Jun 29, 2026, 6:40 PM@evacide Concerning. I don’t want to attribute maliciousness to their actions, but if most of the people they’re quoting don’t exist, it sounds like more than a simple issue such as the AI being out-of-date based on when it was trained. It’s either producing so much content a few fake people are sneaking in, or a few articles with clearly misleading information. Both are bad.
Jun 29, 2026, 6:53 PMJun 29, 2026, 7:06 PM@wcbdata @evacide Attributing “guessing” to AI is once again imbuing AI with human-like qualities you’re trying to say it doesn’t have. It doesn’t guess, and you’re right that it doesn’t know. The people behind it however, do know. You pull inaccurate data offline if you care, leave it if you don’t, or tell the AI to be malicious if you want and it will do as it’s told, much of the time. So what I was saying is, they either don’t care because it’s unattended, or they do and it’s intentional.
Jun 29, 2026, 7:15 PM@david @evacide No, it guesses. It uses a statistical model to come up with the closest thing it can to what could look like a response to a question. That's a guess. The people training the models have ONLY THE VAGUEST IDEA what the models are trained on because the volume of text is so large and the statistical linking is not declarative. So no, they cannot reliably control the guessing. And they do it badly. And yes, the people training and using the models know that, and they do not care.