AI and language learning

Rob asks:

I would love to see you write on how AI changes the value of language learning, at the margin.

Two thoughts:

The value of learning languages in order to “get” a culture more deeply probably doesn’t change much.

On the other hand, international students probably have much less incentive to learn English — they are mostly interacting with other non-native speakers anyway, there is little cultural knowledge to be gleaned from it.

What else?

(I’m thinking of this because I just moved to Paris with my French wife . . . and thinking about the value of improving my French)

Let’s say that soon we have universal translators as good as on Star Trek, except the words are spoken first in the foreign language and then repeated/translated by the AI (i.e., the AI can’t just make “Romulans speak perfect TV English”).  There is then one obvious reason not to learn foreign languages, especially for short interactions and travel interactions.  For short interactions, the doubling time for the verbal exchange isn’t a big deal.  So instead of learning the foreign language, the AI does the work for you.

But under which conditions does this not hold?

Alternatively, let us say you wanted to be “matched” to the most appropriate person in the entire world on some topic.  It could be a marriage match, a friendship match, or simply a “I need to spend many hours discussing Tibetan traditional music with this person” match.  The doubling time for the exchange now is a big deal.  “Yes we are married, but it is not so bad that everything is repeated by the AI — it limits the amount of time we have to talk to each other and it also slows down our arguments” –no, that just doesn’t cut it, perhaps only for Tyrone.

These ideal matches still will provide a robust reason to learn foreign languages, at least provided you know the language you wish to match into, or perhaps you simply wish to match into something.  If the people you wish to match to speak decent English, however, that is again a reason not to learn.

You also might value the “perspectival” benefits of a foreign language more in a world with AI.  Perhaps the humanities and multi-perspectival approaches will rise in value, given the ability of the AI to execute very well on technical tasks.  Under those assumptions, AI increases the returns to learning the appropriate foreign languages.

Plus AI itself will make languages easier to learn — imagine having a companion to practice with, at any level of desired difficulty, all the time.

Since the short, travel-based interactions were never the main reason to learn foreign languages anyway, perhaps the demand to learn other languages will prove more robust than many are expecting?


Comments for this post are closed