@Eric
Probably, we need more experiments / tests and stats on how much “Google Translate” has improved over the last few years. Otherwise, it´s simply “opinion vs. opinion” or “anecdotal evidence vs anecdotal evidence”. And that´s not a good basis for a discussion about the quality of AI translation tools.
And as you wrote it might depend on the language. For example, when I last used Google Translate for Latin texts in LingQ, the translations from Latin into English, German, etc. sentences were almost complete nonsense. But, again, that´s anecdotal evidence.
it’s certainly possible to access these tools from outside of LingQ if one didn’t have the resources to pay for LingQ.
I agree. But, I´m not so much defending LingQ here, but rather the “content-flexible audio reader” tool category itself. If someone is tech-savvy, but doesn´t have the money for a monthly / yearly subscription on LingQ, I´d say: “You should check out LWT. It could be an interesting option for you”.
From my perspective, the main advantage of an Audio Reader à la LingQ is that it can be combined with other tools, platforms, and services. That is: AI translation tools, dictionaries, SRS (the internal SRS or Anki), Youtube videos, Kindle or other e-book formats, all kinds of texts (articles from news sites, blog posts, academic papers, texts from writing forums, etc.) audio files (from Audible & Co). and Netflix videos. In this sense, LingQ is a kind of hub in a zoo of devices/gadgets and media products.
@Rabenwald:
Going old school, we could simulate this with printed (bilingual) books, audio files (on a Walkman, if MP3 players are already too high-tech), bilingual or monolingual dictionaries, and a Leitner cards system. But this is neither cheaper nor more efficient. Often this old-school combination is / was just “frustrating” because of the media gaps (“Medienbrüche”).
This isn´t a “tech revolution”, but simply the next evolutionary step in language learning, like the shift from processing paper-based invoices to processing electronic invoices (diirectly or by means of machine learning tools).