because yes I agree that proposition is so obvious even a fat prediabetic cuck like cowgod would understand it.

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49921821)

something like "convincing women to send nudes given a dataset of male picture + bio and female picture + bio" is the exact kind of thing that a neural network would be super good at if you actually trained one for it using the appropriate data

of course the training data is a relevant and important part of the equation. agentic LLM systems will become better and better at different tasks as they're post-trained by specific data sets for different tasks. that's already how they've gotten way better at stuff like coding

(http://www.autoadmit.com/thread.php?thread_id=5871892&forum_id=2),#49921051)