Curious to know what my fellow micro.bloggers think about ChatGPT!
@joshuapsteele it’s kinda fun. Check out Federico Viticci’s short cut if you are on iOS: www.macstories.net/ios/intro...
@joshuapsteele My thoughts are quite negative I'm afraid. I'm not sure MB is the best place to go in depth (I've toiled for weeks on a blog post that would properly outline my concerns, not there yet), but I am scared of the unintended impact current and next versions of generative AI will have, and suspicious of the motivations of people who are working to advance it. In other words, I am one of those pejoratively referred to as "doomers".
@joshuapsteele I'm very curious about it, and trying to use it to find out how it could help me with daily tasks as a developer. But I'm also very skeptical. I had a discussion about AI safety with ChatGPT a couple of days ago. I think the destruction of humans is far off and there are countless possibilities, but we need to be careful how all of this evolves.
@sherif @joshuapsteele I’m afraid my feelings mirror Sherif’s; not a fan, and greatly concerned about future implications of AI in general. Ignoring whatever nefarious purposes it ends up being used for — it’s already been seen in password cracking — we’re handing over our intelligence and independent thought to computers. I can’t help feeling humanity is too quick to pass off our innate abilities, rather than keeping our minds and decision making sharp. What would be left for us if we pass the buck on everything to machines?
@hutaffe I read your post about AI safety and ChatGPT (sans the transcript). Your thoughts mirror mine very closely and you seem familiar with the same discussions and debates, and so I have to ask, how do you... manage your anxiety about it all? It's been a real struggle for me the last few weeks.
@joshuapsteele With good prompts & coaching replies, it can do cool stuff. For basic prompts, its writing tends to be formulaic & lacking in specificity. For identifying research on a topic, it completely fabricates things that sound very real, which is dangerous.
@sherif honestly… I’m not sure. I think a crucial point is to learn about the capabilities, try to understand how these things work, and to stay up to date on the evolution. Also, trying to raise awareness by telling others about what’s going on. I can’t stop the movement, so I have to make use of it and be aware that it could, theoretically, end horribly.
@joshuapsteele I've got serious concerns about the potential for amplification of bias and misinformation. Have you looked at the stochastic parrots paper? It's got a lot of info on how the training approach introduces bias into the model -- and the idea of "value lock" for reinforcing existing understandings. My notes from the paper
@sherif ugh they rhymed the pejorative with "boomer"
Us neo-Luddites need better branding for this.
If smartphone -> BURNER phone
Generative AI -> ??
Human creativity? Artists?