- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
Anyone who has been surfing the web for a while is probably used to clicking through a CAPTCHA grid of street images, identifying everyday objects to prove that they’re a human and not an automated bot. Now, though, new research claims that locally run bots using specially trained image-recognition models can match human-level performance in this style of CAPTCHA, achieving a 100 percent success rate despite being decidedly not human.
ETH Zurich PhD student Andreas Plesner and his colleagues’ new research, available as a pre-print paper, focuses on Google’s ReCAPTCHA v2, which challenges users to identify which street images in a grid contain items like bicycles, crosswalks, mountains, stairs, or traffic lights. Google began phasing that system out years ago in favor of an “invisible” reCAPTCHA v3 that analyzes user interactions rather than offering an explicit challenge.
Despite this, the older reCAPTCHA v2 is still used by millions of websites. And even sites that use the updated reCAPTCHA v3 will sometimes use reCAPTCHA v2 as a fallback when the updated system gives a user a low “human” confidence rating.
You must log in or register to comment.
Aren’t these Captchas designed to get training data for AI models anyway?
“System does what it was designed to do” doesn’t feel that surprising…
Aren’t these Captchas designed to get training data for AI models anyway?
Yes and no, the captchas are just meant to be hard for computers to solve but easier for humans. People saw that, and thought that “if we’re making people do this might as well have them do something useful” not meant to be malevolent- and the purpose is still stopping bots, training them is a side-effect.
No, you’re wrong, the Traffic Light examples ARE specifically to gather data to train models. Being a good Captcha was just a byproduct of that. If people just wanted a good captcha they wouldn’t need hundreds of millions of photos of street lights and bicycles.
No, you’re wrong, the Traffic Light examples ARE specifically to gather data to train models.
No you’re wrong, because the sites that embed those captchas on their page are not doing that to help good.
If people just wanted a good captcha they wouldn’t need hundreds of millions of photos of street lights and bicycles.
Yes, they are getting something productive out of the human labor that would be done anyways. Trust me as a web developer, and web scraper, some kind of captcha is necessary for many free services to be useful/economically viable. The core of a good captcha is just making it marginally more expensive for the scraper/bot than it is for you.
The sites don’t create the captcha, you yourself just said it was embedded there.
They embed for a reason… And the captchas wouldn’t exist if they weren’t embedded anywhere
Finitebanjo is right. Yes they are used to fight spam and bots but they way they do it us is picked intentionally to train ai.
https://medium.com/@yennhi95zz/how-google-trains-ai-with-your-help-through-captcha-876cb4eb4d01
Also from the Wikipedia article “Google profits from reCAPTCHA users as free workers to improve its AI research.” https://en.m.wikipedia.org/wiki/ReCAPTCHA
they do it us is picked intentionally to train ai.
Yes like I said, the challenges were picked to be useful. But some form of challenge would’ve been chosen regardless.
I just close the page usually if I see one of these ones, I don’t have the patience to click all the boxes and then it just sends you a different one.
Unfortunately they’re on pages that I absolutely need to get into because my money is stored behind them. I cannot stand them, and I generally agree with you, if some random site has me doing a captcha in leaving.
Thank God this means i can stop wondering if i should click on the… the 13 pixels from the fucking bike in that one corner square or wondering if i should count the scooter as a motorcycle fuck i am so tired of that shit
Complete the obligatory “is this a staircase or street crossing” round only to be roundhouse kicked back to the beginning.
Our long international nightmare is finally over!
While a new one has started!
the new ones suck so fucking much though
If I see the newer ones pop up at all I just skip what ever the task is that was requiring me to bother with it.
i love when websites (twitter is a really bad example) hit me with like 8 captchas, and then if i get my username/password wrong i have to do another 8. It’s just so obviously gaming for training data on shit lmao.
What is it actually training? Google owns captcha right?
i have no clue, but i would assume it’s native to twitter if they’re pushing it that hard, either that or someone is paying a lot of money for that captcha access lol.
Cool, so can Google shut it down now?
deleted by creator
deleted by creator
it’s super ableist. if someone has poor vision or colorblindness chances are they’re going to miss things.
FYI as someone that’s colorblind these captcha’s don’t seem to have anything specially relevant to being colorblind in them.
Now if they start showing me a dozen traffic cones and asking me to pick the green one, we might have a problem.
I have regular everything and I still fuck them up. “click the ones with a fire hydrant”. But a tiny piece of fire hydrant is spilling into another box. Does it count? Does it not count? Good luck!!
I had one the other day that was deep fried jpegs to the max. Like, what the fuck am I supposed to do.
Spillovers into other boxes definitely count…
I don’t want to do this next part but I can’t resist…
Just ask my girlfriend…
Ba dum tiss
Sprinkle powdered sugar on them. Delicious deep fried jpegs.
They offer a sound option right below.
a hard to see option, aptly enough
Technically the “correct” answer is set by the highest percentage of people choosing it. EG: 19 people select Box A and 1 selects Box B, then the machine decides Box A is in fact correct.
That means these AI could be selecting the wrong answers for all anybody knows, if enough of them are answering the prompts, and still passing.
I mean, we literally train them by completing the CAPTCHAs. Why do you think you were picking things like bikes, traffic lights, cars, and busses? The only question now is what’s next…
they embed dark souls into the browser
In order to pay your utility bill, you have to beat the Undertale Sans fight in Genocide mode
Lmao
Prove you’re human: Beat this level of DOOM.
There is a Russian captcha solver bot called xevil that costs under $100 (I think, last time I looked) that has been able to solve nearly all captchas for years. You just have to supply it with relatively expensive proxy IP addresses because Google rate limits solve attempts.
So the title of this article has been true for a long long time. Capatchas are absolutely useless except against poor or uninformed script kiddies.
I can see a future where the Internet is completely run by bots and AI to the point where no human actually uses the Internet anymore.
It’s like an island that gets overrun with rats - there are just too many to deal with so you leave.
Basically Cyberpunk, people only interact with the night city intranet because the global internet has been taken over by AIs.
I’m already doing that now. If Lemmy starts showing signs of fuckery I’m out. I’ll switch back to magazines.
I already did… There’s some subscription stuff where you can read pretty much all available magazines and papers, it’s been a long time since I’ve been reading that much “news” and reports
I work in a place with no phones. I bring books and magazines into the shitter.
Yeah, I predict that in the future, you can’t expect that content on the internet is written by humans. If you go to the internet, then it will probably not be to connect to other humans. Maybe you want to know something that a bot can tell you or you have some administrative task to fulfill, like filing a form.
Some believe this happened years ago. Check out Dead Internet Theory.
Pro-tip for webscrapers: using AI to solve captchas is a massive waste of effort and resources. Aim to not be presented with a captcha in the first place.
I think thats much more difficult than it seems, because usually only residential IPs are the ones that don’t get those. And if you start to use a residential proxy too much then that IP can also get flagged.
That’s why companies like Oxylabs exist
That’s when you rotate the proxy. By default most residential proxies will give you a new proxy for each request unless you specify.
Depends on the case, sometimes its unavoidable.
Well we trained them
Unless this was something people could use i dont rly see it becoming much of a problem. Most people dont even use adblockers
I never get the first one and rarely the second one. If it says to click all the squares with motorcycles and it’s just the one big picture, am I supposed to click stuff like the tire and mirrors? I always do and never get it right. Then most of the time they ask me to identify motorcycles, they show me motor scooters and what am I supposed to do then? I think I just need to get one of these bots to do it for me.
Fwiw they aren’t really asking about the motorcycle. I mean they are but they are washing your mouse movements and how fast you click through the images. It’s okay to get a few images wrong.
Not quite.
It’s mostly wisdom of the crowd, as it always has been.
As long as you mostly click the same squares most other people click, you pass.
You often at random get 2-3 images because 2 of them are actual checks, but the third is a new image that you auto pass and they’re using it to gather data on what the average clicks are on it.
I bet you use the word “actually” in conversions regularly.
I bet you make incorrect assumptions about people you don’t know regularly.
A motor scooter is a motorcycle in the eyes of the law.
Not everywhere.
Where I am, you need a special license to drive a motorcycle, classified as having an engine of 51ccs or more, whereas a scooter is any motorcycle with a less than 51ccs and doesn’t require a special license.
That’s a moped, a motor scooter is just a style of bike.
Mopeds are similar but have pedals and can be used as a bicycle. The name itself, Mo-Ped" is a portmanteau for motor and pedal.
Motor scooters are different in that they have a cut out for the rider’s legs/ feet so they don’t have to straddle it the same way they would with a motorcycle. Both mopeds and motor scooters do not require a license endorsement here, while motorcycles, as defined in my original comment, do.
https://en.m.wikipedia.org/wiki/Moped
The definition in most countries is less than 50cc and a maximum power output, they do not need to have pedals. Anything over 50cc is a motorbike.
So a moped with a 49cc engine, astep-through design, and no pedals is a moped but a scooter with a step-through design and 49cc engine is a motorcycle?
That’s confusing as fuck lol
Not really, both your examples are a moped. The definition of a moped in most places has nothing to do with the style of bike.
