A decade or so ago, I pitched an investor on my language-learning business. He looked at me and asked: Why anyone would invest in that? Within ten years, everyone would have language chips in their heads and be able to speak any language.
Well, it's been ten years and I still don't have a language chip in my head. Perhaps more importantly, people haven't stopped learning languages.
They appear to be learning more of them. At the end of 2022, Duolingo had 60.7 million monthly active users. By the end of 2025, it had 133.1 million monthly active users and 12.2 million paid subscribers, compared with 4.2 million in 2022. While this doesn't give us the total size of the language-learning market, it certainly doesn't look like it's going anywhere.
It's funny because the technology that my investor envisioned is starting to become a reality. You can point your phone at something and translate it. You can speak into an app and have it speak another language back to you. Apple's new AirPods can translate a conversation while you're speaking to someone. If both parties have the right devices, they can hear the other person translated through their earbuds.
So why are people still learning languages?
Just because something can be done doesn't mean it will be
When people predict what technology will be able to do, they usually focus on capability. Can a computer translate a conversation? Can an AI write an essay? Can a robot drive a car?
Sure, those are questions that need to be answered, but they're not the only questions. Will people want to use the technology? Will it be easy to incorporate into their daily lives? Will it actually be better than what they currently do?
Talking to someone is a great example. If there is even a slight lag in a conversation, it can change the dynamic between two people. Talking to someone while looking down at your phone, waiting for a sentence to be translated and played back, is very different from simply talking to someone. You lose eye contact, humour can be lost, and it becomes difficult to cut in or respond quickly.
Earbuds solve some of these problems, but they bring up another issue with adopting technology. Sometimes the technology needs to be used by both parties. None of these problems is permanent. Translation will become quicker and devices will become smaller. Eventually, it may become second nature.
I remember using dial-up internet and hearing people predict that one day you would simply open your computer and the internet would already be there, working quietly in the background. At the time, that sounded almost futuristic. Instead, you had to dial in, listen to the strange static, wait for the connection and hope no one needed the phone line. If it dropped out, you started again. Now the prediction seems completely ordinary. Most of the time, we barely think about connecting at all.
I think translation will eventually become something similar. But we should learn from these examples that these things take time. Elon Musk has given us some great examples because he likes to put dates on his predictions. In 2016, he said SpaceX planned to send people towards Mars in 2024, arriving in 2025. Well, that didn't happen. He has also predicted that money may eventually cease to matter once AI and robotics become advanced enough.
Elon may be right about some of these directions. SpaceX has changed the way we think about space travel, and AI and robotics may change the way we think about work. It's just difficult to go from we are advancing this technology to this is what our society will look like in ten years. There are too many variables in between: infrastructure, laws, price, human behaviour and culture.
Sometimes we like to do things the hard way
Another problem with the idea that translation will make language learning obsolete is that it assumes we only learn languages to communicate information.
People learn languages for many reasons. They want to know more about a culture. They want to speak to their partner or family members in their native language. They like the challenge. They want to visit a place and feel like they can communicate with the people around them. Maybe they just like knowing that their brain can learn to think in another language.
Technology can eliminate the need to do something without eliminating the desire to do it.
Computers have been beating humans at chess for years. We still play chess. We don't play tennis because humans are the best machines for hitting a ball over a net. Someday a robot may be able to hit the ball faster, reach more balls and make fewer mistakes. People may watch robots play tennis, but they would be watching because they are interested in robots, not tennis.
We run when we can drive. We grow our own food when we can buy it at the supermarket. We go to live performances when we can listen to the same music on our phones. Being human isn't just about reaching a goal as quickly as possible. Sometimes the activity itself is the point.
What does this mean for AI?
Maybe something even stranger will happen. The more AI takes over tasks that once required effort from us, the more valuable some of those human activities may begin to feel.
There is already a broader cultural version of this happening outside AI. As technology makes things faster, easier and less physical, people sometimes rediscover older ways of doing things precisely because they are slower, more tangible and require participation.
Vinyl records are a good example. When music became digital, we could listen to it more cheaply and easily. Then streaming came along and put almost all music at our fingertips. You would think that an older format requiring large physical discs and a turntable would simply fade away.
It came back.
Part of the appeal seems to be precisely the physical nature of vinyl: owning the record, choosing one, looking at the artwork, placing the needle down and listening to an album rather than just pressing play on any song you want. Things that were once inconvenient about vinyl became some of the things people liked about it.
I think we may see something similar with AI. Maybe writing something yourself will become more appealing because you know an AI could have done it for you. Learning a language may become more appealing because you don't actually need to learn it to communicate. You may draw, do maths, try to remember things or navigate without AI because you want to use those parts of yourself.
As technology advances, we also seem to become more interested in the past. Whether it's cooking from old recipes, using film cameras, listening to vinyl records or doing things our grandparents did, we seem drawn to the idea that at some point people knew something we have now forgotten.
That may not actually be true. But the desire behind it is interesting. I think that as we advance technologically, many of us will want to stay connected to our past as humans.
Which means we have to shape our future.
We have to keep imagining what we want our lives to look like and then build that life. AI will only get better. The companies building it will continue trying to capture our time and money. They will make their products easier and more enjoyable to use. Sometimes it will be hard not to use them.
But we don't have to fight against AI to remain human. We can live with it, use it, challenge it and push back when we want to.
Sometimes pushing back may be as simple as buying a ticket to hear someone sing, even though the same song is already on your phone. It might mean joining a book club and talking about books written by humans, or talking to someone in a language you are still learning even though you have a translator in your pocket.
These are small choices, but I think this is how we will create our future as humans. AI will increasingly allow us to get where we want to go faster. But maybe we don't always want the fastest route.
Sometimes the meaning is in learning the words, turning the pages, hearing the musician breathe between notes, making the mistake, trying again.
Our future won't simply be created by AI. We still get to shape it.