Dystopian Reverse Captchas
You encounter this strange “captcha.” What do you do?
Here’s a creativity assignment for aspiring science-fiction writers. If you saw this captcha, how would you pass it?
In a future run by AI, you, the protagonist, would have to pass reverse Turing Tests to survive. If the dystopia depicted above arrived today, here’s how I’d pass it:
- I’d make a Chrome extension that automatically clicks in the same spot of the captcha every time it encounters one. I’d also program it to do so within a few milliseconds of the page loading. This behavior mimics the way web crawlers bypass login or registration pages today.
- I would also want to use a VPN or proxy to pass the new captcha. Again, this behavior mimics how botnets currently use separate IP addresses to accomplish a million different things. In our emulation, we’d want websites that have safeguards against denial-of-service attacks to pick us up.
- Lastly, I’d try to submit the form without hitting the captcha, possibly skipping the DOM (document object model) rendering before doing so. Robot crawlers can read the web without having to render a page’s DOM. Thus, the trick to browsing this machine-friendly web would be to read it with an alternative browser, like lynx, and attempt to brute-force your way through the form without loading much at all.
These three circumvention techniques tell you that to survive in a world run by robots you’ll have to be fast, thin and multiplied. Bots “machine-read” content, but only from feeds or synopses, and they can be in a million different places at once. A bot, for example, would interface with art galleries by crawling all the photos from Instagram taken at the art opening, cross-referenced with a social media graph and other metadata, to then form an opinion of the artwork. Likewise, to survive in a robot-run world, you will have to reduce the flourish of human expression to a limited and structured palette. We already train ourselves for this future when we clear our throats and enunciate in easy-to-parse phrases when speaking to Siri, Alexa, or Google. We’re finding that it’s easier to build AIs that gets us 80% of the way there while letting humans pick up the slack.