My wife is currently doing some interesting research in English language into medical malpractice in the US etc. by interrogating Deepseek with lot's medical records and so on.
I have seen some of Deepseek's responses/output to her inquiries/prompts. She even uses a long-term memory feature of Deepseek so it remembers past interactions and conversations.
I must say, how Deepseek repeatedly and intensely tries to gain trust, confidence and sympathy with the user is quite disturbing to say the least.
Deepseek also tries to encourage the user to take some legal action of questionable relevance or legitimacy.
No comments:
Post a Comment