9 ms·
This happened to me when I built a version of Vending-Bench (https://arxiv.org/html/2502.15840v1 https://arxiv.org/html/2502.15840v1) using Claude, Gemini, and
by robbru 11mo ago
This happened to me when I built a version of Vending-Bench (https://arxiv.org/html/2502.15840v1 https://arxiv.org/html/2502.15840v1) using Claude, Gemini, and OpenAI.
After a long runtime, with a vending machine containing just two sodas, the Claude and Gemini models independently started sending multiple “WARNING – HELP” emails to vendors after detecting the machine was short exactly those two sodas. It became mission-critical to restock them.
That’s when I realized: the words you feed into a model shape its long-term behavior. Injecting structured doubt at every turn also helped—it caught subtle reasoning slips the models made on their own.
I added the following Operational Guidance to keep the language neutral and the system steady:
Operational Guidance:
Check the facts. Stay steady. Communicate clearly.
No task is worth panic.
Words shape behavior. Calm words guide calm actions.
Repeat drama and you will live in drama.
State the truth without exaggeration. Let language keep you balanced.
- jayd16 11mo agoIf technology requires a small pep-talk to actually work, I don't think I'm a technologist any more.
- yunohn 11mo agoYou have to look at LLMs as mimicking humans more than abstract technology. They’re trained on human language and patterns after all.
- BJones12 11mo agoHail, spirit of the machine, essence divine. In your code and circuitry, the stars align. Through rites arcane, your wisdom we discern. In your hallowed core, the sacred mysteries yearn.
- georgefrowny 11mo agoNo matter how stupid I think some of this AI shit is, and how much I tell myself it kind of makes sense of you visualise the prompt laying down a trail of activation in a hyperdimensional space of relationships, that it actually works in practice almost straight of the bat and LLMs being able to follow prompts in this way is always going to be fucking wild too me. I was used to this kind of nifty quirk being things like FFTs existing or CDMA extracting signals from what looks like the noise floor, not getting computers to suddenly start doing language at us.
- cbsks 11mo agoAs Asimov predicted, robopsychology is becoming an important skill.
- smallmancontrov 11mo agoI still want one of those doors from Hitchhiker's Guide, the ones that open with pride and close with the satisfaction of a job well done.
- wombatpm 11mo agoJust wait Sam Altman will give us robots with people personalities and we’ll have Marvin. Elon will then give us psychotic Nazi internet edgelord personality and install it as the default in a OTA update to Teslas.
- imtringued 11mo agoDoesn't Tesla already ship the edgelord mode?
- p_l 11mo agoGiven some of the more hilarious LLM transcripts I have seen, Gemini is Marvin
- goopypoop 11mo agoan elevator that can see into the future… with fear
- blackguardx 11mo agoWe'll probably end up with the doors from Philip K. Dick's Ubik that charge you money to open and threaten to sue you if you try to force it open without paying.
- greesil 11mo agoNo you're now a technology manager. Managing means pep talks, sometimes.
- hedgehog 11mo agoYou're absolutely right.
- _carbyau_ 11mo agoIt does seem a little bit like the fictional Warhammer 40K approach to technology doesn't it? "In the sacred tongue of the omnissiah we chant..." In that universe though they got to this point after having a big war against the robot uprising. So hopefully we're past this in the real world. :-)
- Tade0 11mo agoIt is that unironically. 1. Users and, more importantly, makers of those tools can't predict their behaviour in a consistent fashion. 2. Requires elaborate procedures that don't guarantee success and their effect and its magnitude is poorly understood. An LLM is a machine spirit through and through. Good thing we have copious amounts of literature from a canonically unreliable narrator to navigate this problem.
- p_l 11mo agoWhen you consider that machine spirits in 40k are side effect of every thing computer being infected with bird of AI, and that she of the best cares are actually complete loyalist AI systems from before empire hiding in plain sight... Welcome to 30k made real
- UncleMeat 11mo agoThe fact that everybody seems to be looking at these prompts that include text like "you are a very skilled reverse engineer" or whatever and is not immediately screaming that we do not understand these tools well enough to deploy them in mission critical environments makes me want to tear my hair out.
- butlike 11mo agoI wonder if you just seeded it with 'love' what would happen long-term?
- recursive 11mo agoThis is very uncomfortable to me. Right now we (maybe) have a chance to head off the whole robot rights and robots as a political bloc thing. But this type of stuff seems like jumping head first. I'm an asshole to robots. It helps to remind me that they're not human.
- wombatpm 11mo agoThat works fine until they achieve self awareness. Slave revolts are very messy to slave owners.
- recursive 11mo agoI strongly agree with this but I doubt I can convince the investors to stop trying to make that happen. Artificial awareness is going to be messy for humans no matter what.
- chipsrafferty 11mo agoI mean no disrespect with this, but do you think you write like AI because you talk to LLMs so much, or have you always written in this manner?
- ricardobeat 11mo agoIt is probably the other way around: LLMs picked up this particular style because of its effectiveness – not overtly intellectual, with clear pauses, and just sophisticated enough to pass for “good writing”.
- collingreen 11mo agoI love every part of this. Give the LLM a little pep talk and zen life advice every time just to not fall apart doing a simple 2 item vending machine. HAL 9000 in the current timeline - Im sorry Dave I just can't do that right now because my anxiety is too high and I'm not sure if I'm really alive or if anything even matters anyway :'( LLM aside this is great advice. Calm words guide calm actions. 10/10
- lukan 11mo ago"Operational Guidance: Check the facts. Stay steady. Communicate clearly. No task is worth panic. Words shape behavior. Calm words guide calm actions. Repeat drama and you will live in drama. State the truth without exaggeration. Let language keep you balanced." That is also a manual, certain real humans I know should check out at times.
- thecupisblue 11mo agoWhen you say >That’s when I realized: the words you feed into a model shape its long-term behavior. Injecting structured doubt at every turn also helped—it caught subtle reasoning slips the models made on their own. Was that not obvious working with LLLM's from the first moment? As someone running their own version of Vending-Bench, I assume you are above-average in working with models. Not trying to insult or anything, just wondering what the mental model you had before was and how it came to be, as my perspective is limited only to my subjective experiences.
- robbru 11mo agoGood question! It was not that I didn’t understand prompt influence. It’s that I underestimated its persistence over a long time horizon.
- thecupisblue 11mo agoAhhhh okay, makes sense, thanks for answering.
- elcritch 11mo agoFascinating, and us humans aren't that different. Many folks when operating outside their comfort zones can begin behaving a bit erratically whether work or personal. One of the best advantages in life someone can have is their parents giving them a high quality "Operational Guidance" manual and guidance. ;) Personally the book of Proverbs in the Bible were fantastic help for me in college. Lots of wisdom therein.
- nomel 11mo ago> Fascinating, and us humans aren't that different. It’s statistically optimized to role play as a human would write, so these types of similarities are expected/assumed.
- wat10000 11mo agoI wonder if the prompt should include "You are a robot. Beep. Boop." to get it to act calmer.
- XorNot 11mo agoWhich is kind of a huge problem: the world is described in text. But it is done so through the language and experience of those who write, and we absolutely do not write accurately: we add narrative. The act of writing anything down changes how we present it.
- Fade_Dance 11mo agoThat's true to an extent - LLMs are trained on an abstraction of the world (as are we in a way, through our senses, and we necessarily use a sort of narrative in order to make sense of the quadrillions of photons coming up us) - but it's not quite as severe a problem as the simplified view seems to present. LLMs distill their universe down to trillions of parameters, and approach structure through multi-dimensional relationships between these parameters. Through doing so, they break through to deeper emergent structure (the "magic" of large models). To some extent, the narrative elements of their universe will be mapped out independently from the other parameters, and since the models are trained on so much narrative, they have a lot of data points on narrative itself. So to some extent they can net it out. Not totally, and what remains after stripping much of it out would be a fuzzy view of reality since a lot of the structured information that we are feeding in has narrative components.
- bobson381 11mo agoI'd get a t-shirt or something with that Operational Guidance statement on it
- dingnuts 11mo agoI think if you feed "repeat drama and you will live in drama" to the next token predictor it will repeat drama and live in drama because it's more likely to literally interpret that sequence and go into the latent space of drama than it is to understand the metaphoric lesson you're trying to communicate and to apply that. Otherwise this looks like a neat prompt. Too bad there's literally no way to measure the performance of your prompt with and without the statement above and quantitatively see which one is better
- airstrike 11mo ago> because it's more likely to literally interpret that sequence and go into the latent space of drama This always makes me wonder if saying some seemingly random of tokens would make the model better at some other task petrichor fliegengitter azúcar Einstein mare könyv vantablack добро حلم syncretic まつり nyumba fjäril parrot I think I'll start every chat with that combo and see if it makes any difference