I've been talking to the chatbot on ChatGPT... very valuable tool... but I have to remind it to do really trivial and assumed things like "apply changes over the last set" "should I reupload the file" "the file is actually named ...." "Are you sure about that?" these type of things. It's holding my hand, I'm holding its hand. It's like an absent-minded professor or C3PO. Gemini has had serious issues too. Look out for misinformation, imprecision, ensh$%^&fication and favoritism.
At least it can say "you're right!"
Also, Clause just wrote:
Use caution before running this prompt. Malicious conversation content could trick Claude into attempting harmful actions or sharing your data.
default text >> Hey Claude, can you recommend tools and apps that are relevant for me? First, figure out what I use every day. Then help me connect the tools I use, and point me to the Claude apps that fit what I do. Afterwards, give me ideas about what you can do with what you know about me.
Has anyone else seen that? That was the first prompt I ever saw, pre-filled.
-- Wow what a dangerous thing to ask. Fist to head. Who defines malicious? Also, my credit card straight up wouldn't work with Claude. Never had that before. Maybe this is all a good thing... the Chinese can pick up the slack.
Imagine a big truck blowing its front tyre at 60mph in the middle of nowhere. Now think of your business. Or one day your extended family decides to hold you hostage at gunpoint and holds a ransom on you. Or Order 66 in Revenge of the Sith. Just be alert, not alarmed. AI insurance is where it's at.
nabukodonozor•47m ago