The Technium

Arrival of the Babel Fish


In the very near future, maybe in ten years, we’ll have earpods that will do real time language translation. Someone speaks Greek to you, and with the slightest delay, you’ll hear English. You respond in English, they’ll hear Greek. It’ll work for most spoken languages, x to x. You might recognize this as the Babel fish in Douglas Adams’ fiction, but this one will be real. We are not far from it today. I’ve been using Google Translate on my phone when traveling in China. I can speak or write English through it, or listen or read Chinese from it. It’s about 80-90% accurate, which is good enough to speak with taxi cab drivers, or navigate as a tourist. I have also been using a couple of different AI translation services, such as Trint, to create a text transcript from podcasts. It listens to the podcast audio file and puts the words into text with about 95% accuracy. It does this in minutes and for a few dollars.

When even more accurate machine translation becomes available in ever more handy forms — like earbuds, or embedded into smart glasses — I can imagine huge economic changes arising from this technology. The first thing it will do is to enable people around the world who have very desirable skills, except the skill of English, to participate in the global economy. This Babel fish would permit a talented programmer in Jakarta who spoke no English to work for a Google. It would allow a talented programmer in Utah to work for a Chinese company, in Chinese. Nor does the translation have to happen online. Two employees in the same room could each be wearing the Babel fish. Of course it is immensely effective combined with virtual telepresence. When a colleague is teleporting in from a remote place to appear virtually, it is relatively easy to translate what they are saying in real time because all that information is being captured anyway. For even greater verisimilitude, their mouth movement can be reconfigured to match what they are saying in translation so it really feels they are speaking your language. It might be even be use to overcome heavy accents in the same language. Going further, the same technology could simply translate your voice into one that was a different gender, or more musical, or improved in some way. It would be your “best” voice. Some relationships might prefer to meet this way all the time because the ease of communication was greater than in real life.

This unleashing and liquidity of talent would be a huge boost to the global economy and would help in leveling some of the inequality between wages around the world.

There would be other effects: films, music, videos, books would not need to be laboriously and expensively translated beforehand, or to reach some level of popularity before getting dubbed. Now with the Babel fish they would be instantly subtitled, dubbed, translated in real time, on demand. Over time, even regional differences (American vs Australian) could be accounted for. This universal translation-on-demand (UTOD) immediately increases the potential audience size for creative works, increasing the probability that obscure interests can find the thousand true fans around the world it’ll need to be sustainable.

I can also imagine this UTOD technology aiding migration and human mobility.  When the global population plunges later this century, mega-cities around the world will begin to compete for workers and citizens; without the added hurdle of having to speak a new language will make it much easier to migrate. Many might move to Tokyo if they could virtually speak Japanese fluently.

UTOD might diminish the dominance of English as a second language. Why bother with it? On the other hand it is very possible that having simultaneous translation whispered into your ear all day for years would, over time, with the right attention, act as a teacher and help a person learn another language. Or the program could be modified to accelerate such learning if someone desired.

Today I can use Google Translate for free, just like other Google products. Ideally there would be a free version of Babel fish so that those to whom this would most make a difference would have full access to it. But we know free has its own costs. There will be pressure to insert advertising into UTOD. One could imagine how annoying it would be to be conversing with someone when every now and then you are interrupted with an ad that you both hear in your language. Worse, the ad could be related to what you were talking about, since the machine would “know” exactly what you are talking about in order to translate it. Other biz models would not interrupt you in conversation, but would try to exploit that very specific data in other modes or parts of your life. The poor and desperate are likely to take that bargain, but their data is less valuable (being poor and desperate). Alternatively, there would be a paid (no ad, no track) version.

UTOD, encased in a wearable like a Babel fish, is almost here. If adopted widely its consequences would be enormous, and I think, sudden. Even though it has been gradually improving, it might come as a huge “overnight” surprise to the world.




Comments