[minimaxir]

Gambling addition implies dopamine hits from irregular and uncertain outcomes: as I note in the post, I don't get a dopamine hit from running agents. I know people say "LLMs are just gambling because they're next-token-maximizers that can't write real code" but with GPT 5.6 Sol (and a few new tricks I discovered) the outputs are much less irregular and uncertain. It's just typical engineering.

"Not wanting to waste money" is the polar opposite of gambling.

[dlenski]

> Gambling addition implies dopamine hits from irregular and uncertain outcomes

Your post literally describes your fascination with trying to figure out the pattern of a "random" reward that you get, and trying to maximize the value you get out of it.

I put "random" in scare quotes because I strongly believe that—just as slot machine payouts are carefully structured to keep you playing—these LLM resets are structured to keep heavy users like you coming back to max out their usage, and to progressively upgrade it.

Several other commenters have also stated this same suspicion about the pattern of resets you're describing.

> "Not wanting to waste money" is the polar opposite of gambling.

From everything I've read about gambling addiction, particularly Jay Caspian Kang, that seems wrong.

The desire to "not waste money" and "get back to even" seems like a huge part of what motivates gamblers to keep gambling.

[drekipus]

>The desire to "not waste money" and "get back to even" seems like a huge part of what motivates gamblers to keep gambling.

As someone who had family members go through gambling addiction this is the primary mechanism behind it.

Addicts don't see it as "cool fun dopamine kicks" but instead find it the only way they can get back to normal/where they are supposed to be

[therealpygon]

Don’t worry, they have a system, they can’t lose, and honestly, it’s like the outcome is almost guaranteed.

[win311fwg]

That seems like an incomplete explanation.

Logically, gambling is like going to the movies. You expect to pay x currency for y value of entertainment. If y falls short of expectations you might feel like you wasted your money, but who becomes addicted to going to the movies to try to get even? There is probably someone who has, but I’ve never heard of it and it doesn’t seem to be common; not like gambling addictions. For all intents and purposes it doesn’t happen.

But gambling addictions do happen, fairly regularly. Perhaps it is loss aversion coupled with the aforementioned dopamine hit associated with gambling that makes it so prevalent?

[ifwinterco]

From what I've heard from gambling addicts (thankfully this is not something I'm prone to), there is something about the actual process of gambling that makes it addictive beyond the loss/reward system.

People who are really into roulette get a buzz off seeing the wheel spin, hearing the ball bouncing etc.

Probably that comes after the initial addiction to the reward function but it then strongly reinforces it.

I don't think there's an equivalent for paying to watch a movie because the time to payoff (or not) is too long and the sensory experience is too inconsistent to elicit a conditioned place preference.

LLMs on the other hand... the time to payoff is shorter and the experience is consistent every time. It's just lacking the tactile/sensory elements

[dlenski]

> Logically, gambling is like going to the movies. You expect to pay x currency for y value of entertainment.

I don't think gambling is at all like paying a set price for a ticket and having a pretty good idea of how long the entertainment will last. If "gambling" means making a series of short-term bets for entertainment value, you don't have any clear idea how long you'll be entertained for or how much it will cost.

People will show up at a casino with let's say, $200 and a debit card, and expect that they'll be able to spend $50, be entertained for 2 hours, and then just leave… while secretly hoping that they'll actually leave with more than they came with.

Then they burn through their $50 in half an hour, and dip into their remaining stash in order to keep playing. OR they triple their money in half an hour and feel such a rush that they want to keep playing with more money. Then, repeat repeat repeat.

[win311fwg]

> I don't think gambling is at all like paying a set price for a ticket and having a pretty good idea of how long the entertainment will last.

Generally you can gain a pretty good understanding of how long the entertainment will last. Maybe not down to the millisecond, but you know a spin on the one arm bandit won't take hours. It will give on the order of seconds. Your willingness to give up x is contingent on the perceived entertainment value of those seconds. If you choose to play again, that is an independent event — like deciding to watch a second movie while you are already at the theatre.

> while secretly hoping that they'll actually leave with more than they came with.

Yes, this may be the undefined variable. The dopamine hit of believing you can come home better than you started, with little tastes of the possibility, coupled with loss aversion when it isn't being realized. This is what earlier comments seem to be speaking about, so perhaps, despite your insistence, a consensus was already reached.

[somewhereoutth]

> and a few new tricks I discovered

This right here. Any gambler would recognize that statement.

[minimaxir]

Said tricks improve the output in an objective measurable manner, not theoretical, vibes, or gambler's fallacy. (blog post forthcoming on that)

I've been researching LLM prompt optimization for longer than ChatGPT has existed; I was successfully optimizing the output of GPT-2 back in 2019.

[chaps]

Respectfully, a lot of what you're saying in this thread sounds a lot like the lies that gamblers tell themselves. Saying this as someone with a strong tendency towards addictions.

Some of these things are only possible to really see in hindsight. Yes, you've been working on these things for a while, but these systems are notably different in their capacity and strings they pull on us.

Be well, please.

[minimaxir]

Yesterday (unrelated to quotamaxxing described in the article), I made an Apple TV macOS menu bar remote app: https://cdn.bsky.app/img/feed_fullsize/plain/did:plc:oxaerni...

Every single prompt worked without issue, and it got most of the way on the first try with the initial prompt (+ a couple visibility bugs due to the agent not having Computer Vision to see said menu bar app) such as:

> Create a SwiftUI menu bar app named `swiftmote` using theto create the most user friendly app following Apple's HID guidelines for creating a remote that can operate a Apple TV on a local network. Instead of reimplementing the protocols needs to interface with an Apple TV, use the Python package `pyatv` and host it within the SwiftUI app as a sidecar along with a Python installation.

I have my own Apple TV I can manually verify that it worked as expected, which is notable because the agent can't test or lie about this pipeline because it does not have access to the Apple TV.

That is not hallucination or psychosis. If you want, I can release all the prompts I used. (EDIT: Sure, why not, here are the prompts. If I don't complain about something in a followup prompt, assume it worked correctly: https://gist.github.com/minimaxir/30fa820daa1392da13026ec6aa... )

[chaps]

You're, almost like clockwork, responding to the wrong thing. Not contesting what's been made or what hasn't or what's possible or what's not. That's not the issue I'm bringing up.

Just -- do well for yourself. Deflection aint it.

[minimaxir]

You accused me of "lies I'm telling myself" about the robustness of current LLMs and I take offense to that when I have hard evidence that it's not a lie I'm being manipulated by. That's not deflection and I'm not sure what else you think I should be responding to.

[rcxdude]

A lot of this really seems to be pattern matching on superficial similarities when the meat of the issue is probably the more important one. I think for LLM usage to be compared to gambling, it has to be a) universally negative-sum in the long run, and b) extremely addictive. a) seems very unlikely to be true, even if it is sometimes negative. b) seems to be the case for some subset of programmers, but again I would not assume this just because someone uses some statements that are sometimes uttered falsely by gamblers (the siren song of those statements being that if they were true, the gambling would be a good thing. That means the truth of those statements matters and should not be assumed to be false like they are more obviously so for gambling).

[somewhereoutth]

the key comparison is variable reward.

That 'triggers a surge of dopamine and creates highly addictive habits' [thanks Gemini!]

LLM use for code generation does exactly that, sometimes it works amazingly, sometimes it fails inexplicably. Whether it is negative sum or not doesn't really matter. Indeed it may well prove to be negative sum, especially if we step back a bit and consider the business benefit of the code produced, not just lines of code or even features produced.

[rcxdude]

I think whether it's negative sum or not matters quite a lot, really. The variability of the outcome is a thing that might create some addictive components, but it matters a lot whether it's something with positive effects with some potential negative effects that may need to be managed, or whether it's the same as gambling, and I think it's extremely unlikely that it is the same as gambling.

[dehrmann]

> dopamine hits from irregular and uncertain outcomes

Like the LLM getting the solution right?

[minimaxir]

For the problems I work on with GPT 5.6 Sol and the checks and balances I have in place, I estimate:

- 80% of prompts get everything correct and are confirmed correct with manual validation

- 19% of prompts make a minor mistake based on an ambiguity of the original prompt (user error not LLM error), but then reliably fixed in a followup prompt

- 1% of prompts causes more problems than it solves and is more pragmatic to just revert

For 99% good output, there isn't much of a dopamine rush when there is good output. The dopamine rushes are for the <1% odds.

From the other replies on this post, I suspect no one believes me, but I am offering these numbers in good faith.

[SyneRyder]

I think those of us who are using AI consistently believe you and understand. I'd say roughly the same thing about Claude in terms of numbers.

I think many people who don't believe you just haven't built-up the kind of prompt history & MCP / CLI tooling etc that lets you get to the point where things work at that level of accuracy.

Hope it helps to know that at least some of us here understand and are seeing the same thing. And if it's anything like my experience with Fable, "always be more ambitious". The capabilities of the models are often limited only by what you're brave enough to ask for. I keep finding I'm not ambitious enough.

[epitrochoid413]

Yes I agree with you here.

LLMs tend to be a bit random, but are still more consistent and predictable than slot-machines. Also, gambling tends to exert lower effort and higher dopamine hits than vibe-coding, making it way more addictive.

But LLMs are still addictive to some extent. Maybe its around the same level as other behavioral addictions like food, social media, or gaming addictions.