• Nix@merv.news
    link
    fedilink
    English
    arrow-up
    86
    ·
    1 year ago

    Wow that’s awesome hopefully they open source it and make it easy for anyone to use

      • AliLunaCat@lemmy.world
        link
        fedilink
        English
        arrow-up
        64
        ·
        1 year ago

        Google’s recaptcha is used to help train image-recognition programs (like all those ones having you identify signs/bicycles/stoplights help Google’s self-driving cars)
        They used to have the 2 words, where one word was the control and one word was from a physical book that Google was trying to digitalize

        • WhyIDie@kbin.social
          link
          fedilink
          arrow-up
          15
          ·
          1 year ago

          same deal with the ones with matching the orientation of something with a given direction being pointed at; it’s training 2d-to-3d AI image generators

        • antlion@lemmy.dbzer0.com
          link
          fedilink
          English
          arrow-up
          7
          ·
          1 year ago

          How does that make it acceptable? I’ve had to do like 6 minutes of free labor in the past 10 years. Google owes me like $6-10 for that bullshit.

          • DreadPotato@sopuli.xyz
            link
            fedilink
            arrow-up
            8
            ·
            1 year ago

            Since i started getting in to online privacy, I’ve been stuck in captcha-purgatory a lot…it just keeps loading a never ending stream of new captchas for me to solve.

            • Ryantific_theory@lemmy.world
              link
              fedilink
              arrow-up
              2
              ·
              1 year ago

              Yeah, there was a chunk of time where there was some spat going on between them and VPN providers, and whenever I would wind up on an exposed VPN server every single captcha would take over a minute of clicking through different prompts. Happened so frequently for the fastest server that I just switched to Firefox and DuckDuckGo because I couldn’t stand getting hit every single time I googled something. Not just every session, but literally every single search.

              The worst was when it would test me for several minutes straight, and then have the gall to tell me to start over again. Google’s really been racing to the bottom lately.

        • Mr_Blott@feddit.uk
          link
          fedilink
          arrow-up
          3
          ·
          1 year ago

          Yeah I noticed now that Google maps has little icons for traffic light positions. Years of our work and AI learning for…fuck all. I CAN SEE THE FUCKIN LIGHTS IRL

      • edric@lemm.ee
        link
        fedilink
        arrow-up
        14
        ·
        1 year ago

        Does lemmy signup use reCAPTCHA? Because I’m starting to think I’m a robot when trying to sign up new accounts. lol. I’ve never had such a hard time getting it right on the first time. Or maybe my eyesight is just getting bad.

        • crab@lemm.ee
          link
          fedilink
          arrow-up
          3
          ·
          1 year ago

          I always use the audio for Lemmy captcha because some of the letters are ambiguous

      • Solumbran@lemmy.world
        link
        fedilink
        arrow-up
        25
        ·
        1 year ago

        Encryption is generally “open source” and that’s what makes is strong. Security does not come from people not knowing how things work, but by having properly designed things that work whether people know how they work or not.

        • poke@sh.itjust.works
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          It doesn’t seem to me like encryption is comparable here. With encryption we have known algorithms that are harder to reverse than initially run. This is a completely different problem, where many inputs are taken and some algorithm has to decide if they are human or not. What digital task can a human do that a robot can’t in the same way, especially if the robot knows exactly the measures it should aim for?

          • BOB_DROP_TABLES@lemmy.ml
            link
            fedilink
            arrow-up
            1
            ·
            1 year ago

            exactly what recaptcha does, for example. Knowing that you have to type a word because a computer failed to identify which word is it makes creating a program that does that no easier. Same with the image ones. While criptography is a different problem, the argument is the same: you want something that can be verified to be hard to break otherwise someone will eventually figure it out

            • poke@sh.itjust.works
              link
              fedilink
              arrow-up
              1
              ·
              1 year ago

              If you have a known algorithm for generating those hard-to-read images, then it really wouldn’t be that difficult to generate a large enough set yourself to train a custom ML model to solve them. The same would apply to audio challenges.

              Only one person would need to do it then they could share the process, potentially automating others being able to bypass as well.

              I like the idea of captcha being open, but unlike encryption as far as I know we don’t have a starting point on something that is actually easier for humans when all information is available. Until something like that exists, open sourcing to implement and improve it doesn’t make sense if you want an effective product.

              • BOB_DROP_TABLES@lemmy.ml
                link
                fedilink
                arrow-up
                1
                ·
                1 year ago

                The text is not generated. It’s from photos of books that failed ocr. The photos are then distorted to make it even harder in order to become that captcha. 2 words are used 1 is a control (to know if the response is correct), the other is one they what to know what says (to add to the pool of words and finish digitizing the book).

        • poke@sh.itjust.works
          link
          fedilink
          arrow-up
          1
          arrow-down
          1
          ·
          1 year ago

          This isn’t a problem of security, this is a problem of deciphering between human and non human users.