Saturday, November 30All That Matters

Why can’t robots check the box that says ‘I’m not a robot’?


Why can’t robots check the box that says ‘I’m not a robot’?




View Reddit by fumifeiderView Source

13 Comments

  • Stolehtreb

    This is one of those things where there’s an arms race that just improves both sides. Creating a bot that mimics human behavior isn’t outrageously hard, and as long as you know what it assumes is human, bots can be patched to bypass. And gathering data on what actions spoof human behavior is also easier than it’s ever been with machine learning. If there’s ever actually an ai singularity in the future, google is the one that’s been training it as it’s enemy for decades.

  • phpworm

    In my experience with web dev, old school methods still work better than captchas. For example bots can’t distinguish the difference between required fields and optional ones, so they just try to complete everything. Dead giveaway. Captchas on the other hand not only require your visitors to jump through hoops making it less user friendly, but they fail frequently.

  • LilDutchy

    I had to do one recently that was slide the puzzle piece into place, except that when you started sliding the puzzle piece the image shifted slightly to the right but the place where it expected the puzzle piece didn’t move. Took me a few tries to figure it out.

  • Anom8675309

    Just google hu? just wait till Microsoft builds it into your PC’s OS.

    Really nice that we don’t have those capcha checks anymore right?

    How about an entire OS that deems whats worthy of you to find and use as software? How about your browsing history and surfing behavior being used during a court case as a character witness against yourself? Just a couple Patriot act subpoena away from pulling that data ‘for the protection of the greater good’.

    Spend 2 hours surfing a knife website? guess who’s on the top of the list of knife crimes in your area? Well knife shoppers aren’t a protected class, so why not just pay a global analytics cooperation for the info on potential knife crimes and simply not hire them for that job. Or any other factor that deems slightly problematic for corporations. Fat people, smokers, reddit subscribers… anyone who shows up to that political rally that might be a problem in the future for the company being associated with that potential employee. Somewhere some time you’ve violated your companies polices and siri/allexa/google whatever recorded it.

    All that data won’t be collected by just Google anymore, it will be on your computer and phone OS’s.

    These conspiracy theorists are always worried about a secret government organization ruling them all the while they’re simply ruled by their own laziness.

    Edit: no clue why people are dving.. this is happening right now, if not within a year or less. Even right now you can’t apply to a job without going through several AI filters. If you think people aren’t going to take the data you freely provide and market it, you haven’t been paying attention to the world.

  • neihuffda

    So a program can pass by first figuring out where the box is on the screen, then move the cursor in a random fashion over to the box to click it.

    For browsing history, you could probably just copy some files over to where ever the browser stores browsing history. Or, have the start of each run google some random shit.

  • awp_india

    I got into sneaker botting a few years ago. I got to the point of using these services. I’ve never felt so dirty and such a slime ball. On top of the bots, proxies, and servers I’m renting monthly. I’d pay a few bucks each “hot sneaker drop” to have some captcha slaves.

    Such a weird fucking service. Literal internet slaves

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.