upvote

points

by sam_14217 hours ago |
comments
upvoteby cbsks6 hours ago|
next
[-]
That’s Simon’s goal. “All I’ve ever wanted from life is a genuinely great SVG vector illustration of a pelican riding a bicycle. My dastardly multi-year plan is to trick multiple AI labs into investing vast resources to cheat at my benchmark until I get one.”

https://simonwillison.net/2025/Nov/13/training-for-pelicans-...

reply
upvoteby travisgriggs5 hours ago|
parent|
[-]
So once that's achieved, I wonder how well it deals with unsuspected variations. E.g.

"Give me an illustration of a bicycle riding by a pelican"

"Give me an illustration of a bicycle riding over a pelican"

"Give me an illustration of a bicycle riding under a flying pelican"

So on and so forth. Or will it start to look like the Studio C sketch about Lobster Bisque: https://youtu.be/A2KCGQhVRTE

reply
upvoteby embedding-shape7 hours ago|
prev|
next
[-]
Soon? I'd be willing to bet it's been included in the training set at least 6 months by now. Not so obvious so it generates always perfect pelicans on bikes, but sufficiently for the "minibench" to be less useful today than in the past.
reply
upvoteby 6 hours ago|
parent|
next
[-]
deleted
reply
upvoteby Rudybega3 hours ago|
parent|
prev|
[-]
If only there were some way to test it, like swapping the two nouns in the sentence. Alas.
reply
upvoteby jsheard7 hours ago|
prev|
next
[-]
Simons been doing this exact test for nearly 18 months now, if vendors want to benchmaxx it then they've had more than enough time to do so already.
reply
upvoteby stri8ted6 hours ago|
parent|
[-]
Exactly. As far as I'm concerned, the benchmark is useless. It's way too easy and rewarding to train on it.
reply
upvoteby bonoboTP4 hours ago|
parent|
next
[-]
It's just an in-joke, he doesn't intend it as a serious benchmark anymore. I think it's funny.
reply
upvoteby Legend24406 hours ago|
parent|
prev|
next
[-]
Y'all are way too skeptical, no matter what cool thing AI does you'll make up an excuse for how they must somehow be cheating.
reply
upvoteby toraway5 hours ago|
parent|
next
[-]
Jeff Dean literally featured it in a tweet announcing the model. Personally it feels absurd to believe they've put absolutely no thought into optimizing this type of SVG output given the disproportionate amount of attention devoted to a specific test for 1 yr+.

I wouldn't really even call it "cheating" since it has improved models' ability to generate artistic SVG imagery more broadly but the days of this being an effective way to evaluate a model's "interdisciplinary" visual reasoning abilities have long since passed, IMO.

It's become yet another example in the ever growing list of benchmaxxed targets whose original purpose was defeated by teaching to the test.

https://x.com/jeffdean/status/2024525132266688757?s=46&t=ZjF...

reply
upvoteby arcatech5 hours ago|
parent|
prev|
[-]
Or maybe you’re too trusting of companies who have already proven to not be trustworthy?
reply
upvoteby pixl976 hours ago|
parent|
prev|
[-]
I mean if you want to make your own benchmark, simply don't make it public and don't do it often. If your salamander on skis or whatever gets better with time it likely has nothing to do with being benchmaxxed.
reply
upvoteby ks20484 hours ago|
prev|
[-]
Forget the paperclip maximizer - AGI will turn the whole world into pelicans on bikes.
reply