“The only way you’ll have success in the stock market is if you have a variant perception, something different from the masses.” – Ted Weschler

On May 13th, OpenAI demo’d real-time translation capabilities of GPT-4o. You can watch the video using this link.

Needless to say, it is quite remarkable what OpenAI has been able to develop with the technology they have at hand. I would much rather talk about the reaction to the announcement. Upon announcement, shares in Duolingo were down ~3.8%, with a noticeable decline at 1:00 PM EST (time of announcement). Obviously correlation is not causation but let’s assume so for illustrative purposes. Let’s imagine the entire decline in Duolingo’s market value for the day can be attributed to the OpenAI announcement. In effect, OpenAI’s real time translator would’ve erased ~$250 in Duolingo’s market value. But does that make sense?

I have little knowledge of the economics of Duolingo. Of what I do know, the business has 90 million MAUs of which 8% subscribe for additional quirks. The primary motivation of Duolingo subscribers is to learn the language. Surprising, right?! So why would the market have decided to reprice the stock? The most common opinions I’ve seen is that GPT’s real-time translator wipes out the need to learn languages. I believe this perception still misses the mark on why people use Duolingo and why those who pay for it do.

There has been live translation on Google for years (albeit, not as seamless as the demo). So while impressive, it is not novel. I hold the opinion, the majority of Duolingo subscribers pay because of a necessity (English is the most learned language) or intrigue. Live translation does not reduce necessity nor does it impact intrigue. If you must learn a second language, GPT’s live translation will only enhance not disrupt the need. I suspect this would be due to relocating to a foreign country. Real-time translation would undoubtedly help, but, learning the language would most definitely occur in concert.

Additionally, should you become fascinated with a new culture/language, you’ll learn as you wish. These consumers are either driven by trends or general interest. To satisfy the thirst of these consumers, real-time translation isn’t enough. They are more likely to hope for immersion into the culture of the language and probably willing to enter foreign countries with the aim of understanding the language better. So the Duolingo value proposition is left unchanged for them regardless of GPT-4o capabilities.

In my opinion, real-time translation isn’t really in the mind of these consumer groups, as it is does not serve their ambition. These groups of consumers will only be satisfied once assimilated – the constant need to have an AI speak for you is not assimilation. If it does not assimilate, it does not threaten Duolingo either in my view.

I haven’t began to factor in other reasons users subscribe to Duolingo either(e.g., the games, test administration, etc.). The thing with all great public equity investing is that to generate above-average returns you must be willing to take idiosyncratic positions. If we all follow the crowd, then you’ll earn the returns of the crowd – and that’s just not what we’re all here to do. So, despite the wisdom of the crowd, I heed to Benjamin Franklin’s advice when he said: “If we all think alike, then no one must be thinking.”

“Having opinions about facts is what makes markets.” – Jim Chanos

*Just to reiterate, no one can for certain know why Duolingo shares were down on the day or even in the past month (if you zoomed out further). I did, however, think this would make an excellent example of variant perceptions.*