I think HN moderation guardrails should be public.
AI behavior guardrails should be public
91–100 of 350 posts
Re: AI behavior guardrails should be public
#92gemini seems to have problems generating white people and honestly this just opens the door for things that are even more racist [1], the harder you try the more you'll fail, just get over the DEI nonsense already 1. https://twitter.com/wagieeacc/status/1760371304425762940
Is there any evidence that this is a consequence of DEI rather than a deeper technical issue?
When DALL-E 2 was released in 2022, OpenAI published an article noting that the inclusion of guardrails was a correction for bias: https://openai.com/blog/reducing-bias-and-improving-safety-i...
It was widely criticized back then: the fact that Google both brought it back and made it more prominent is weird. Notably, OpenAI's implementation is more scoped.
Re: AI behavior guardrails should be public
#93Earlier quoted context omitted.
> Where do we go from here? opensource models and training sets. So basically the "secret sauce" minus the hardware. I don't see it happening voluntarily.
Absolutely it won't. We've armed the issue with a supersonic jet engine and we're assuming if we build a slingshot out of pop sticks we'll somehow catch up and knock it off course.
i can't predict the future but there is precedent. Models, weights, and dataasets are the keys to the kingdom like operating system kernels, databases, and libraries use to be. At some point, enough people decided to re-invent and release these things or functionality to all that it became a self-sustaining community and, eventually, transformative to daily life. On the other hand, there may be enough people in power who came from and understand that community to make sure it never happens again.
Re: AI behavior guardrails should be public
#94Earlier quoted context omitted.
I'm convinced this happens because of technical alignment challenges rather than a desire to present 1800s English Kings as non-white. > Use all possible different descents with equal probability. Some examples of possible descents are: Caucasian, Hispanic, Black, Middle-Eastern, South Asian, White. They should all have equal probability. This is OpenAI's system prompt. There is nothing nefarious here, they're asking…
> As these systems get better, they'll figure out that "1800s English" should mean "White with > 99.9% probability". The thing is, they already could do that, if they weren't prompt engineered to do something else. The cleaner solution would be to let people prompt engineer such details themselves, instead of letting a US American company's idiosyncratic conception of "diversity" do the job. Japanese people would pro…
I believe we can reach a point where biases can be personalized to the user. Short prompts require models to fill in a lot of the missing details (and sometimes they mix different concepts together into 1). The best way to fill in the details the user intended would be to read their mind. While that won't be possible in most cases getting some kind of personalization to help could improve the quality for users.
For example take a prompt like "person using a web browser", for younger generations they may want to see people using phones where older generations may want to see people using desktop computers.
Of course you can still make a longer prompt to fill in the details yourself, but generative AI should try and make it as easy as possible to generate something you have in your mind.
Re: AI behavior guardrails should be public
#95Earlier quoted context omitted.
Why would this be flagged / shut down? Also, what Gemini stuff are you referring to?
Carmack’s tweet is about what’s going around Twitter today regarding the implicit biases Gemini (Google’s chatbot) has when drawing images. Will refuse to draw white people (and perhaps more strongly so, refuses to draw white men?) even in prompts where appropriate, like “Draw me a Pope” where Gemini drew an Indian woman and a Black man - here’s the thread: https://x.com/imao_/status/1760093853430710557?s=46 Maybe in…
I don't even know how people get it to draw images, the version I have access to is literally just text.
Re: AI behavior guardrails should be public
#96It's super easy to run LLMs and Stable Diffusion locally -- and it'll do what you ask without lecturing you.
If you have a beefy machine (like a Mac Studio) your local LLMs will likely run faster than OpenAI or Gemini. And you get to choose what models work best for you.
Check out LM Studio which makes it super easy to run LLMs locally. AUTOMATIC1111 makes it simple to run Stable Diffusion locally. I highly recommend both.
Re: AI behavior guardrails should be public
#97Earlier quoted context omitted.
Is there any evidence that this is a consequence of DEI rather than a deeper technical issue?
I dont think so. My boss wanted me to generate a birthday image for a co-worker of a John Cena flyfishing. ChatGPT refused to do so. So I had to move to describing the type of person John Cena is instead of using his name. I kept giving me bearded people no matter what. I thought this would be the perfect time to try out Gemini for the first time. Well shit, It wont even give me a white guy. But all the black dudes a…
It feels that the image generation it offers is perfect for some sort of a California-Corporate Style, e.g. you ask it for a "photo of people at the board room" or "people at the company cafeteria" and you get the corporate friendly ratio of colors, ability-levels, sizes etc. See Google's various image assets: https://www.google.com/about/careers/applications/ . It's great for coastal and urban marketing brochures.
But then then same California Corporate style makes no sense for historical images, so perhaps this is where Midjourney comes in.
Re: AI behavior guardrails should be public
#98Earlier quoted context omitted.
Absolutely it won't. We've armed the issue with a supersonic jet engine and we're assuming if we build a slingshot out of pop sticks we'll somehow catch up and knock it off course.
i can't predict the future but there is precedent. Models, weights, and dataasets are the keys to the kingdom like operating system kernels, databases, and libraries use to be. At some point, enough people decided to re-invent and release these things or functionality to all that it became a self-sustaining community and, eventually, transformative to daily life. On the other hand, there may be enough people in power…
No, compute is keys to the kingdom. The rest are assets and ammunition. You out-compute your enemy, you out-compute your competition. That's the race. The data is part of the problem, not the root.
These companies are silo'ing the worlds resources. GPU, Finance, Information. Those combined are the weapon. You make your competition starve in the dust.
These companies are pure evil pushing an agenda of pure evil. OpenAI is closed. Google is Google. We're like, ok, there you go! Take it all. No accountability, no transparency, we trust you.
Re: AI behavior guardrails should be public
#99Earlier quoted context omitted.
You want large tech companies "creating reality" on behalf of everyone else? They're not even democratic institutions that we vote on. You trust they will get it right? Our benevolent super rich overlords.
Its not really a question about want, its a question about facts. Their actions will make a significant mark on the future. So far it seems like they are trying to promote positive changes such as inclusion and equality. Which is far far far fucking really infinitely far better than trying to promote exclusion and inequality
This switch might flip instantaneously.
Re: AI behavior guardrails should be public
#100Earlier quoted context omitted.
"The only way to deal with some people making crazy rules is to have no rules at all" --libertarians "Oh my god I'm being eaten by a fucking bear" --also libertarians
"can you write the rules down so i know them?" --everyone
"No" --Every company that does moderation and spam filtering.
"No" --Every company that does not publish their internal business processes.
"No" --Every company that does not publish their source code.
Honestly I could probably think of tons of other business cases like this, but in the software world outside of open source, the answer is pretty much no.