Jaime Buelta
Jason GormanMaybe should have asked ChatGPT about "the tortoise and the hare"
bposi@jasongorman Yeah a damn great idea - in the future you ask for a web site programming and let's see what will be the outcome.
Shannon Quinn 
@jasongorman Did… did Lucas Carlson actually *read* the tortoise and the hare?
@magsol I'm guessing no.
NK30 
@magsol @jasongorman Probably asked ChatGPT to read it.
Rafael@NK30 @magsol @jasongorman I think he did. At least the draw is very good and precise on the analogy…
Robin Ward@jasongorman Also the tortoise isn't facing the computer screen
Timo the timo@eviltrout @jasongorman the tortoise knows exactly what it's doing, you can see from the image that it's a two-row keyboard, probably a wicked custom chord-based keyboard with macros out the wazzoo.
actually, come to think of it, could be steno typing!
in any case, the tortoise is hard core flexing on the rabbit here "i don't even have to look at my screen most of the time to know what's happening" meanwhile the rabbit is grinning ear to ear, probably enjoying the dopamine rush of seeing whether the prompt gives something that works or not, and prepared to "pull the lever" again.
meanwhile, poor tortoise is probably being assaulted by incredible opinions from rabbit over here and has to resist the temptation to explain all the mistakes, but when you're actually doing work, and trying to get done in a reasonable time, it just wouldn't do to be sidetracked into a lecture on why "you can't just verify a person's name" ...
anyway, that's my rabbit-hare-code-competition fanfic
Martin Owens 
@timotimo @eviltrout @jasongorman
The turtle doesn't have to use the screen because they've got their own shell.
I'll get my coat.
dbat 
econads@doctormo oh dear. I salute you.
Zip@timotimo
"You can't just verify a person's name"
Say that to all the sites who say my last name is invalid.
@eviltrout He's *that* good!
Boris@jasongorman @cocoaphony very sad childhood Lucas had… 😎
Iris Young (he/they/she) (PhD)@jasongorman ugh, and the AI generated image can't represent a race track correctly, despite it being... just an oval...
@iris It's one of those races where you go round and round in circles very fast and the one who doesn't get dizzy wins
@jasongorman ah, seems like the tortoise has an advantage there.
@iris Now that I think about it, it's a very apt analogy
Camelia 
🇵🇸@iris @jasongorman there's also grass coming into it lol
fuzzle @ cccfr@jasongorman guess rabbit is fastet, unclear if in scope and target
Large Language Madam@jasongorman Alexa, what is a self-own?
tofuwabohu@jasongorman damn, the tortoise can code without even looking at the screen!
@tofuwabohu That's how I know it was generated by a diffusion model 🙂
@jasongorman the road marking look suspicious as well. And the fact that it's promoting AI :-)
Laurens@jasongorman It writes 50% of it, but the other 50% is needed to correct all the mistakes it's making 🙃
Fabien@jasongorman @ggete Also much slower if your screen is behind your back 🤪
Antiquity/Midnight Spire Games@jasongorman Ah, so Google's applications are written by a mindless token generator and overseen by individuals whose level of media literacy falters at a simple children's story, the moral of which is typically explicitly stated to the listener at the end in six words.
Yes, that all checks out.
airakose@jasongorman also like... google's products have been notoriously bad for a long while so I dont think that's a good metric to compare against
if they didn't already have market saturation from their pre-LLM work, their current output would be failures
Giles Goat@jasongorman I think all this "AI makes things faster" will bring us to the same race to the bottom of "This item is cheaper than the other buy it ! Buy it !" Not to mention "how much land will be converted to run AI servers ?" .. or "How we'll ever go green with more and more power required to run AIs ?".
toerror@jasongorman OP is referencing this 2024 article from google :
research.google/blog/ai-in-software-engineering-at-google-progress-and-the-path-ahead/
The line he appears to be getting the 50% figure from states:
"with an acceptance rate by software engineers of 37%[1] assisting in the completion of 50% of code characters"
@jasongorman I'm not actually sure what the stats there mean in reality, or whether the "AI" suggestions are being mixed with standard static analysis, but it does suggest that two thirds of the suggestions are trash, and you have to read the trash first before going on to write the code yourself anyway.
Michal Měchura@jasongorman Someone didn't read their Aesop!
Tacitus 🇮🇪 
Even ignoring the hilarity of how applicable the tortoise and the hare parable is in this instance, they are still basically saying "my vibe coding is faster than someone who can't even use a computer, their screen is behind them and their keyboard is upsidedown".
Alex Darby@jasongorman Ignoring everything else, "The debate ended while you were still proofreading" gives me anxiety. Is this guy really suggesting using AI to generate code means you don't have to spend time proof reading? That's literally the oppoiste. AI code needs _more_ proofreading than human code, and it's almost entirely proofreading from a human standpoint. Actual coders are being productive while AI coders are still proofreading, unless they don't and the AI kills the codebase.
Tissa The Artista@jasongorman The turtoise is facing away the computer, the racing track behind them goes nowhere. The Hare's hand is entering the laptop's screen.
kieliscalling
JRT@jasongorman "Maybe you're fed up. Maybe you want to be by yourself. Who knows? You look down and see a tortoise, Leon. It's crawling toward you..."
Dwemthy@jasongorman the hare has accidentally deleted the database that holds the race results, but they did it damn fast!
Dustin Mitchell@jasongorman also that 50% number is dubious at best. I suspect it's near to 50% only if you count all the boilerplate config files and whatnot.
Oliver Schafeld
Denis 
@jasongorman It looks like the tortoise is frowning because it is debugging the rabbit code.
myrmepropagandistThis is like some kind of subconscious tell. Good lord. Is kindergarten-level literacy dead?
I would love to see a redraw of this image for use with various essays critical of AI... because that's what it's giving.
George BThis is on par with people defending police corruption as being "just a few bad apples" (a few bad apples what exactly? Finish the sentence)
Daughter of Discord@gbargoud @futurebird @jasongorman
...spoils the whole bunch
Garrett Wollman@gbargoud @futurebird @jasongorman I'm betting a lot of people who use the phrase "bad apple" don't even have a broader context, it's just a common noun phrase.
Log 🪵@gbargoud @futurebird @jasongorman ...spoil the barrel! Which doesn't fall far from the tree. Compared with oranges, which also don't fall far from the tree, and all get moldy if there was a moldy orange in the bag. Anyway, corrupt cops in Florida, California, and Texas might be a few moldy oranges in the mesh bag, but north of the citrus line, they're bad apples in the barrel.
@futurebird I don't think anything could undermine generative A.I. more effectively than its loudest cheerleaders.
chop wood, carry water@futurebird @jasongorman watch this turn into the next "bootstraps" and "bad apple"
God. The bootstraps things drives me nuts. The whole point is that it's nearly impossible, circular.
Though, now "bootstraps" makes me think of systems design first and everything else second.
clewAlso, there’s some “pay aircraft designers by the pound” thinking here. @futurebird @jasongorman
Ben AvelingThere is the theory that scammers like to be a little bit obvious because they only want to deal with ppl who can’t spot an obvious scam. Maybe AI companies feel the same. @futurebird @jasongorman
@futurebird @jasongorman i was thinking of there was an aesop fable that was more apt, but the "quality > quantity" one was really fucking gross (and the "quantity" animal being the fox would really send mixed messages in this social context)
and then i found something far more on point than i was looking for https://aesopfables.co...
and then i found something far more on point than i was looking for https://aesopfables.co...
@futurebird @jasongorman ...or we could go back to hares and shallow quantity versus deep quality https://aesopfables.co...
walkies enthusiast@jasongorman using AI to absolutely crush this doltish moron who codes facing away from the monitor
Alison Wilder@jasongorman I don't really know what's going to happen with AI but I'm pretty sure literacy is out the window
@alisynthesis That's why for and how they
FoolishOwl@jasongorman Aside from obvious ignorance of one of the most famous children's stories of all time, I notice the original comment implies that coders not using LLMs are spending more time proofreading the code, and what I've heard from developers is that any time saved by using LLMs to generate code is less than is lost in additional review and revision time.
So the implication is either the poster wasn't aware of that, or worse, is just committing code without properly reviewing it.
Morten@foolishowl @jasongorman Yeah, I was like: Who is still proofreading? Because surely you are proofreading the 50% of your code made by a computer? A computer that doesn't have ideas, it just fills in the blanks. A computer that doesn't have the full context of the problem you are trying to solve.
abon999@jasongorman The turtle won't get hacked or sql injected. So, the turtle, definitly.