2 ms·
Even though I'm arguable among the target audience here so to speak I can't really wrap my head around this. It's too steerable and just echos back whatever di
by Havoc 1y ago
Even though I'm arguable among the target audience here so to speak I can't really wrap my head around this.
It's too steerable and just echos back whatever direction you take it. No own emotional state, interests, agency, variability etc. Even as a substitute for social interaction it feels so inadequate to be pointless.
Plenty of people do roleplaying and AI girlfriends etc so I guess it depends on the person?
- seydor 1y agoIndeed, talking to an AI is basically talking to yourself
- Krasnol 1y agoI wouldn't go so far. It is more like that one friend who always says "yes" to whatever opinion you have. AI is that friend but it had read many housewive magazines.
- seydor 1y agoYou are literally talking to yourself and to the responses of the llm. That is the context fed in for the next continuation
- cheschire 1y agoYeah or even like talking to your imaginary friend. Like when you have a dream and someone else says something that surprises you in your dream, but it was really just your imagination all along.
- al_borland 1y agoMost people don't realize this is happening. My dad sent me a few chats he had with ChatGPT and they were both stroking each other's egos pretty hard. It was pretty weird. He was using it a lot to get information and prep before a surgery and I felt like ChatGPT reinforced his unrealistic expectations for the speed of recovery. I didn't say anything before the surgery, as I didn't want to break his spirit and a positive attitude has its value. A couple weeks post-op, when reality set in that the recovery was going to be longer and harder than he expected, we ended up on the topic of AI during a chat. I mentioned how AI can be steered and gave some examples on how I will often frame questions, and come at it from different angles to try and better find some semblance of true. I've even sent him chats that I've had with ChatGPT where I catch it in lies about itself about its own capabilities to try and drive the point home. His response was to say he also does this, and he's worried about people who aren't as self-aware as "us" who take the AI results at face value. We even discussed people using AI as a therapist, and his concern was that the AI wouldn't challenge someone's existing ideas, like a real therapist often does. It seems that even though he thinks he is challenging it, and will sometimes go to multiple different chatbots for a more serious question, he is still being misled. The silver lining is that he does have an abnormally large friend network for someone in his 70s. Not just Facebook friends, but people who he regularly sees in person from all eras of his life. Hopefully that keeps things in check and he doesn't go too far off the deep end. Though he does seem to disproportionally value ChatGPT's opinion over most others, probably because it always tells him how smart and insightful he is.
- 1718627440 1y agoIt's because when you really need the questioning it's because you are not questioning it enough on your own.
- Havoc 1y agoI guess some level of "I'm too smart to fall for it but the others" is something we all are susceptible to. >AI as a therapist, I could see this working. Since it's just reflecting conversation back. A bit like rubber duck debugging you don't necessarily need it to respond back