Universal Paperclips
decisionproblem.com
decisionproblem.com
https://en.wikipedia.org/wiki/Instrumental_convergence#Paper...
https://en.wikipedia.org/wiki/Autofac
https://www.vulture.com/2018/01/electric-dreams-recap-season...
This isn’t Jurassic Park.
LLMs are pretty basic stuff but we are all struggling with what to use them for!
OpenAI is manually playing whac-a-mole with ChatGPT saying the darndest things!
It's also hard to understand why an AI smart enough to paperclip the world wouldn't also be smart enough to realize the futility in doing so. So while alignment remains an issue, the existential alignment threats are too ill-specified. AGIs would understand we don't want to paperclip the world.
Fun game though.
Terminal goals can't be futile, since they do not serve to achieve other (instrumental) goals. Compare: Humans like to have protected sex, watch movies, eat ice cream, even though these activities might be called "futile" or "useless" (by someone who doesn't have those goals) as they don't serve any further purpose. But criticizing terminal goals for not being instrumentally useful is a category error. For a paperclipper, us having sex would seem just as futile as creating paperclips seems to us. Increased intelligence won't let you abandon any of your terminal goals, since they do not depend on your intelligence, unlike instrumental goals.
Of course the premise becomes that the AI has been instructed to make paperclips. They should have hired a better prompt engineer, capable of actually specifying the goals more clearly. I don't think an AI that eradicates humankind, will have such simplistic goals, if an AI ever becomes the end of humans. Cybermen, though, are inevitable.
Even if they did, what if they aren't smart enough for eloquent humans to convince them it's for the greater good. True AGIs will need a moral code to match their intelligence, and someone will have to decide what's good and bad to make that moral code.
"Futility" is subjective, and the whole purpose of the thought experiment is to point out that our predication of "futility" or really any other purely mental construct does not become automatically inherited by a mind we create. These imaginary arbitrarily powerful AIs would definitely be able to model a human being describing something as futile. Whether or not it persues that objective has nothing to do with it understanding what we do or don't want.
See also big tobacco. They exactly what their additives to the product did.
See also 3M and PFAS. See also Big Oil. See also, see also...
Why would I expect anything different from any other branch of business using the precedence laid before us?
Eichmann.
For example, perhaps the lesser-evil argument played a role in the decision process: would a world where deep fakes are ubiquitous and well-known by the public be better than a world where deep fakes have a potent impact because they are generated seldomly and strategically by a handful of (nefarious) state sponsors?
if we build ai AND THEN we give it a stupid goal to optimize AND THEN we give it unlimited control over its environment, something bad will happen.
the conclusion is always "building AI is wrong" and not "giving AI unrestricted control of critical systems is wrong"
Replace the word "we" with "a psychotic group of terrorists" in your post and see how it reads.
Super easy to not think about something if your job depends on it. And even if you do, things don't go as you think (see bombings of civilian Hiroshima and Nagasaki despite objections of nuclear physicists).
Some others entered the field, made progress, and apparently regretted it.
Others are willing to put their concerns aside for money. Salaries get very high in that field.
Sci-Fi Author: In my book I invented the Torment Nexus as a cautionary tale
Tech Company: At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus
It is nothing close to being an actual intelligence, regardless of how much we anthropomorphize it. We also anthropomorphize stick figures, stuffed animals, and weighted companion cubes.
How is this different from maximizing paperclips? It's the same thing, just with a much more direct basis for instrumental convergence!
Ok, so it's a more general example of worries about "managing superintelligence" but IMO it does the debate a disservice by being so obviously ludicrous that it's hard to square "naive paperclip-maximizing AI" with "superintelligence."
I think if we're going to survive all this stuff it's much more likely to be because the private parties with the wherewithal to unleash an AI with the ability to affect the world to that extent will largely be ones with enough resources to have narrow banal goals and narrow banal constraints including self-preservation too, not because we figure out some sort of general purpose "AGIs that are aligned with humans" solution.
Kinda like with nukes.
It's too intelligent to restrict our constraints.
but
It's not too intelligent to be "aligned" to underlying intentions and values?
That approach doesn't even work on humans, why would it work on a superintelligence?
The thing is, being aligned cannot be solved with intelligence per se.
Say, you are (far) more intelligent than a spider. There's no way you can get aligned with (all of) its values unless the spider finds a way to let you know (all of) its values. Maybe the spider just tells you to make plenty of webs without knowing that it might get entangled in them by itself. The webs are analogous to the paperclips.
Intelligence makes that harder, not easier. Just because it can work out the underlying intentions doesn't mean it cares about them. Remember, this is an optimisation process that maximises a function; deciding to do something that doesn't maximise that function will not be selected for
Whenever you do something, do you think "the underlying goal of my behaviour set by evolution is to reproduce and have children, so I'd better make sure my actions are aligned to the goal of doing that"? No, you don't care what the underlying "intentions" are, and neither does an AI. Because our environment has changed, many of our instincts no longer line up well with that goal, which is actually another problem with aligning AI because it can do the same thing if the environment changes since training as the training process can create an AI with goals that aren't exactly the same as the training goal but line up well during training
If you're saying such an AI would be too smart to be a simple paperclip maximizer, then I'd agree, but then what's the point of the thought experiment if a paperclip maximizer is impossible.
The first is that these constraints aren’t easy. Make paperclips in a way that doesn’t hurt anyone. Ok, so it’s going to make sure every single part is ethically sourced from a company that never causes any harm to come to anyone ever, and doesn’t give any money to people or companies that do? That doesn’t exist. So you put in a few caveats and those aren’t exactly easy to get right.
The second part is an any versus all issue. Even if you get this right in any one case, that’s not enough. We have to get this right in all cases. So even if you can come up with an idea to make an ethical super intelligence, do you have an idea to make all super intelligences act ethically?
I actually believe in the general premise of this question as being the biggest threat to humans. I don’t think it’s a doomsday bot that gets us. It’s going to be someone trying to hit a KPI, and they’ll make a super intelligence that demolishes us like a construction site over an anthill.
What constraints do you suggest? If it's just changing "make as many paperclips as possible" to "make at least x number of paperclips" (putting a cap on the reward it gets), here's a good explanation of why that doesn't really work: https://www.youtube.com/watch?v=Ao4jwLwT36M
If you're suggesting limiting the types of actions it can take, then to do that to the point that a superintelligence can't find a way around it (maybe letting it choose between one of two options and then shutting it down and never using it again) would make it not very useful, so you'd be better off just not making it at all
> If you're saying such an AI would be too smart to be a simple paperclip maximizer
No, that's not what I'm saying. Any goal is compatible with any level of intelligence, there is no reason why it wouldn't be possible to follow a simple goal in a complex way. Again here's a video about that: https://www.youtube.com/watch?v=hEUO6pjwFOo
Second, if you are able to set a goal then during this setting you can set many constraints, even fundamental ones. There is no reason the goal is more fundamental than the constraint. If I approve, make paperclips. Efficiently make 100 paperclips.
It's the duality of being able to set a rule but not being able to set a constraint that I find a strange concept. I lean towards the picture of not being able to set goals nor constraints at all.
You can set constraints just fine. It’s simply a part of the goal: “do x without doing y”. It’s just really hard to find the right constraints, no simple one works.
For example “if I approve, make paperclips” - so it gets more reward if you approve? What’s to stop it from manipulating you into thinking nothing is wrong so you always approve? “Efficiently make 100 paperclips.” I already linked a video on why capping the reward like that doesn’t work, but if you don’t want to watch it the gist is that for your suggestion it may just make a maximiser which is pretty guaranteed to make at least 100, and is pretty efficient because it’s not doing much work itself. Then the maximiser kills us all
I don't think there's any industry on earth that doesn't need humans in the loop somehow. Whether is mining raw material from the ground, loading stuff in machines for processing, and most importantly fixing broken down machines, robots are really bad at these things for the foreseeable future.
Not to mention AI needs constant electricity, which is really complicated and requires humans fixing a lot of stuff.
Also it’s a simplified example, it wouldn’t literally be paperclips but some other arbitrary goal (it shows how most goals takes to their absolute extreme won’t be compatible with human existence, even something that sounds harmless like making paperclips)
A human who wanted to "build as many paperclips as possible" could cause a great deal of destruction today.
A human who wanted to accumulate as much wealth as possible could, too.
EDIT: maybe a better way of articulating my complaints about this famous thought experiment is that it's supposed to be making a point about superintelligence but it's talking about a goal that has sub-human-intelligence sophistication.
The "taken to the absolute extreme" part.
> A human who wanted to "build as many paperclips as possible" could cause a great deal of destruction today.
Maybe, but a) no one really wants that (at least not as their only desire above all else) and b) we aren't superintelligent so it's hard to gain enough control and power and plan well enough to do it that well
> talking about a goal that has sub-human-intelligence sophistication
There is no reason a simple goal can't be followed in an intelligent way or vice versa. This is called the "orthogonality thesis". There's a good video about it here: https://www.youtube.com/watch?v=hEUO6pjwFOo
this nightmare scenario really only plays out if paperclip machine develops some sort of self-preservation instinct and has the means to defend/protect itself from being disabled. Building a machine capable of that seems a) like fantastical scifi and b) easily preventable.
You could turn this into a "this is why superintelligence will good" thought experiment, though! Maybe "the superintelligence realizes that optimizing for these short term metrics will harm the company's position 30 years from now in a way that isn't worth it" - the superintelligence is smart enough to be longtermist ;) .
I realize that the greater point is supposed to be more like "this agent will be so different that we can't anticipate what it will be weighing or not, and whether it's longterm view would align with ours", but the paperclip maximizer example just requires it to be dumb in a way that I don't find consistent with the concern. And I find myself similarily unconvinced at many other points along the chain of reasoning that leads to the conclusion that this should be a huge immediate worry or priority for us, instead of focusing on human incentives/systems/goals.
Of course the paperclip company can try to add constraints to their AI in order to prevent naive paperclip maximization, but what if they screw up those constraints as well? The whole premise of Asimov’s Three Laws is that AI has these sorts of constraints, but even in his stories these constraints still lead to unexpected outcomes.
All programming bugs are the result of a programmer encoding an instruction or statement that doesn’t imply what they think it implies and the computer following it literally. A more capable and autonomous computer that approaches what we might call “intelligence” is also going to be more capable of doing harm when it runs into a bug. And if it’s something like an LLM where the instructions are natural language, with all its ambiguity and vagueness, you have a whole other issue compounding it.
If you study philosophy you end up running into the exact same problem. The object of the game of philosophy is to make the most general true statements possible. One philosopher might say something like, “knowledge is defined as true justified belief”, or maybe “moral good is defined as whatever delivers the greatest good to the greatest number”, or maybe even, “the object of the game of philosophy is to make the most general true statements possible”. And then another philosopher comes up with a counterexample or counterargument which disproves the first philosopher’s statement, usually because—just like a programming bug—it entails an implication that the first philosopher didn’t think of. We have been playing the game of philosophy for thousands of years and nobody has managed to score a point yet.
Another thing. Human beings have a lot of needs, imperatives, motivations, and values. Some of them, like food, are built in. Others are learned through culture. But we end up with a lot of them, and it’s easy to take them for granted. With a machine, you have to build those things in yourself. There’s no getting around it. But we don’t actually have a complete, hierarchical set of imperatives/motivations/values for a decent human being. The philosophers have been working on it for millennia but keep running into bugs. So how can we expect to solve the problem for non-human AI? True, we are unlikely to screw up so badly that we end up with a literal paperclip maximizer, but we are bound to make some far more subtle mistake of the same general kind.
It's interesting to transfer the idea of the Turing Test onto other "agent" scenarios.
Financial trading bots have been a thing for a long time without any need to pretend that they're human.
The legitimation of property and capital depends on human owners though.
The writing and the story is pretty damn good though.
It's an absolute grind, but once you pick up some 500% productivity multipliers it gets much easier
Honestly kinda ruined it, I don't know if I can play the game again knowing I lost a couple hundred hours of progress
- Kittens: https://kittensgame.com/web/
- Evolve: https://pmotschmann.github.io/Evolve/
There are plenty of really crappy clickers that don't do anything for me. And there are some that are too complex to be really fun (Kittens), but the ones in the middle, like this one, really feel good.
I would put Universal Paperclips and A Dark Room as exceptions though in the sense that they're still fully engrossing, but there's a little bit more depth and discovery than just "click the thing until you have enough clicks to get the next thing".
It's about as barebones as you can get - it's literally just "numbers go up" with a few simple names attached to them. It turns out that even that is enough to get me addicted for a bit.
Universal Paperclips was a latecomer to the clicker genre. It starts off making you think its a simple clicker game, but then it turns out that Universal Paperclips has an ending.
Once you achieve the ending, you then have permission to turn off the Universal Paperclips (in fact, its an explicit option), and it installs a cookie or something that prevents the game from starting up again.
Its this "anti-clicker" mindset, despite looking like a simple clicker game, that makes me... ironically... come back to Universal Paperclips over the years.
Mind you, I'd have liked a lite version that reduced the tech tree by maybe a third but still enabled you to get all the advancements without devoting your whole life to it ;o)
I wonder what other examples there are?
I thought I had won the game with full automation.
There is an entire new level.
12 hours later.
Let it run in background before figuring out to add points to 'explore'.
Best game ever.
Then I tried again, with a Javascript console timer 'clicking' 10 times per second. Let it go overnight. Got to 200,000 clips or somesuch and ... it was again boring.
Use the same timer technique on the quantum computer to get the stock market. Didn't see why I cared to make more money on the market since apparently all I could spend it on was marketing.
Messed up with a timer on the quantum computer, ended up with -87,554 operations and growing (shrinking?). That was enough. The "quantum temporal reversal" at -10,000 operations is a nice touch, but with "revert to the beginning" and a screwed up timer I gave up.
So, the way to not get addicted is to deliberately play it wrong. ;)
> Welcome to Universal Paperclips|
Universe: 49 / Sim Level: 5
Paperclips: 183,735,416
Make Paperclip
Specifics sent to dev:
>I progressed to world level 2. At the beginning I did not activate the new artifact but instead went back to world level 1 via the map then went back to world level 2 via the map. This process removed the world level 2 alien artifact. If this was by design it would make map usage conflict with artifact progression.
>Also, a second bug is that when clicking "Activate" in the artifact section with no artifact selected throws the error.
for me at least, just seeing numbers go up and get huge (Swarm Simulator) doesn't really do it for me. part of what makes Universal Paperclips so good is that, like Candy Box, a huge part of the joy is uncovering entirely new gameplay systems as you progress. A Dark Room was neat in that it brought the idea of a coherent narrative that you (sometimes subtly) uncover as you progress, too.
I feel like there's a lot of room left to explore in the space: different mechanics and systems to explore (outside of just clicking and upgrading), the possibility of cooperation with other players... the browser-based incremental game is pretty versatile in what it could do.
one of the most interesting one of these I've seen is Parameters (http://nekogames.jp/swf/prm.swf — download & play locally with Ruffle or some other SWF player). it's like an abstract RPG where you go on quests (or something) by clicking squares to fill them up. it's kinda crazy to me that nobody seems to have iterated on this concept.
Spaceplan is a relatively short and simple experience (there's not much to optimize, just do), but it's so well done that I highly recommend it.
The mobile port is pretty good too, functional and has some minor tweaks to make it more playable with mobile's more idle style. And the periodic check-ins to build more stuff are quick and easy - you might need a lot of them, but it doesn't waste your time or penalize you for taking longer. Nor does it really benefit from an auto-clicker, except perhaps very early on.
It is pretty slow though, and you'll have to experiment or check the wiki to figure some things out. I personally enjoy that, and I find it more figure-out-able than many of Antimatter Dimensions' challenges (which are essential to progress, and sometimes require hitting things you can't even see happening). Just don't expect to see real endgame stuff in a few days. It'll probably be months before you even see sephirots, much less as a viable target.
Fun game though! Definitely worth a play if you're interested in clicker games with an unfolding narrative.
Re planets: yeah just don't even bother. By the time they arrive, your production will have outstripped anything that gets delivered. I had a bit of fun setting up huge quantities of cheap things on nearly every planet and then racing the stream of pointless supplies around the map... but I was already making far more supplies per second than was delivered and able to be used due to storage and production caps (ignoring that I had no useful reason for increasing further, after some point you just trounce everything trivially).
I can elaborate if you'd prefer, just figure I should try to stay safe by default :)
I just remember the last time I played, I got to some point where nothing new was happening for a good long while, so I assumed I was at the end of the content.
It progressively introduces new game mechanics, which you automate away when they start being annoying.
Also: At this point I probably spent about the same time building excel sheets for figuring out prioritization of upgrades as actually playing the game.
(I'm not associated with the developer in any way, I'm just a satisfied customer.)
[1] In-browser demo (older version, saves may not transfer): https://marple.itch.io/orb-of-creation
[2] Steam purchase page: https://steamcommunity.com/app/1910680/
:/
My first time through.
Couldn't stop. Like a good human trained to be a bot.
Click games are good example of what humans will become, cogs, pulling levers for the AI.
Look at manufacturing or fast food, or many industries, can see it already.
In Game, I chose to eliminate the Drift. Since an AI would. The offer of other universes could have been a ploy to fool the AI. Don't think an AI would take the bet.
Is this a typical score?? Or can it go higher.
Paperclips: 30,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000
30.0 septendecillion
Good Night.
All productivity for the day is gone.
Haven't stayed up to play like this since Civilization.
Don't think I'll play this one again.
~55,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 (55 Quattuordecillion)
but alas, Firefox crashed.
https://news.ycombinator.com/item?id=24389655
https://news.ycombinator.com/item?id=15439569
https://news.ycombinator.com/item?id=36972227
https://news.ycombinator.com/item?id=22394560
https://news.ycombinator.com/item?id=32125372
https://news.ycombinator.com/item?id=35378453
https://news.ycombinator.com/item?id=20955089
https://news.ycombinator.com/item?id=15504187
https://news.ycombinator.com/item?id=15444762
https://news.ycombinator.com/item?id=30837131
https://news.ycombinator.com/item?id=29637543
https://news.ycombinator.com/item?id=32016242
https://news.ycombinator.com/item?id=31998798
https://news.ycombinator.com/item?id=30256647
https://news.ycombinator.com/item?id=29187050
https://news.ycombinator.com/item?id=22236702
https://news.ycombinator.com/item?id=25552628
omg, "A filmmaker thinks he can turn Universal Paperclips into a movie (2019)" https://news.ycombinator.com/item?id=24405682 I would rent that movie for sure. Sounds like a wonderful sci-fi B movie to watch.
(these are links to former discussions)
Universal Paperclips - https://news.ycombinator.com/item?id=33446121 - Nov 2022 (170 comments)
Universal Paperclips - https://news.ycombinator.com/item?id=30837131 - March 2022 (3 comments)
Universal Paperclips - https://news.ycombinator.com/item?id=29496595 - Dec 2021 (82 comments)
Universal Paperclips – play the role of an AI programmed to produce paperclips - https://news.ycombinator.com/item?id=27121348 - May 2021 (2 comments)
Universal Paperclips - https://news.ycombinator.com/item?id=26524117 - March 2021 (1 comment)
A filmmaker thinks he can turn Universal Paperclips into a movie (2019) - https://news.ycombinator.com/item?id=24405682 - Sept 2020 (2 comments)
Universal Paperclips - https://news.ycombinator.com/item?id=24389655 - Sept 2020 (84 comments)
Universal Paperclips - https://news.ycombinator.com/item?id=22394560 - Feb 2020 (1 comment)
The Unexpected Philosophical Depths of the Clicker Game Universal Paperclips - https://news.ycombinator.com/item?id=19513089 - March 2019 (52 comments)
Universal Paperclips – A Paperclip Production Simulator - https://news.ycombinator.com/item?id=15439569 - Oct 2017 (3 comments)
(Btw the convention is to omit links to past threads that have no comments, or only trivial comments. Otherwise people click on the links, find nothing of interest, and come back complain. Not a criticism! just FYI)
I think it can be done faster.