78 comments

[ 3.3 ms ] story [ 175 ms ] thread
https://archive.ph/RUNYd

Interesting, especially in light of yesterday's assertions that they're supporting Phillips: "Silicon Valley insiders are trying to unseat Biden with help from AI" https://archive.ph/DSrM5

OpenAI recently removed the restriction in their ToS against military uses. Maybe there was a merge conflict when another team tried to removed the restrictions on electoral campaigning.
Or they got a tasty military contract, and part of that was conditioned on not upsetting the current administration.
In short, trying their hardest to follow "avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power"
> Or they got a tasty military contract, and part of that was conditioned on not upsetting the current administration.

Killing people is okay, political bot making fun of current admin is not.

I'm hoping they modify the language to something to the effect of banning use related to enacting violence (and the other similar cases) as opposed to removing it entirely. My understanding was it was pretty blanket in terms of what wasn't allowed to the point that mundane use cases still weren't allowed. Maybe it's not a weaponizing contract. Maybe I'm too optimistic
Killing people outside of US is more ok than meddling with US politics and/or elections.

I mean that as not at all a dark joke, it’s just a fact.

> A new super PAC tied to OpenAI CEO Sam Altman and billionaire Bill Ackman is backing Democratic challenger Dean Phillips

Oh, well that's a pairing I didn't see coming for some reason.

PAC website: https://www.wedeservebetter.us/

I'm suspicious that any organization Sam Altman and Bill Ackman are involved in would have a spelling mistake on their front page.

> " Stagnate wages for two decades"

Dean Phillips seems relevant to this community. He has proposed cabinet positions for Elon Musk and Bill Ackman, and announced yesterday that he might be interested in running in the general under the No Labels ticket.

My understanding is that to make that work, he would need to switch parties pretty soon in order not to be affected by sore loser laws in many states that would prevent him from switching parties after the primary.

To me he appears to be an obvious Republican spoiler effort pushed by some of the wealthiest people in tech including Musk and Altman. If it were my boss pushing this I would be incensed.

And I would vote for him over the nonsense the Republicans and Democrats are offering any day.
That's not really the point, the point is it's not an honest campaign. Elon Musk isn't donating money because he thinks Phillips will win, he's doing it because he thinks it'll help Trump win. Because Phillips needs to act soon, he can't wait as he says until Joe Biden appears unable to win. He needs to make that assessment in a week or two, before the general campaign even really starts -- so it isn't honest.

IMO it is fine to run for president, and to vote for whoever you want. But running a campaign that is really intended to just give the edge to another guy, is dirty politics intended to fool voters with an illusion of choice.

When the richest people in the country engage in those tactics, isn't democratic, it's plainly oligarchic.

>the point is it's not an honest campaign

So it's a dishonest campaign because some rich people are giving him money? By this logic RFK Jr's campaign should be seen as a dishonest campaign as well since he's raised more money from more people than Phillips has.

RFK's campaign is also a spoiler, propped up in part by the same people. My sense is what happened is they realized he's a little too crazy to be effective, so they are hedging their bets with Phillips.

Exactly the same story though.

If RFK ran as a third party he'd take votes from the Republican candidate, not the Democrat one. What kind of Democrat is going to vote for someone who believes vaccines cause autism and HIV doesn't cause AIDS?
> What kind of Democrat is going to vote for someone who believes vaccines cause autism...

Before COVID, I would have guessed the biggest reservoir of anti-vax sentiment was among very liberal Democrats (think the people who are super into yoga, organic food, and naturopathic doctors).

COVID became a culture war issue, which lead to partisan Democrats lining up on the pro-mask, pro-lockdown, pro-vaccine side, but I have no idea if that disrupted the views of existing anti-vax Democrats (of the "vaccines cause autism" variety).

(comment deleted)
I think your observation is why you read a lot more about Phillips these days than RFK.

RFK's promoters switched to promoting Phillips when they realized RFK was drawing more conservatives for the antivax stuff, than democrats for the Kennedy name.

Wait till you hear where Trump and Biden are getting their money from
Do you have any evidence of these claims? I have seen such tactics employed before (by Democrats and Republicans) but only rarely is there evidence of the altier motives.
I just learned about him and find his campaign platform to be agreeable.

Unfortunately, in a two-party system one is compelled to vote for the official party candidate if they want their party to win. So voting for a challenger during primaries is great in that it helps send a message to the party, but in the general one must hold their nose and vote for the party's nominee.

If he runs in the general election it will be as a spoiler candidate

I dunno, I thought he came off pretty level-headed when I listened to him talk on All In. I don't think he's polling well enough for me to feel good throwing away my vote, but I'm certainly more interested in him than any other new candidate.

I'm not very politically savvy, though, so take everything I'm saying with a grain of salt.

https://www.youtube.com/watch?v=1hh8lcoJ1NA fwiw

Interesting conspiracy theory.

Musk, Ackman, Altman historically vote Democrat.

Dean didn't propose them as cabinet positions. He made an off the cuff joke.

Most Democrats want a different choice.

Why is competition within the party a bad thing? That anyone trying to compete must be a conspiracy by the others?

There's nothing wrong with competition within the party -- that's just a primary. What should clue you off is that Phillips announced yesterday he's considering a third-party run.
> What should clue you off is that Phillips announced yesterday he's considering a third-party run.

I don't think that follows. Biden is not a popular guy, and it appears his strategy is to crush all internal opposition while engineering a rematch with Trump (probably because he'd lose against anyone else).

What possible engineering could Biden do, even if he wanted to, to run against Trump? He’s ludicrously popular among the GOP.
>Biden is not a popular guy, and it appears his strategy is to crush all internal opposition

For example, the New Hampshire Democratic primary is officially unsanctioned by the Democratic Party.

Just listen to him speak for a single minute and you know that he is not relevant for this community...he might be more successful on LinkedIn.
The market for completely anonymous and uncensored models via API must be huge. How come nothing exists yet?
https://replicate.com does in fact already exist.
Can’t this bot developer just switch to them then? as replicate can’t know what it is being used for I would assume. Perhaps there was fine tuning or RAG being used that still needs a private and anonymous service solution.
They have a list of prohibited uses which, though not has extensive as OpenAI's, includes this gem at the end (emphasis added):

> To engage in any other conduct that restricts or inhibits anyone’s use or enjoyment of Service, or which, as determined by us, may harm or offend Company or users of Service or expose them to liability.

Edit: Also, their privacy policy makes it clear that "completely anonymous" is not a thing they offer— they collect emails and IP addresses, transmit them to the United States, and will deliver them to law enforcement upon request.

https://replicate.com/privacy

So it’s not anonymous then because they see what the customer is generating? Otherwise how can they know what it’s being used for?
Yes, it's definitely not anonymous. I added to my comment right as you replied—they collect IP addresses and usage data and explicitly say they will share that with law enforcement upon request. It's a US-based company, so they kind of have to.
Let’s say there exists some “illegal” or “criminal” output of an LLM, doesn’t that mean by definition the government has criminalized speech and thought? since the output of an LLM is by definition tokens of language.

Now image generation I guess we can see as different since we do agree some images or videos have content society sees as illegal.

The whole topic is fascinating to me. Censorship and regulation around generative AIs.

Why trust someone else if you want it to be anonymous? Buy a proper GPU, run a local model and that's it handled. Basically all open source inference engines support OpenAI's API format and act as drop in replacements.
"just spend $1000 on a new GPU"
How much do you think those API calls will cost you that require someone else to buy the GPU, maintain the server architecture, and also take a cut in the process? It only takes a 3-4 months to amortize if you want the instance to be up 24/7. And in the end you can still resell the GPU.
I rent spot instances on GCP with big GPUs on demand. It works quite well and since I'm just playing around I pay a very small amount per hour and kill it after I'm done.
Does this work on-demand? Could I build something I can query from my phone as needed?
Depends on your budget. If it's the absolute minimum, it won't be real time responses. As provisioning infra isn't instantaneous.
Just want to point out that $1000 is nothing to a lot of potential bad actors. $1 million is nothing.
Plenty exists

LM Studio lets you load in models from a built in HuggingFace browser, and sets up a server that accepts OpenAi 3.0 API requests, so everyone that built an app around ChatGPT API plugs in exactly the same

I'm working on this problem in my spare time, it requires a GPU (or two) RTX 3090 or 4090 because of the need for vRAM memory but I could also forsee APIs in front that allow worker nodes to run the quantized models for cheaper. the problem is model switching is hard so you would have to pick which model to run upfront for inference

openchat in particular ships with an inference server compatible with openai

follow me here: https://www.youtube.com/@fxhp1

Completely anonymous and uncensored anything is hard to come by. You really can't legally provide a service like that in most Western countries. Every service that has tried has ended up realizing why it doesn't exist.

Even if you tried to base your business outside of the West, you're left figuring out how to collect payments for it without getting blocked. LLMs are far more expensive to provide than plain text communication or a VPN.

Eventually it'll probably happen, but there are a lot of logistics to sort out first, and in the meantime the open source models still have a ways to go before they're good enough at arbitrary problems out of the box to be worth paying for.

The issue with this is the same one that exists with commercial VNP offerings. It is impossible for the user to verify claims of any company that guarantees anonymity/doesn't track/doesn't keep logs/etc.
> Dean.Bot, which could converse with voters in real-time through a website, was an early use of an emerging technology that researchers have said could cause significant harm to elections.

Goodness, imagine if people vote for the wrong candidates! How would democracy recover from that?

I can't really fault them for hard banning anything to do with politics given that "propaganda but tailored to the individual, is reasonably articulate, and it scales!" is one of the things that might earn this technology heavy restrictions and they don't want any more attention on them than they already have.

I'm surprised that a forum of technologists doesn't even attempt to see the somewhat accidental strings that allow society to run efficiently as somewhat trusting with a cheap bullshit detector. If you break that down everything grinds to a halt.

Ha! I’m sure they’re fine with anything that purely encourages voting. As long as the right folks are the ones being encouraged.
It's propaganda when we don't like it, and a civil duty when we do!
So they should just say "we're worried about our reputation" rather than say it's a threat to democracy. Why is push-polling not a "threat to democracy"? How about emotionally manipulative imagery in television ads? Or use of detailed demographic, geographic, and purchasing information for ad targeting?
Washingtonpost.com - "Democracy Dies Behind a Paywall"
It’s hard to believe ChatGPT would even work for this, unless this politician’s values and positions are identical to ones OpenAI have trained ChatGPT to have, the pretraining is going to leak through no matter what instructions they stuff into the prompt.
>unless this politician’s values and positions are identical to ones OpenAI have trained ChatGPT to have

That's not far-fetched if he's Sam Altman's preferred candidate.

I certainly wouldn't trust it to not say some seriously embarrassing stuff as people mess with the prompts.
Can we just cut out the middleman and vote for ChatGPT directly? I'd trust it more than any of the last few presidents.
Propaganda by the opposition - Bad. Propaganda by the leadership - Good.

Got it!

I believe we will see the day when we have an LLM model running for President. Thinking about it...maybe we should start with the Supreme Court...
> maybe we should start with the Supreme Court...

sounds like that's going to be more of a textualist than living document LLM ;-)

> I believe we will see the day when we have an LLM model running for President.

An LLM would probably win a landslide victory over its (human) opposition, if the voters don't know that they are voting for a robot.

We (the voters) could also have our own LLMs that the candidate LLM would have to convince to vote for them. We can totally abstract away the whole campaigning BS, just let the LLMs duke it out for us!
> LLM model running for President

It would need to be 35 years old and born in the USA.

Just use a proxy human to pass through the output from the LLM
I think Neal Stephenson’s vision in Interface is more likely. An AI augmented candidate that adjusts and responds to sentiment in near real time with the right signals coming in, trained to manipulate and appeal.
I can appreciate why companies would not want to be attached to politics in this way, or accused of influencing elections.

At the same time, as long as this person is consenting to this (which clearly seems to be the case), I don't think the developers did anything wrong (malum in se).

I suppose this is a use-case where people will turn to locally-run models.

Yes, it's apparently consensual and transparent:

> Dean.Bot didn’t all-out pretend to be Phillips himself; before engaging with Dean.Bot, website visitors would be shown a disclaimer describing the nature of the chatbot.

This is politics - attempting to treat specific situations reasonably is the fast path to a kerfuffle. Note that if Phillips' campaign thinks this would have been good for him, it might easily be perceived bad for someone else. And that is an unstable equilibrium that easily leads to political escalations using OpenAI technology.

Clear lines on what shan't be done (lines drawn far away from the real grey areas) is a position that is both prudent and respectable by OpenAI.

Yes, I agree. OpenAI is wise to take this course, and the campaign won't be wrong to work around it by using an open model (in my opinion).
It feels kind of weird to do this to me. Isn’t this what is often touted as free speech?
OpenAI is not a government entity.
Power corrupts. Who would have thought that descendants of freedom hippies will turn out to be petit tyrants!