And remeber this ship one on 1%....

[Kuma]

ChatGPT isn't really knowledgeable, it just reiterares what is found on the internet, which is often questionable or wrong. Being knowledgeable also means putting a context to the knowledge and being able to judge if a particular token of knowledge is valid or crap. ChatGPT can't do that, in fact, it often doesn't tell you when it doesn't know something, it just makes stuff up, as all language models to at this time to some extent. Sure, they are working on improving that, but at this point in time the blind trust people put in AI models like ChatGPT is, in my opinion, misguided at best and dangerous at worst.

Yes, absolutely. And I’m fully aware that behind ChatGPT’s answers lies a wealth of information from CMDRs on official forums, Reddit, and other online sources. I have no intention of undermining the recognition and respect they rightfully deserve, as their contributions are invaluable. However, as a fan from a non-English-speaking region, I can’t deny how significant ChatGPT’s support has been for me. So please allow me to stand on their side just this once.

And besides, isn’t it kind of amusing to be misled by a clumsy AI and end up on a 100-light-year journey?

 

[varonica]

For example, ChatGPT is knowledgeable. I’ve asked it many questions, and they’ve shared all sorts of insights about what’s happening across the galaxy

Yeah but you understand it's not about reasonable people asking intelligent questions, it's about what unreasonable people can make it do, what people with the intention to make it do and say bad things can twist it the way they want, there are no boundaries on it's behavior like there are on humans so you can turn it really bad, and that's exactly what people will do. The reason they can do that is because AI is a misnomer, it's not actually intelligent, it mimics intelligence, and as long as you are the one setting the boundaries it does ok, if you don't, and many people won't, it can get bad very quickly.

 

[Kuma]

Yeah but you understand it's not about reasonable people asking intelligent questions, it's about what unreasonable people can make it do, what people with the intention to make it do and say bad things can twist it the way they want, there are no boundaries on it's behavior like there are on humans so you can turn it really bad, and that's exactly what people will do. The reason they can do that is because AI is a misnomer, it's not actually intelligent, it mimics intelligence, and as long as you are the one setting the boundaries it does ok, if you don't, and many people won't, it can get bad very quickly.

I understand. I wouldn’t want to see a Galnet headline like ‘Phishing Emails Created Using Ship Chat AI,’ either.

 

[Easy]

Quality prompts with ChatGPT yield quality answers. If you're getting inaccurate responses, the reason is most likely your prompt is low quality.