10 ms·

For the car captchas, I've found actually clicking all the boxes with part of a car will always be a wrong answer (distinct from when it just makes you answer t

For the car captchas, I've found actually clicking all the boxes with part of a car will always be a wrong answer (distinct from when it just makes you answer twice). Instead, you have to click on the squares that you know it thinks are cars. This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is.

js8 9y ago
That was my experience with car and sign captchas as well.. But interestingly, it also depends on my mood, when I feel lazy, I click fewer boxes.
pbhjpbhj 9y ago
The problem with the cars for me is their definition -- is that van classed as a car, how about the half-back, what about a 4x4, what about a 4x4 with no back windows, ...
- indigochill 9y ago
  What about a square that contains just a sliver of the car from the next box over? Does that still count? I hypothesize that my attempt to classify every pixel related to a car as a "car" may contribute to my failing of these tests.
  - toyg 9y ago
    This is what kurtisc is saying. The algorithm is unable to classify those pixels, so you have to guess which bits of the car can be detected by the algorithm and then select only those. So you have machines asking humans to think like a machine in order to prove they are human.
zer0t3ch 9y ago
Isn't it supposed to learn from you? IE, answering technically correctly but what it thinks is incorrect is slightly annoying for you, but better for the system in the long run. (ie better for everyone)
- moreless 9y ago
  Everyone? Or Google? I don't feel particularly happy about being used as a lab rat, so no, thank you - I don't care about the quality of Google's AI. If anything, I would purposefully mislead it if I knew how.
  - amrrs 9y ago
    Don't you think that's how a lot of training data can be generated efficient for future ML/AI breakthroughs?
    - kbart 9y ago
      So we will get even more "targeted ads" in our faces? No, thanks. I think ML/AI has a great potential (especially in medicine), but I just don't trust ad companies to use it for any good cause.
    - JshWright 9y ago
      Has Google published this training set somewhere? Until they do, you're absolutely right that this is a great way to build a training set, but I don't see how it's to anyone's benefit but Google's.
wott 9y ago
I have more problems with bridges than with cars. The damn thing forces you to select 3 bridges, except that there are only 2... So you are forced to select what it thinks is a bridge, and confirm his erroneous bias even more.
- Chaebixi 9y ago
  Storefronts are difficult too, I don't think I'm ever good enough at those to satisfy it. The most reasonable one seems to be street signs, but I think it fails me for not flagging the unpainted back of one.
- zhte415 9y ago
  Sounds like poor question asking. "Click any square with a bridge" would be better wording.
aswanson 9y ago
Twisted Turing test. There's a novel waiting to be written about this.
0xTJ 9y ago
Does it tell you that you're wrong, or simply give you another set. If it is just giving you another set, it may be that it thinks that you can provide more useful data.
- csallen 9y ago
  It just gives you another set, but it's so frustrating and annoying as a user that I can't believe they're doing this on purpose.
- Chaebixi 9y ago
  If you fail, it tells you you were wrong (some red text at the bottom) and gives you another challenge. I think it will sometimes just give you another challenge for more data, but it won't have red failure text.
robryan 9y ago
I had to do 20 or more of these to get some post tracking data recently. Found the same thing with most objects where a very small part of an object hadn’t been classified as containing that object.
bildung 9y ago
It was the same when this was about words from old books. I always had to fill in letters the average person would have thought it to be, not what it actually was (e.g. the letter "f" for what really was an "s" in gothic type). Nowadays it's much easier, you can click anything that looks vaguely the same (e.g. boxy things for cars, ads for traffic signs, traffic signs for store fronts etc.). The fact that it's so easy to poison the training set makes me very wary about the autonomous car future...
- MisterTea 9y ago
  I actually like poisoning them. Not to be malicious but I feel manipulated into training their software for free. "Oh you wanted to sign up for that web forum? Sorry, but you have to do some free work for us first" And if you think that it's somehow good because it's mutually beneficial to train AI to better the future of humanity, don't. That is what their marketing department wants you to think.
  - dazmax 9y ago
    The value they're providing is to the forum owner, who has reduced spam-handling workload. So the forum is providing you value, you are providing Google value, and Google is providing the forum value.
    - kuschku 9y ago
      Sure, but if the public has to provide Google with free data, we should make laws requiring Google to open source the entire ReCaptcha training set.
      - operon 9y ago
        Why?
        kuschku 9y ago
        Anything created with free work should be free in return. Anything created by the public should be available for the public.
thom 9y ago
I went through exactly this today. Wasn't sure if it was the computer being dumb, or other people missing corners of signs or bridges etc. Still, takes me 5 goes every time.
nerdponx 9y ago
Street signs are a problem as well. Is the post part of the sign or not?
dboreham 9y ago
I find the same thing with the road signs : other people are giving (imho) the wrong answer by not including squares that cover a small, but non-zero part of the sign.
Chaebixi 9y ago
> This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is. Exactly. I think Recaptcha was better when it was looking for consistency with other human answers. Using "AI" has the same problem you mentioned, plus its more vulnerable because it has the assumption that your "AI" is unapproachably far ahead of competitors.

js8 9y ago

That was my experience with car and sign captchas as well.. But interestingly, it also depends on my mood, when I feel lazy, I click fewer boxes.

pbhjpbhj 9y ago

The problem with the cars for me is their definition -- is that van classed as a car, how about the half-back, what about a 4x4, what about a 4x4 with no back windows, ...

indigochill 9y ago
What about a square that contains just a sliver of the car from the next box over? Does that still count? I hypothesize that my attempt to classify every pixel related to a car as a "car" may contribute to my failing of these tests.
- toyg 9y ago
  This is what kurtisc is saying. The algorithm is unable to classify those pixels, so you have to guess which bits of the car can be detected by the algorithm and then select only those. So you have machines asking humans to think like a machine in order to prove they are human.

zer0t3ch 9y ago

Isn't it supposed to learn from you? IE, answering technically correctly but what it thinks is incorrect is slightly annoying for you, but better for the system in the long run. (ie better for everyone)

moreless 9y ago
Everyone? Or Google? I don't feel particularly happy about being used as a lab rat, so no, thank you - I don't care about the quality of Google's AI. If anything, I would purposefully mislead it if I knew how.
- amrrs 9y ago
  Don't you think that's how a lot of training data can be generated efficient for future ML/AI breakthroughs?
  - kbart 9y ago
    So we will get even more "targeted ads" in our faces? No, thanks. I think ML/AI has a great potential (especially in medicine), but I just don't trust ad companies to use it for any good cause.
  - JshWright 9y ago
    Has Google published this training set somewhere? Until they do, you're absolutely right that this is a great way to build a training set, but I don't see how it's to anyone's benefit but Google's.

wott 9y ago

I have more problems with bridges than with cars. The damn thing forces you to select 3 bridges, except that there are only 2... So you are forced to select what it thinks is a bridge, and confirm his erroneous bias even more.

Chaebixi 9y ago
Storefronts are difficult too, I don't think I'm ever good enough at those to satisfy it. The most reasonable one seems to be street signs, but I think it fails me for not flagging the unpainted back of one.
zhte415 9y ago
Sounds like poor question asking. "Click any square with a bridge" would be better wording.

aswanson 9y ago

Twisted Turing test. There's a novel waiting to be written about this.

0xTJ 9y ago

Does it tell you that you're wrong, or simply give you another set. If it is just giving you another set, it may be that it thinks that you can provide more useful data.

csallen 9y ago
It just gives you another set, but it's so frustrating and annoying as a user that I can't believe they're doing this on purpose.
Chaebixi 9y ago
If you fail, it tells you you were wrong (some red text at the bottom) and gives you another challenge. I think it will sometimes just give you another challenge for more data, but it won't have red failure text.

robryan 9y ago

I had to do 20 or more of these to get some post tracking data recently. Found the same thing with most objects where a very small part of an object hadn’t been classified as containing that object.

bildung 9y ago

It was the same when this was about words from old books. I always had to fill in letters the average person would have thought it to be, not what it actually was (e.g. the letter "f" for what really was an "s" in gothic type). Nowadays it's much easier, you can click anything that looks vaguely the same (e.g. boxy things for cars, ads for traffic signs, traffic signs for store fronts etc.). The fact that it's so easy to poison the training set makes me very wary about the autonomous car future...

MisterTea 9y ago
I actually like poisoning them. Not to be malicious but I feel manipulated into training their software for free. "Oh you wanted to sign up for that web forum? Sorry, but you have to do some free work for us first" And if you think that it's somehow good because it's mutually beneficial to train AI to better the future of humanity, don't. That is what their marketing department wants you to think.
- dazmax 9y ago
  The value they're providing is to the forum owner, who has reduced spam-handling workload. So the forum is providing you value, you are providing Google value, and Google is providing the forum value.
  - kuschku 9y ago
    Sure, but if the public has to provide Google with free data, we should make laws requiring Google to open source the entire ReCaptcha training set.
    - operon 9y ago
      Why?
      - kuschku 9y ago
        Anything created with free work should be free in return. Anything created by the public should be available for the public.

thom 9y ago

I went through exactly this today. Wasn't sure if it was the computer being dumb, or other people missing corners of signs or bridges etc. Still, takes me 5 goes every time.

nerdponx 9y ago

Street signs are a problem as well. Is the post part of the sign or not?

dboreham 9y ago

I find the same thing with the road signs : other people are giving (imho) the wrong answer by not including squares that cover a small, but non-zero part of the sign.

Chaebixi 9y ago

> This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is. Exactly. I think Recaptcha was better when it was looking for consistency with other human answers. Using "AI" has the same problem you mentioned, plus its more vulnerable because it has the assumption that your "AI" is unapproachably far ahead of competitors.