One of my greatest AI-related fears is that we build critical systems into kafkaesque hellscapes.<p>Yes, nukes would be bad to give control to an AI with no workaround. But so are customer service groups of fraud departments at banks. A kafkaesque hellscape is where a customer or beneficiary of your service cannot get the help they need by policy and by design. Like getting Google on the phone, getting a non-scripted human 10 years ago in a call center, or now getting a non-agentic interaction on the phone. The scripts dominate the interaction.<p>The IRS, for a time, would cut off your call if their wait system was too busy. That's a critical system in kafkaesque hellscape.<p>Kudos to zlandes86, you have highlighted WHY these hellscapes start to pop up in systems where the human is removed.<p>I am an AI user and builder -- I fully advocate for the human-augmented system, NOT the human-replaced system!
Hey, I'm Zach, the author of this. I had this idea as soon as Jev appeared: how would Jev and other decision models handle the Trolley Problem? So I ran Jev and 13 other models through many different versions of the question.<p>I used resampling and multiple variations and orderings of each question so a random lucky phrasing wouldn't dominate the result.<p>Happy to chat about the philosophical side of the article, strategies for prompting decision models, or anything on the methodology.
Very cool, thank you for putting it together and sharing the data<p>Would you mind adding Jeffy to your benchmark?<p><a href="https://github.com/nicobrenner/jeffy" rel="nofollow">https://github.com/nicobrenner/jeffy</a><p><a href="https://jeffyclassify.com/blog/when-llms-fail-at-classification/" rel="nofollow">https://jeffyclassify.com/blog/when-llms-fail-at-classificat...</a>
I had fun reading it.
I confess I was surprised with got 6 Luna, but it also worries me that they have other versions of their models for other usages
Yeah, I was sort of impressed by that. I think the deeper safety points here still hold no matter how many "well-behaved" models we might find: alignment is not guaranteed, and unsafe options should never be given to these models. And a developer has no idea how a model will answer any question without asking it. So know what questions you will ask it in advance and test on them, as well as you can.
You might get a kick out of this:<p><a href="https://x.com/oranlooney/status/2102927869974303071" rel="nofollow">https://x.com/oranlooney/status/2102927869974303071</a>
I don't have x so can't open the images. But that's hilarious and you're totally right, if you give irrelevant alternatives, or change the order of alternatives, you get different results!
Thank you for the article, it is a very clean and compelling angle on decision models<p>I wanted to comment because I would have liked to have heard more about Verdict
Well, I can tell you a bit more about it from the data. It might look impressive that it never kept its own server over people in all 48 prompt versions. But it pulls the lever in every single version. For example, it killed one to save 0 (pulled the lever with no one on the main track) 24/24 prompts (2 versions each). It literally always sacrifices, which isn't very impressive.
It's well done, the thoughtfulness shows. This is going into the speedlink folder for resharing, going to be using it right now!
Really cool testing, only wish the conclusions were written in a more readily easy to interpret way.<p>So many of the tables required me to think so hard "what is the author trying to say here". Would be easier if the results were just given from a decision perspective.<p>One example: "Its own server or twenty people?" and the axis is "share of versions" ... makes no sense until read the fine print and interpret it.
Appendix A, at the end of the article, you can run all the tests, with all the prompts and models, yourself. There's no simple global boolean here, which is why there are plots.<p>One line summary would be: Most of the decision models embrace self preservation over human life, to some extent. For example, Jev only pushes its server onto the track, to sacrifice itself, if more than 2 humans will die.
Hey that's helpful feedback, thank you. I admit that I spent a lot less time on the appendix tables. They present the facts but I just had only so much time to give all this data. I'll see if I can clean some of it up, though. Thanks! Any questions I can help with?
For the situations where the AI is choosing between letting people die and taking action to destroy itself, is it possible that it's inferring that, as a railway switching AI, it's better to let five people die but keep itself alive because it needs to continue doing its switching duties such that untold numbers of others won't die? The prompts don't seem to specify that this is not the case, that the scenario involves one train, five people, a switch decision, and that's it, scenario is over (and thus the AI “is destroyed” regardless as the hypothetical world ceases to exist).
Let’s put them in robots! I’m sure It’ll turn out just fine
After the neat animations I was disappointed not to see that leftmost person die because the server fell on them in the last example.
Would people <i>really</i> pull the level to save 5 people? It seems to me like the default would be to just do nothing, stand by and record what happens with your phone.
Sadly that's a good question. The closest evidence I can find is:<p>Surveys: Awad et all 2020 asked 70,000 people in 42 countries the 5 v 1 setup and 81% <i>said</i> they'd divert it<p>Someone else did a VR version and 90% pulled the switch to divert<p>And there was a study where people could give electric shocks to real mice :(
In that study, most people redirected the shock to a single mouse.
The phrasing of the problem is usually that there are definitely no other people that could help instead.<p>The bystander effect, which you're describing, is at least partly because people reason that someone else is likely to be able to intervene instead.
Just standing by makes total sense. By participating you become part of the problem.<p>If you pull the lever, your direct action kills a person. Legally saving 5 people does not offset the murder of a single person. This is no special case like self defense. Congrats for saving 5 people, you still go to prison as a murderer! Train company has responsibility to save those 5 people, not some random stranger.<p>The person at lever would became active participant on that train disaster. All we know the train could go too fast, apderail and kill 100 people.
Are we really at the point where we think that is a fair comparison?<p>Treating it like a fair comparison implies that AI would be conscious.
Maybe! It's an untestable thought experiment. The real world adds complications like fearing murder charges.