To be frank, their curriculum (IME) is so bad already that even if implementing more AI could only maintain continuity of quality (it can't, it'll be worse), they're still a doomed company. AI might help them make more languages available; or it could help them design deeper & longer courses. But neither of those things remotely capture Duolingo's problem; their problem is that a ton of their courses suck, from day 1. They teach the wrong things, in the wrong way, there's no logic in the order of why words and concepts are taught in the way they are, there's no attempt at understanding why someone is learning the language and catering courses to their needs, they constantly teach incorrect translations or use really bad AI voice-overs which don't capture native accents, and they don't actually correct problems when they're reported.
Duolingo is, in my mind, the categorical example of a tech company. Think about what it means to say "we're an AI first company". Who is the audience of a statement like that? Who reads that and thinks "heck yeah"? Its not customers; customers, at best, don't care, but at worst have had so many negative experiences with AI that it reflects very poorly. Its not employees; Luis von Ahn is digging a grave while saying "no we're not going to kill you, we're just killing the contractors". The answer is: Its a statement for Duolingo's real customers, not their users, but the US Financial Markets.
Duolingo is, in my mind, the categorical example of a tech company. Think about what it means to say "we're an AI first company". Who is the audience of a statement like that? Who reads that and thinks "heck yeah"? Its not customers; customers, at best, don't care, but at worst have had so many negative experiences with AI that it reflects very poorly. Its not employees; Luis von Ahn is digging a grave while saying "no we're not going to kill you, we're just killing the contractors". The answer is: Its a statement for Duolingo's real customers, not their users, but the US Financial Markets.