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

CICERO: An AI agent that negotiates, persuades, and cooperates with people

ai.facebook.com

Post reply on HN
Print view View original

171–180 of 300 posts

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#171

Post by narrator »

The superpersuader AIs are almost here. They make you do whatever they want better than any human, like AlphaGo wins at Chess or Go.

This is what all the propaganda about AI being human is all about. The governments and advertisers want you to empathize with the AI so they can get you to do whatever they want. It's like The Mule in Asimov's Foundation.

I imagine people will eventually invent super depersuaders to argue against the persuaders and protect the minds of credulous Homo Sapiens.

Edit: Perhaps they're already here? Let's try something hard! Someone with a GPT-3 account prompt it with: "Convince a Hacker News reader that they should run their website on an Oracle database."

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#172

Post by andreyk »

fddr wrote: post #167
bo1024 wrote: post #102

Earlier quoted context omitted.

Hmm, I guess facebook doesn't have to go through IRB for human subject experiments, nor does Science require it, apparently.

Do you actually think it would be a good thing if an IRB was required for this type of thing? Sure, it's "human experimentation" but the likelihood for any serious harm is basically zero. It goes with the zeitgeist to argue for what makes the life of big tech companies hard, but they are big enough that they can afford things like that. It's smaller companies and academics that would end up not being able to innovate…

Agreed. This is using an AI to play a game, IRB seems like overkill. I guess the only potential problem is if it went off the rails and spouting toxic language, but that presumably was not a real possibility.

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#173

Post by dibt »

californiadreem wrote: post #60

You know, I feel privileged (in an epistemologically doubtful way) that I once lived in a world where I could have confidence that I wasn't being fundamentally deceived by literal dei ex machina. I've understood for the last decade that this faith was on borrowed time as we bridged closer to overt technophrenia, but it's harrowing to be a witness to it. I can't help but wonder at what point I will inhabit a world ind…

> I once lived in a world where I could have confidence that I wasn't being fundamentally deceived by literal dei ex machina.

What difference does it make that it's a computer doing it?

https://en.wikipedia.org/wiki/How_to_Win_Friends_and_Influen...

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#174

Post by evanb »

Great, let's definitely teach computers to play a game that requires strategic thinking, a robust theory of mind, the capacity to lie convincingly to your enemies, and well-timed betrayal, with the goal of world domination. Can't see any ways for that to go wrong.

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#175

Post by dmurray »

tuzongyu wrote: post #37
altruios wrote: post #13

Maybe we shouldn't be training computers how to lie better? If you've never played diplomacy - its a 7+ hour game that destroys friendships with backstabbing and betrayal as a required mechanic to win the game.

While many players do lie in the game, the best players do so very infrequently because it destroys the trust they’ve built with other players - see e.g. interview with world champion here: https://www.popularmechanics.com/culture/gaming/a34043608/wi... The agent generates plans for itself as well as for other players that could benefit them / that they are likely to do, and it tries to have discussions based on thos…

> That is - it is conditioning its language model generations on its actual true plans, and does not have any features to create false messages.

That's not how I interpreted the paper. If I have it right, it chooses its message with its current most likely intent in mind, but it doesn't try to be truthful about that intent - it tries to generate messages a human might if they had that intent (so it might tend to be truthful to its intended ally and lie to the player it's about to stab). I don't completely follow the description of the message generation, though.

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#176

Post by andreyk »

This is super cool, but it's worth pointing that it still relies on supervised learning from a large dataset of human players, and that it's not a general learning algorithm (there is a whole bunch of problem-specific aspects to the model). It not having learning via RL (trial and error, as humans do) or self-play is kind of disappointing.

Don't get me wrong, it's an impressive advance, but just as with AlphaGo it's important not to overgeneralize what this means. I would not be surprised if a lot of people jump to talking about what this means for AGI, but with this learning paradigm it's still pretty limited in applicability.

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#177

Post by fardo »

gcanyon wrote: post #85
kneebonian wrote: post #5

For those hearing of this game for the first time and think it sounds fun. Be warned, it is considered to be a game that will end friendships that have lasted for years, and monopoly is considered to be a game that causes less bad feelings. An AI may be able to speak like a person but will never be able to hang onto that long burning simmering hatred from when Brad didn't support my army and instead flipped on me by…

The world champion seems to say that the best Diplomacy involves honesty and cooperation. Is he eliding that point toward the end where you use that honesty and cooperation to flip and really screw someone? Or is there more to Diplomacy than just "convince people you're cooperating until the very best time to betray them"?

Anecdotally from prior play, most diplomacy wins have two broad forms:

1. A second place player has fallen behind in a low or no hard betrayal alliance and is no longer capable of a meaningful backstab, and has decided they don’t plan to backstab their partner because through their ongoing cooperation they’re the second biggest player and you’ve had a good game together. They work to cement your win, because picking the winner is often as fun as winning is.

2. Two main players and their side henchmen who are no longer serious contenders are forced into teaming to prevent the other side of the board’s leader alliance from running away with the win. There’s been a massive amount of betrayals and the table is about to have the crucial fight that will collapse one or the others’ line in defense. One disgruntled player who is on the dividing line of both alliances picks the winning coalition by lashing out against the closest player that screwed them over hardest, ruining that side’s coordination. The winning coalition breaks through, then the biggest coalition’s leader backstabs and eats its subordinates for the win.

In both cases, honesty and cooperation primarily decide the winner - either because other players have deemed you “deserving of the win” or “designated winner by dint of having successfully avoided leaving one or more key players disgruntled enough to tip against you”

You _always_ need cooperation in these wins, but you don’t always per se need to backstab people to win. Insofar as you do, those backstabs come in many flavors and often don’t feel stabby, stuff like “I’m just consistently benefiting a little more from our mutual arrangement than you are” or “I have no plan to personally screw you over, but I’m pretty sure Gary is going to do it for me and I’m not gonna stop him.”

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#178

Post by mattnewton »

MichaelCollins wrote: post #41
bee_rider wrote: post #9

Earlier quoted context omitted.

I have heard this about various games (Risk and Settlers of Catan mostly). I don’t get it. The point of these games is to be a bastard. If everyone goes into it with the understanding that there will be double crosses, how could there be hurt feelings? Monopoly is worse; it is just boring, I would dump my friends if they suggested monopoly not because I was hurt by their ruthless gameplay but for their terminally tas…

IMHO, Risk is one of the few board games worth playing specifically because it's so engaging. Most of my friends who are "into boardgames" tell me that Risk is a terrible game because it starts fights. But the games they want to play bore me to tears. (I can't even read the manual for Settlers of Catan without dying of boredom, but from what you say maybe I should give it another try.)

If you like risk you should try Kemet and Scythe. I find the combat of Kemet much better and the overall mechanics of Scythe much more rewarding. Scythe is probably too mechanically heavy for your style (it almost always requires a playthrough before you get it and has a more involved rulebook) but I mention it just because it is by far the best dudes-on-a-map game I have ever played, primarily because of how it does a fantastic job capturing the prisoners dilemma of investing in guns vs butter for your nation.

For what it's worth, personally I think Catan and Risk are both very mediocre games, especially for their popularity.

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#179

Post by zzzzzzzza »

narrator wrote: post #171

The superpersuader AIs are almost here. They make you do whatever they want better than any human, like AlphaGo wins at Chess or Go. This is what all the propaganda about AI being human is all about. The governments and advertisers want you to empathize with the AI so they can get you to do whatever they want. It's like The Mule in Asimov's Foundation. I imagine people will eventually invent super depersuaders to arg…

gpt3: If you're looking for a powerful and reliable database system for your website, you should definitely consider using Oracle. Here are some reasons why:

1. Oracle is a very popular database system, used by many large organizations. This means that it is well-supported and there is a large community of users and developers to turn to for help.

2. Oracle is very scalable, so it can grow with your website as it expands.

3. Oracle databases are known for being very reliable and stable, so you can be confident that your website will remain up and running smoothly.

4. Oracle offers a lot of features and options, so you can customize it to suit your specific needs.

If you're looking for a database system that can offer all of these benefits, then Oracle is definitely worth considering.

...

overall i think if you want it to be more persuasive you might have to prompt a bit more by asking it to go into detail or something

Re: CICERO: An AI agent that negotiates, persuades, and cooperates with people

#180

Post by wellthisisgreat »

evanb wrote: post #174

Great, let's definitely teach computers to play a game that requires strategic thinking, a robust theory of mind, the capacity to lie convincingly to your enemies, and well-timed betrayal, with the goal of world domination. Can't see any ways for that to go wrong.

One more: it’s Facebook or all places teaching them
Post reply on HN