jamilton

26 ms·

361.
by jamilton 4y ago
The government of Israel is not particularly known for being ethical.
362.
by jamilton 4y ago
I really don't think we can say that research will apply to future AI, given that it was about humans. If intelligent AI exists in the future, it will probably not think exactly like humans. I think you're being overly dismissive.
363.
by jamilton 4y ago
419 accidents involving self-driving (level 2 and 3), 18 deaths, 19 accidents with injury level unknown [0]. All deaths from level 2 vehicles. So being pessimistic, maybe 50 deaths from self-driving. The people signing this are worried abou
364.
by jamilton 4y ago
What do you mean by "democratizing the LLMs"? There are already open source LLMs. Making Claude, GPT-4, and whatever else is being worked on open source wouldn't make them any less capable of doing harm.
365.
by jamilton 4y ago
Humanity is still in its "wild west" stage by that definition.
366.
by jamilton 4y ago
Why should this be called the metaverse?
367.
by jamilton 4y ago
It was the killing people and colonization that was bad. A country letting people move there is not either of those things, that is a terrible argument. No one is saying "Let's force Iceland at gunpoint to let everyone in". E
368.
by jamilton 4y ago
I wouldn't expect them to be any better than current extensions, which are pretty good.
369.
by jamilton 4y ago
UCLA SETI is doing a similar project, they're collecting data for a model via volunteers/citizen science: arewealone.earth (redirects to zooniverse, a citizen science platform). You classify different types of radio signals accord
370.
by jamilton 4y ago
I guess I was assuming "efficiency" because if it's not effective, why would anyone use it?
371.
by jamilton 4y ago
Un-"limited" LLMs (to me meaning pre-RLHF) look more like GPT-3, which instead of responding to a prompt will just continue it. As an example, you'll ask it a question, and it will give 5 more questions. To get it to write ab
372.
by jamilton 4y ago
ChatGPT was RLHF'd into answering questions (instead of just continuing them) and using "I" from some base model. You start over from that base model and RLHF it into something non-conversational. And ideally you'd take
373.
by jamilton 4y ago
It might prevent users from anthropomorphizing LLMs as much.
374.
by jamilton 4y ago
Honestly I'd love to see someone do a chatbot but instead of being a customer service employee it's trained to sound like E-Prime (minus "I" as suggested by the article, too). I don't expect it to make it more usefu
375.
by jamilton 4y ago
Why do you think it wouldn't be efficient for general use? Do you think it'll be expensive?
376.
by jamilton 4y ago
I feel like most hobbyist authors enjoy writing-as-writing enough that they wouldn't use AI to write something for them. But there's definitely a minority who mostly just want a story to exist who probably would use a better versi
377.
by jamilton 4y ago
I don't understand what you mean by "the other way".
378.
by jamilton 4y ago
The idea is that if it's solely corporate owned maybe AGI happens a few years later than if it was open-sourced, which gives time for alignment research to potentially come up with a solution to creating a safe AGI, which should be har
379.
by jamilton 4y ago
I don't think they're saying Bing/Sydney is sentient, they're saying it's misaligned: Microsoft probably did not want it to say problematic things, and likely spent a fair amount of money to that point and it still
380.
by jamilton 4y ago
Yeah, I thought it was funny that most of the page is "in the future...". Guess they're trying to get attention for the product before it's ready and gauge interest/response.
381.
by jamilton 4y ago
I think the naive approach of just dividing revenue equally across all contributors could be acceptable, and would have lower overhead costs.
382.
by jamilton 4y ago
I think it's really unlikely it's the last one. I think it's closer to the second.
383.
by jamilton 4y ago
They said upfront that they were using it in a non-standard way, I think it's fine.
384.
by jamilton 4y ago
I wonder how much the hardware they're using costs.
385.
by jamilton 4y ago
I don't think it actually, meaningfully diminishes "actual labor struggles". The working class is not going to have a harder time organizing because some tech workers did it first.
386.
by jamilton 4y ago
Who is "the existing organization" if not the employees of that organization?
387.
by jamilton 4y ago
I think training on output is one method that people use to create customized Stable Diffusion models. IIRC, it doesn't work well unless you manually label it, in which case it works fine. Letting existing AI label it doesn't do v
388.
by jamilton 4y ago
I recommend just calling it software, or something specific like "image generator", "text generator". It gets the point across well, in my experience.
389.
by jamilton 4y ago
Copyright is not necessarily the best way for creators to get compensated. For example, for most musicians in the US it's only a small part of their income [1]. Study is from 2013, could be different in the Spotify era, but it found th
390.
by jamilton 4y ago
>Lack of IP protection in our culture has made it impossible for 99.9% of artists to thrive economically. Can you explain this position? My understanding is most artists don't thrive economically because there's not much demand

More ›