News for hackers. Stuff that matters.
Re:DALL·E Now Available Without Waitlist by Uehreka on Wednesday September 28, 2022 @03:21PM (#33012056)
>contrary to their initial openness goals

Their goals were never about openness at all though. From the beginning I’ve felt like they should’ve called themselves something like “SafeAI”, since their stated goal was basically to develop advanced AI first, then keep a lid on it until they could somehow ensure it was “safe” or would only be used by “good” people.

Sure, OpenAI might sound nicer, but it also drags this contradiction into the foreground whenever someone says their name.

Re:DALL·E Now Available Without Waitlist by somenameforme on Wednesday September 28, 2022 @04:28PM (#33012839)
Wow, I was about to argue against this 'unfairly cynical' take, but it's completely correct.

---

(2015) OpenAI's original "Introducing OpenAI Post" : https://openai.com/blog/introducing-openai/ : "As a non-profit, our aim is to build value for everyone rather than shareholders. Researchers will be strongly encouraged to publish their work, whether as papers, blog posts, or code, and our patents (if any) will be shared with the world. We’ll freely collaborate with others across many institutions and expect to work with companies to research and deploy new technologies."

(2018) OpenAI's "Charter" : https://openai.com/charter/ :

"We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions. Therefore, if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project."

"We are committed to providing public goods that help society navigate the path to AGI. Today this includes publishing most of our AI research, but we expect that safety and security concerns will reduce our traditional publishing in the future, while increasing the importance of sharing safety, policy, and standards research."

---

Provides some interesting context to the fact that Elon left the company's board in February 2018 over "disagreements about the company's development."

Re:DALL·E Now Available Without Waitlist by moffkalast on Wednesday September 28, 2022 @04:32PM (#33012882)
> if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project

Haha, and then they would proceed to get told to politely piss off.

Re:DALL·E Now Available Without Waitlist by bogomipz on Thursday September 29, 2022 @01:56AM (#33016959)
That's interesting. Might you or someone else be able to say what was the catalyst for this change in mission? What happened between 2015 and 2018?
Re:DALL·E Now Available Without Waitlist by ryanmcbride on Wednesday September 28, 2022 @04:15PM (#33012687)
For real it's insane to me how much I bump up against their community guidelines. For example, you'll get a community guidelines block if you enter a prompt like "An illustration of a computer in the style of Henry Vandyke Carter".

removing "Vandyke" from the prompt lets it go through[1], but doesn't result in the style I want. Because there's no artist that I'm aware of that goes by "Henry Carter". The middle name is important.

It reminds me of the old 2D Runescape days where the language filter would convert "dictionary" to "**tionary".

[1]https://ibb.co/4106mfF

Re:DALL·E Now Available Without Waitlist by Miraste on Wednesday September 28, 2022 @04:49PM (#33013073)
It's doubly pathetic because of how they frame it:

-We are the world's most advanced AI company.

-Our filter verifiably acts as a simple blacklist.

-You aren't allowed to see the blacklist because it's really a "contextual" filter, so you'll have to guess.

-If you guess wrong too many times you'll be banned.

-Using our service more often increases the chance you'll hit the number of wrong guesses.

-No, you can't know what that number is.

Re:DALL·E Now Available Without Waitlist by pmarreck on Thursday September 29, 2022 @01:52PM (#33024591)
I ran up against a warning because I tried to enhance/expand old videogame cover art which happened to feature... sit down because this will shock you... fighting!
Re:DALL·E Now Available Without Waitlist by wyldfire on Wednesday September 28, 2022 @04:46PM (#33013038)
Scunthorpe prob [1]. Apparently they couldn't just unleash a model to fix this [2]

[1] https://en.wikipedia.org/wiki/Scunthorpe_problem

[2] https://www.techdirt.com/2018/08/31/scunthorpe-problem-why-a...

Re:DALL·E Now Available Without Waitlist by solidasparagus on Thursday September 29, 2022 @02:40AM (#33017224)
If only there existed some tool that understood language deeper than substrings!

(That second link describing how AI can't understand language well enough to solve the problem predates GPT-3 by 2 years)

Re:DALL·E Now Available Without Waitlist by pmarreck on Thursday September 29, 2022 @02:04PM (#33024760)
I never heard of the name of that problem and now I will never forget it! Thank you!

I'm actually on a 3 day Facebook ban because I posted a very legit medical NIH link that looked quite innocent (despite dealing with sexual organs- it was an article about a case of urethral intercourse, which I was including to demonstrate how little people, or perhaps just Americans, seem to understand human anatomy) and unbeknownst to me, the preview card picked a closeup of some kind of vaginal surgery to feature, and that resulted in an instaban. Fffffuuuuuuuuu

It was basically the visual version of this

Re:DALL·E Now Available Without Waitlist by msoucy on Wednesday September 28, 2022 @04:44PM (#33013001)
My favorite example of an issue like this (the Scunthorpe problem) is from the mobile game Kingdom Hearts Unchained X. In it, players used Medals based around Disney characters, including experiment 626. However, for a long time after the game's release, players were unable to say that name in chat, because it got rendered as s***ch
Re:DALL·E Now Available Without Waitlist by Blahah on Wednesday September 28, 2022 @04:28PM (#33012837)
H V Carter
Re:DALL·E Now Available Without Waitlist by ryanmcbride on Wednesday September 28, 2022 @04:31PM (#33012874)
I'm out of credits but I'll try that later
Re:DALL·E Now Available Without Waitlist by metacritic12 on Wednesday September 28, 2022 @04:24PM (#33012796)
Yup, OpenAI is founded by AI-safety-as-a-religion people. They're essentially single-issue voters, who believes earnestly their issue is the only issue that matters. You see analogues of them in e.g. climate change (right or wrong).

This religion definitely has a parentalist bent to it that rubs a lot of people the wrong way. I vaguely recall them floating on Twitter the theoretical idea of whether murdering people to prevent AI-takeover is acceptable, due to how bad AI-takeover is.

Not surprising limiting access, spying on what its users are using their tools for, etc, is acceptable to them.

This is much in the same vein as how for Lenin, the eventual triumph of the working class is so important as to justify a little bit of interim violence, dictatorship, and summary executions.

Re:DALL·E Now Available Without Waitlist by Miraste on Wednesday September 28, 2022 @04:45PM (#33013021)
The difference being that climate change activists have mountains of data, decades' worth, backing their cause, while OpenAI and friends have a sci fi story they made up, based on nothing. The whole "AI alignment" movement is the worst example of arrogance in modern tech. Even the nomenclature screams condescension - the imaginary AGI needs "aligned values?" Aligned with whose values? Invariably it ends up being the creators', at the expense of squashing everyone else's. The DALL-E "acceptable use" rules are a dystopian nightmare and they are born of incredibly pompous self-righteousness.
Re:DALL·E Now Available Without Waitlist by nearbuy on Thursday September 29, 2022 @12:15AM (#33016259)
Ironically, your comment also sounds arrogant and condescending.

The AI safety crowd's main concern isn't that an "unaligned" superintelligence will have some other people's values. It's that an unaligned AI might kill everyone.

Of course, that isn't a concern with DALL-E. It's like if the safety crowd was worried about a ferocious tiger and OpenAI was like, "we got a kitten. Let's keep the public safe from it while we develop better kitten-handling gloves."

Then Midjourney and Stable Diffusion get their own kittens and let everyone play with them and OpenAI finally says, "Okay, everyone can safely play with our kitten now because we carefully developed great kitten-handling gloves" and proceeds to hand you plain dollar store gloves.

Re:DALL·E Now Available Without Waitlist by Miraste on Thursday September 29, 2022 @03:20PM (#33025731)
I know how they define "alignment." I'm picking on the tortured nomenclature, and the tendency in current AI research for it to look a lot like "my values."

In my view, speculating on AGI ethics is at best pointless. It's like trying to write laws for the Internet during the invention of the telephone. If you imagine how it could operate, you'll be wrong, and the details change the whole problem.

Re:DALL·E Now Available Without Waitlist by addingadimensio on Wednesday September 28, 2022 @04:54PM (#33013131)
Shutting down nuclear power is also dystopian.
Re:DALL·E Now Available Without Waitlist by Miraste on Wednesday September 28, 2022 @05:02PM (#33013217)
Anyone who calls themself a climate change activist and supports shutting down nuclear plants is - well, I prefer not to use invective on HN, so let's say extremely misguided.
Re:DALL·E Now Available Without Waitlist by addingadimensio on Wednesday September 28, 2022 @05:53PM (#33013759)
So the green party in basically every country?
Re:DALL·E Now Available Without Waitlist by jiggawatts on Wednesday September 28, 2022 @07:47PM (#33014642)
Yes.

We had a Greens candidate here running for election, and his main campaign promise was to shut down a 10+ billion dollar motorway underpass that was 99% completed at the time of the election.

His approach for "saving the environment" is to convert efficient motorway travel into inefficient stop-start traffic. And to throw away tens of billions of dollars of already spent resources.

Re:DALL·E Now Available Without Waitlist by Tostino on Thursday September 29, 2022 @12:13AM (#33016245)
As much as I'd like to support actual competent Green candidates, this example is what I've seen my entire life in the US.
Re:DALL·E Now Available Without Waitlist by addingadimensio on Thursday September 29, 2022 @10:39AM (#33021327)
You know what would actually take some political willpower?

Banning plastic packaging and waterproofing.

My bet is that just like before i was born, after I’m gone these things will only increase.

Re:DALL·E Now Available Without Waitlist by circuit10 on Thursday September 29, 2022 @07:48AM (#33019274)
"It sounds like sci-fi" isn't a valid reason to think something is impossible. Many things in the modern world would have sounded like sci-fi 50 or 100 years ago. This type of reasoning has a name: https://en.m.wikipedia.org/wiki/Appeal_to_the_stone
Re:DALL·E Now Available Without Waitlist by Miraste on Thursday September 29, 2022 @03:13PM (#33025626)
I don't think human or greater artificial intelligence is impossible. I do think the specifics of how it works are both very important to any ethical problems and not currently predictable. My objection isn't that alignment problems sound like sci-fi but that they're very soft sci-fi, and aren't based on meaningful predictions.
Re:DALL·E Now Available Without Waitlist by alasdair_ on Wednesday September 28, 2022 @07:20PM (#33014460)
>the imaginary AGI needs "aligned values?" Aligned with whose values?

This is something that is pretty well addressed in their writings and is definitely a heavily considered topic. It gets quite heavily into philosophy because ideally you want to count future-you's preferences as well as current-you and you want to avoid, say, genocide just because most people dislike a certain group.

A good place to start is here: https://www.lesswrong.com/tag/coherent-extrapolated-volition but there is a LOT of discussion of this topic.

Re:DALL·E Now Available Without Waitlist by Miraste on Wednesday September 28, 2022 @09:50PM (#33015456)
I'm familiar with the Yudkowsky school of thought here; it's largely where my complaint about basing philosophy off of sci-fi stories comes from. Picking on some of the more obviously inane ideas that have come out of it, like Roko's Basilisk, is too easy, but the page you linked is a great example of the problem. I quote in full:

"In calculating CEV, an AI would predict what an idealized version of us would want, "if we knew more, thought faster, were more the people we wished we were, had grown up farther together". It would recursively iterate this prediction for humanity as a whole, and determine the desires which converge. This initial dynamic would be used to generate the AI's utility function."

This is too detached from the reality of AI development to be useful. I can't make a utility function like this, nor can you, nor can humanity. There's no reason to think a human-equivalent AI could either - the data for such a function doesn't exist, and can't exist because of how vague all the terms are. The current ML revolution is built on statistical pattern recognition. This definition would better fit a genie.

The fact that MIRI does no ML research and has no dialogue with the state of the craft only furthers my impression that it produces a lot of words with little substance.

Even if one still accepts the ultimate conclusions of the movement, using its AGI-focused rhetoric to justify restrictions on a simple image generator is silly.

Re:DALL·E Now Available Without Waitlist by mach1ne on Thursday September 29, 2022 @03:01AM (#33017346)
I wouldn't make too restricted assumptions on the form of a future AGI though. I find theoretical projections from a more axiomatic level quite important. It's like making rules for nukes before they were invented by assuming an abstract apocalypse-capable weapon without knowledge of missiles or nuclear fission.

That being said, the absolute majority of AI safety theory seems to fall into the same pothole where philosophy falls; modelling the world through language-logic rather than probabilities. The example you quoted fits this category - it's way too specific and thus unlikely to be useful in any way, even though its wording may deceive its author to believe it to be an inescapable outcome.

Re:DALL·E Now Available Without Waitlist by antupis on Wednesday September 28, 2022 @11:27PM (#33015970)
Thing with climate change is that at like semi bad scenario it will kill about million people yearly https://www.nature.com/articles/s41467-021-24487-w and those everybody is going to die scenarios are also very unlikely and basically scifi story at this point.
Re:DALL·E Now Available Without Waitlist by gzer0 on Thursday September 29, 2022 @12:36AM (#33016401)
It's completely insane to me to see comment chains like this. Who cares if climate change is man-made or not; we are witnessing an evolving climate landscape, and irrespective of who or what caused it, we can do things to help mitigate some of the damage.

We are quite literally watching the world burn before our eyes, and we are too stupid and self-centered to do anything about it.

> it will kill a million people yearly

And it's even more outlandish that such a number and atrocity upon mankind is brushed away. That is one million humans.

Re:DALL·E Now Available Without Waitlist by mach1ne on Thursday September 29, 2022 @02:46AM (#33017257)
I see that the reason for climate change denial is two-fold; on one hand, climate change wrecking chaos is mainly restricted to projections, and on the other, climate change activists as a group express many qualities which make people not want to give weight to their claims, mainly naivete and arrogance. Since we're not seeing too significant effects of climate change yet, people would need to research the subject in order to get an accurate image of possible future scenarios. But how they see climate change activist group expressing themselves puts them off the subject. Psychologically the best option in this cases is denialism of some amplitude.

I'm not trying to crucify any group here by the way. This is just how I see the issue.

Re:DALL·E Now Available Without Waitlist by freedomben on Thursday September 29, 2022 @10:39AM (#33021335)
I think you nailed it. This is exactly what I think too. I've had many a conversation with climate change deniers and the rebuttal is pretty much always a (mostly valid) list of examples of naivete and arrogance. Often times the people best equipped to identify the problem are too close to it (or too myopic) to come up with a good solution to the problem that balances environmental efficacy with economic stability/growth, etc. Since a discussion on climate change almost always includes the person's preferred solution, it's easy to attack/dismiss. I try to point out that having bad solutions doesn't invalidate the existence of the problem, but at that point the motives are suspect and presumption of good faith is completely gone. It certainly doesn't help that a very small but vocal minority have boldly predicted the end of the world many times in the past. It's easy for a denier to look at that and laugh and dismiss the current warnings based on the past, much like we do with people predicting the end of the world based on Biblical numerology.

I wish there were enough naturally curious people willing to dive into the data and come to their own conclusions. When you look at the history of Earth's climate, and the speed with which greenhouse gases natural release into the environment, and compare it to the speed with which they now release (due to burning of fossil fuels, reduction of trees/plants that trap it) it is really quite common sense that what we're doing is not a good idea to continue indefinitely.

Re:DALL·E Now Available Without Waitlist by Miraste on Thursday September 29, 2022 @03:31PM (#33025886)
I wouldn't rely too much on that paper. Modeling deaths per additional ton of CO2 is bound to be inaccurate because the effects of increased atmospheric carbon dioxide aren't linear.
Re:DALL·E Now Available Without Waitlist by hwers on Wednesday September 28, 2022 @05:16PM (#33013382)
“They have literally floated the idea of whether murdering people to prevent AI-takeover is acceptable,”

Where? I probably believe you but it almost makes me worried about the well being of the stability ai founders (on a long term horizon)

Re:DALL·E Now Available Without Waitlist by metacritic12 on Wednesday September 28, 2022 @06:08PM (#33013888)
To be honest, I thought I saw it somewhere on Twitter but can't find it right now after a few minutes of search. It was proposed as a theoretical question, not as a statement -- like is AI so bad that if you had a time machine, it would be worth killing the pivotal people to prevent it. Terminator style.

I've modified my OP to clarify that.

Re:DALL·E Now Available Without Waitlist by moffkalast on Wednesday September 28, 2022 @04:30PM (#33012857)
> stated goal was basically to develop advanced AI first, then keep a lid on it until they could somehow ensure it was “safe” or would only be used by “good” people

That was the stated made up bullshit they spun because "we're keeping this walled to figure out how to squeeze the most profit out of it" doesn't go as well with their focus groups.

Re:DALL·E Now Available Without Waitlist by pmarreck on Thursday September 29, 2022 @01:50PM (#33024546)
I'm glad I can now download models to my computers in order to perform "UnsafeAI". I'm tired of this nanny-state "nudity is worse than violence" puritanical BS. If the cat is out of the bag, it's too late and you'll have to rely on the good-actor faith of the majority to keep nefarity in check, anyway. Any extra inconvenience or impediment you apply after that to good actors will just frustrate them, without any compensatory "win."
Re:DALL·E Now Available Without Waitlist by samstave on Wednesday September 28, 2022 @08:34PM (#33014963)
So what is the evil AI we all should worry about but may not know about? NSA.?