[@rysiek](<https://mstdn.social/@rysiek):> once on my blog I have had [wrote a piece (in Polish, though)](<https://blog.oliwierjaszczyszyn.com/menstruacja-wypowiedzenia-czyli-jak-technologia-nie-rozumie-kontekstow>) about how technology in general does not understand any contexts unless we direct them to do so (for non-Polish speakers, I’ve translated to English the ending of my post):

> First of all, the speaker meant a language model, and not the artificial intelligence at all. In my opinion the artificial intelligence goes way too far beyond some huge word set. Although the responses coming from it may sound natural, often these present false information (again, this is just a word set made upon automated Internet browsing and downloading all necessary data to build such). \[…\] Opponents of LLMs raising concerns about “creating bias towards any side” seem to be right at least in regards of the quoted response about euthanasia “being the sole good solution in regards to the \[…\] father, and rest of family”. What if the person in question \[the LLM asker\] is against the euthanasia and want to resolve the problem not necessarily by terminating the father’s life despite his old age?

[@travisfw](<https://fosstodon.org/@travisfw>)