Some years back, Carnegie Mellon professor Luis von Ahn helped invent ‘Captcha,’ that equally ingenious and annoying reverse-Turing test that prevents spammy machine algorithms from infiltrating services meant for humans on the web. What you probably didn’t know about it is that every time you punch in one of those squiggly, barely-legible words, you’re helping a computer digitize a book. Accounting for all the various sites using Captcha, that’s about 2.5 million books per year. The only problem is that those 10 seconds of time you spend typing Captchas still offer no immediate benefit. That’s going to change with von Ahn’s next project, which is using similar technology in a way that teaches you another language — and helps translate the entire internet in the process.
The project, Duolingo, which is currently in a closed beta, is addressing an increasingly important problem with online crowdsourcing (people doing for free what they could be paid to do) by tackling another issue (the high costs of learning a new language). Meanwhile, the entire web is being translated into every major language, a task which is currently far too complex for computers and far too expensive for humans.
How? By giving users simple sentences taken directly from the source material that needs to be translated. It sounds crazy, but it actually works — von Ahn shows that the results they’re getting from this method are comparable to those from paid professional translators.
The downside is that it threatens to put some members of that vocation out of work. But even considering the dragging economy, perhaps having more bilingual people on the planet — in particular, those who can’t afford to spend $500 on language learning software — can help individuals increase their own worth, and bring us all closer together. That, at least, is something that no machine can be programmed to do.