Early on, the Duolingo stated goals was to teach language to the point where it's learners could ultimately start translating documents. They were going to sell cheap mechanical turk style translation services. (Think captcha style translation)
Unfortunately as they got popular automated translation services got good enough that nobody was going to pay for a slightly better and slower translation enmass.
Once that happened, that's when it seemed like they dumped their goals of teaching language and instead focused on dark pattern money extraction.
I started using DL in 2019 and it was then still very much a respectable language-learning app for semi-casual users up to CEFR A1/A2/B1/B2 standard, with good user-contributed community forums (they killed those 3/2022, so they could milk them with AI to resell as an add-on - for me that act was the jump-the-shark telltale). Don't know about the goal of translating documents but that's something they'd want to pivot away from to get beyond a tiny niche audience. Some gamification and user evangeliation is good and necessary (users should be able to turn it off). If nothing else, DL's antipattern whenever they eventually crash will show limits on how far it can be pushed in the name of $$$ growth hacking. Meanwhile, bona-fide langage learners are jumping ship off DL. (Analogously, what is the point when Zynga/FB games Candy Crush/FarmVille/MafiaWars stopped being cool? or Pokemon Go? Ingress?) How much of the stickiness is from the thing being inherently hooked, vs the social competition agaonst your friends?
Unfortunately as they got popular automated translation services got good enough that nobody was going to pay for a slightly better and slower translation enmass.
Once that happened, that's when it seemed like they dumped their goals of teaching language and instead focused on dark pattern money extraction.