https://blacktwitter.io/@bibliotecaria/109650353375080864 I would expect people like librarians and teachers to have a lot more stories like this in the near future. Text generators have been marketed in an incredibly irresponsible way and laypeople don't understand the most fundamental things about them: that they don't actually *understand* anything, have no concept of truth, are just as willing to tell you something completely fabricated, and increasing the % of truth isn't a simple matter of more compute or training data.
And that's by design! Because companies like OpenAI and their investors stand to make a lot more money if people treat these things like The All-Seeing Oracle rather than "that one hyperconfident white dude acquaintance who claims to have read everything ever written, whose answers on anything important you always need to fact check".
Actually Open AI (the parent company) is a non profit, with the mission to prevent the scenario you (rightly!) fear. At least the development is open source and transparent. Just imagine a for profit was sitting on this code......
