A lot of the infrastructure of capitalism is oriented around generating and sustaining small- and medium-sized businesses, and while large corporations will always want to increase market power most sectors have antitrust reasons to avoid outright monopoly. The tech sector feels free to aim for monopoly and generate monopoly rents.
The whole business model of Silicon Valley is finding companies that can grow to become abusive monopolies, and then growing them into that. Cut off one head and two will take its place.
People aren't mad because Facebook runs a social network or presents users with a newsfeed or provides a messaging service or stores users photos. Lots of companies do those things without complaint. People are mad at Facebook because the company constantly lies about privacy issues, makes one agreement with its users about their privacy and then does something else, and just generally treats the privacy of its users with complete contempt.
Language learning courses and tools are just ways to kickstart normal language acquisition, so you don't have to go through the years of gurgling like a baby has to. Learning to communicate in a new language isn't exactly like learning some other kind of skill--it's much, much huger.
When you're speaking your native language, you're decoding and processing complex social signals in real time at multiple levels of structure (phonological, morphological, syntactic, semantic, pragmatic). Even the best AIs can only get a couple of these levels of structure, with significant error rates, and only in a handful of well-studied languages (there are hundreds or thousands of natural languages). How much time does it take for you to learn 1000 vocabulary words? A natural language will have tens of thousands. A comprehensive grammar of the English language would be dozens of heady volumes and have less information about English grammar than any competent native speaker.
Learning a language is just an astonishingly huge task. No course or system can cover it all. Mostly they're just trying to make things easier for you.
Oh, absolutely. But I think they're going to be more specialized tools to work on different skills.
For example, I used to use the perapera-kun plugin and I guess yomichan does a similar thing, but that sort of tool—but say, for closed captions—could help listeners get more comprehensible input from movies. Like, here's one I have to try. https://chrome.google.com/webstore/detail/language-learning-...
Or just apps or social networks that help connect speakers in other languages. I know of a Discord server, that lets me practice a target language; the group just meets at an inconvenient time for me.
The problem is that the tools you need are "listening to and interacting with people in the target language". Human language is a problem of interpreting the creative social signals of other human beings, and there's no way to automate that. With the Web, YouTube and internet radio, the actual amount of target-language material that a person can expose themselves to is hugely, hugely more available than it was 20-30 years ago.
You have to keep in mind what a tool is trying to accomplish relative to the personal development it takes to grow into another human language. Different levels of learners require different kinds of instruction. Bilingual instruction is quickest if a person has zero knowledge of the language whatever. Absolute beginners need direct instruction in phonology. An Anki deck is to build basic vocabulary or literacy for a person who has some. Duolingo is there to get you comfortable with basic grammar or phrases, to get you interacting with native speakers in the first place once you stumble off the plane. A more advanced speaker might need accent reduction work.
Anki is fantastic at flashcards. Duolingo is fantastic at translation exercises, way better than what we had a few decades ago. Definitely the author in the OP should be using Anki rather than postcards on her bed.
"Mastery"-level linguistic tasks include things like writing creative poetry in the language that another person finds moving, describing the movements of a complex machine, or composing an essay in a specific literary style—things that native speakers may find difficult without formal training. There's just no way Duolingo can do that.
I took a survey recently that clarified a lot of this for me. The questions were all things like: What percentage of your time do you spend listening to music in the target language (and your native language)? What percentage of your time do you spend watching moveis in the target language (and your native language)? What percentage of your time do you spend reading books in the target language (and your native language)? What percentage of your time do you spend thinking and interacting with other people and thinking in the target language (and your native language)? If I spend 1% of my time or less in the target language and 99% in my native language, it's no wonder that I have plateaued in it.