Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI behavior guardrails should be public

twitter.com

Post reply on HN
Print view View original

111–120 of 350 posts

Re: AI behavior guardrails should be public

#111

Post by transitionnel »

ceejayoz wrote: post #101
DebtDeflation wrote: post #90

Earlier quoted context omitted.

Surely there is a middle ground. "Generate a scene of a group of friends enjoying lunch in the park." -> Totally expect racial and gender diversity in the output. "Generate a scene of 17th century kings of Scotland playing golf." -> The result should not be a bunch of black men and Asian women dressed up as Scottish kings, it should be a bunch of white guys.

You can see how this gets challenging, though, right? If you train your model to prioritize real photos (as they're often more accurate representations than artistic ones), you might wind up with Denzel Washington as the archetype; https://en.wikipedia.org/wiki/The_Tragedy_of_Macbeth_(2021_f... . There's a vast gap between human understanding and what LLMs "understand".

If they actually want it to work as intelligently as possible, they'll begin taking these complaints into consideration and building in a wisdom curating feature where people can contribute.

This much is obvious, but they seem to be satisfied with theory over practicality.

Anyway I'm just ranting b/c they haven't paid me.

How about an off the wall algorithm to estimate how much each scraped input turns out to influence the bigger picture, as a way to work towards satisfying the copyright question.

Re: AI behavior guardrails should be public

#112

Post by mike_d »

Workaccount2 wrote: post #34

Harris and who I think was either Hughes or Stewart a podcast where they talked about how cringey and out of touch the elite are on the topic of race or wokeness in general. This faux pas on google's part couldn't be a better illustration of this. A bunch of wealthy rich tech geeks programming an AI to show racial diversity in what were/are unambiguously not diverse settings. They're just so painfully divorced from r…

I've found that anyone who uses the term "wokeness" seriously is likely arguing from a place of bad faith.

It's origins are as a derogatory term, which people wanting to speak seriously on the topic should know.

Re: AI behavior guardrails should be public

#113

Post by rahidz »

lp0_on_fire wrote: post #84
janalsncm wrote: post #59

Earlier quoted context omitted.

I’d be curious to hear that podcast if you could link it. If that was genuinely his opinion, he’s missed the forest for the trees. Brand safety is the dominant factor, not “wokeness”. And certainly not by the choice of any individual programmer. The purpose of these tools is quite plainly to replace human labor and consolidate power. So it doesn’t matter to me how “safe” the AI is if it is displacing workers and dump…

> As long as the Left is worrying about whether the chatbots are racist, people won’t pay attention to the net effect of these tools. It's by design. A country obsessed with racial politics has little time for the politics of anything else.

Exactly. What a coincidence that the media's obsession with race and gender inequalities began right after Occupy Wall Street.

Re: AI behavior guardrails should be public

#114

Post by CuriouslyC »

photoGrant wrote: post #98
chasd00 wrote: post #93

Earlier quoted context omitted.

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. OpenA…

Compute lets you fuck up a lot when trying to build a model, but you need data to do anything worth fucking up in the first place, and if you have 20% of the compute but you fuck up 1/5th as much you're doing fine.

Meta/OpenAI/Google can fuck up a lot because of all their compute, but ultimately we learn from that as the scientists doing the research at those companies would instantly bail if they couldn't publish papers on their techniques to show how clever they are.

Re: AI behavior guardrails should be public

#115

Post by mp05 »

AlecSchueler wrote: post #104
DebtDeflation wrote: post #90

Earlier quoted context omitted.

Surely there is a middle ground. "Generate a scene of a group of friends enjoying lunch in the park." -> Totally expect racial and gender diversity in the output. "Generate a scene of 17th century kings of Scotland playing golf." -> The result should not be a bunch of black men and Asian women dressed up as Scottish kings, it should be a bunch of white guys.

As soon as you have then playing an anachronistic sport you should expect other anachronistic imagery to creep in, to be fair.

> anachronistic sport

Scottish kings absolutely played golf.

Re: AI behavior guardrails should be public

#116

Post by photoGrant »

CuriouslyC wrote: post #109
photoGrant wrote: post #54

Earlier quoted context omitted.

Remind yourself we're discussing censorship, misinformation, inability to define or source truth and we're concerned on Day 1 about the results of image gen being controlled by a for profit single entity with incentives that focus solely on business and not humanity... Where do we go from here? Things will magically get better on their own? Businesses will align with humanity and morals, not their investors? This is…

The ridiculous degree of PC alignment of corporate models is the thing that's going to let open source win. Few people use bing/dall-e, but if OpenAI had made dall-e more available and hadn't put ridiculous guardrails on it, stable diffusion would be a footnote at this point. Instead, dall-e is a joke and people who make art use stable diffusion, with casuals who just want some pretty looking pictures using midjourne…

No, ignoring laws and stealing data to increase your Castle's MOAT is the win. Compute isn't an open source solvable problem. I can't DirtyPCB's an A100

Making the argument open source is the answer is an agenda of making your competition spin wheels.

Re: AI behavior guardrails should be public

#117

Post by sho_hn »

transitionnel wrote: post #111
ceejayoz wrote: post #101

Earlier quoted context omitted.

You can see how this gets challenging, though, right? If you train your model to prioritize real photos (as they're often more accurate representations than artistic ones), you might wind up with Denzel Washington as the archetype; https://en.wikipedia.org/wiki/The_Tragedy_of_Macbeth_(2021_f... . There's a vast gap between human understanding and what LLMs "understand".

If they actually want it to work as intelligently as possible, they'll begin taking these complaints into consideration and building in a wisdom curating feature where people can contribute. This much is obvious, but they seem to be satisfied with theory over practicality. Anyway I'm just ranting b/c they haven't paid me. How about an off the wall algorithm to estimate how much each scraped input turns out to influen…

An LLM-style system designed to understand Wikipedia relevance and citation criteria and apply them might be a start.

Not that Wikipedia is perfect and controversy-free, but it's certainly a more sophisticated approach than the current system prompts.

Re: AI behavior guardrails should be public

#118

Post by redox99 »

slily wrote: post #81
matt3210 wrote: post #58

How is this any different than doing google image searches of the same prompts. Exmaple: Google image search for "Software Developer" and you get results such that there will be the same amount of women and men event though men make up the large majority of software developers. Had Google not done this with its AI I would be surprised. There's really no problem with the above... If I want male developers in image sea…

Google injecting racial and sexual bias into image search results has also been criticized, and rightly so. I recall an image going around where searching for inventors or scientists filled all the top results with black people. Or searching for images of happy families yielded almost exclusively results of mixed-race (i.e. black and non-black) partners. AI is the hot thing so of course it gets all the attention righ…

I tried "scientists" and there's definitely a bias towards the US vision of "diversity".

Re: AI behavior guardrails should be public

#119

Post by fatherzine »

hackerlight wrote: post #15
kaesar14 wrote: post #3

Curious to see if this thread gets flagged and shut down like the others. Shame, too, since I feel like all the Gemini stuff that’s gone down today is so important to talk about when we consider AI safety. This has convinced me more and more that the only possible way forward that’s not a dystopian hellscape is total freedom of all AI for anyone to do with as they wish. Anything else is forcing values on other people…

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…

BigTech, which critically depends on hyper-targeted ads for the lion share of its revenue, is incapable of offering AI model outputs that are plausible given the location / language of the request. The irony.

- request from Ljubljana using Slovenian => white people with high probability

- request from Nairobi using Swahili => black people with high probability

- request from Shenzhen using Mandarin => asian people with high probability

If a specific user is unhappy with the prevailing demographics of the city where they live, give them a few settings to customize their personal output to their heart's content.

Re: AI behavior guardrails should be public

#120

Post by CuriouslyC »

photoGrant wrote: post #116
CuriouslyC wrote: post #109

Earlier quoted context omitted.

The ridiculous degree of PC alignment of corporate models is the thing that's going to let open source win. Few people use bing/dall-e, but if OpenAI had made dall-e more available and hadn't put ridiculous guardrails on it, stable diffusion would be a footnote at this point. Instead, dall-e is a joke and people who make art use stable diffusion, with casuals who just want some pretty looking pictures using midjourne…

No, ignoring laws and stealing data to increase your Castle's MOAT is the win. Compute isn't an open source solvable problem. I can't DirtyPCB's an A100 Making the argument open source is the answer is an agenda of making your competition spin wheels.

You're on a thread about how people are lambasting big money AI for being garbage, and producing inferior results to OSS tools you can run on consumer GPUS, tell me again how unbeatable google/other big tech players are.
Post reply on HN