48 minutes ago · Tech · hide · 0 comments

Social media and the Blogosphere (what’s left of it) and many other media publishing some individual’s view is full of people who claim that “AI” (meaning Large Language Models) only produces what they call “slop” that is full of errors. In programming they claim that it only reproduces code that was part of its training data and if it does not find anything matching it will just produce some hallucination that is mostly wrong. Also the code is supposed to be inefficient. And of course that it does not work at all for niche programming languages (which alas Delphi has become). I have stated before, that they are wrong but I am getting tired of this. But before I simply give up, here is an example how I just found and fixed a problem in a proprietary tool. Today I got an email from a user that contained only one sentence: “Anscheinend kann der DIR-Checker keine Messungen mit R erkennen bzw. checken.” (English: Apparently the DIR checker tool cannot detect or check measurements with R.)…

No comments yet. Log in to reply on the Fediverse. Comments will appear here.