besterman23

It’s like trying to create a new type of imposter syndrome, but instead of referring to complex data structures and architectures, it’s the most effective esoteric inactions to chant at the Jira ticket resolving machine.

> am I committing violence against him? Should I be force to work for him to avoid that "violence"? I don't think so.

Being forced to do work is violence yes? Slavery precisely. Why can’t you be forced to work? Because you’re protected against it, only because you have the capacity for violence, or at the very least the benefit of promised violence. Otherwise you could be coerced to do anything as you would have no recourse.

We have laws that protect us from coercion, both physical and otherwise. Those laws only exist because of promises of retaliation if they are not adhered to. So yes, refusing labor is a violent act (or at the very least, violence by proxy).

Perhaps I should say “refusing labor with the expectation of not being assaulted or killed, is a violent act”.

I’m saying the bet is in them becoming THE System One Model company.

Probably the “nobody ever got fired for buying IBM” effect. If there’s a use case for the tech, buying the most well known implementation of it will always be useful for people who want credit without the threat of blame. This funding is based entirely on the hype and a bet that TypeSafe will have name recognition.

Exactly why I’m confused with the “techno-feudalism” point as well. What are the serfs for if they’re useless (sans some physical labor maybe?). The AGI dream is for the smallest amount of humans needed to keep the system running, and if AI can do everything feasible, then that group is basically those that “own” the AI. Techno-feudalism, if even necessary, would be a fleeting blip in the end.

Unhinged conspiracy, if you take the implicit goals of AI research orgs at face value (I.e., replacing all knowledge work) and assume it possible; why would they drag a few thousand knowledge workers along to the promised land when they can just shed them after their usefulness has worn out?

What does adding a few more suckers to the pile of tens of millions of other suckers do to those that get the real benefits? It’s not like after they achieve their goals, the people who helped them would have the ramp or resources to recreate the process.

Why even respond to these comments anymore?

Every new model you make this post, every time there’s someone who posits it might be trained on, and every time the answer is “maybe but probably not” it’s not a worthwhile conversation to have at this point, either the models can do some arbitrary thing or they can’t.

I guess it would result in the same outcome if it knew it exploited a bug (and didn’t disclose that) or not.

I agree with this, however I’m not sure where the idea of refusing labor as not a form of violence comes from. The WHO describes violence as “the intentional use of physical force or power, threatened or actual, against oneself, another person, or against a group or community, which either results in or has a high likelihood of resulting in injury, death, psychological harm, maldevelopment, or deprivation” and refusal of labor as a political tactic definitely falls under the category of “use of power, threatened or actual, against another person, group, or community, resulting in deprivation”.

That falls cleanly in the realm of a violent act, unless you disagree with that definition of violence, or my interpretation of the quote. I suppose you could interpret it as only applying to physical violence, but I don’t think anyone would agree that psychological or emotional violence simply don’t exist.

I think there’s also something to be said about how social technologies are used to reduce the efficacy of democracy as representative of the people within them.

We are so divided and mislead that I feel confident there is a significant sum of non-ruling class individuals who are completely for their own oppression for no reason other than spiting a perceived “other”.

Some other specific examples that don’t all flow together:

Globalism promised efficiency and a way to materially improve conditions for people as consumers and producers. However it has been used as a cudgel to threaten workers that would ask for more and keep down workers who have no other option. Also pitting working class people against each other for the benefit of a few.

Social media, it promised untold communication between people who would never have been able to communicate before, and it would allow them to spread ideas. Instead it is used as a dumping ground of nonsense information, drowning out any semblance of coherent thought.

I don’t think it’s useful to argue whether withholding any particular labor is violence or not. In some cases it is apparent (refusing to maintain key infrastructure is akin to actively destroying it) and some cases it’s not (who cares about nobody wanting to build your app).

However, what I was trying to convey is that voting is not inherently something that holds sway over anything. Votes are sort of like the currency of democracy, and like currency they need to be backed by something. U.S dollars are backed by the countries capacity to physically control strategic resources like oil, land, etc. (I.e., the U.S capacity for violently controlling resources), or emit soft-power (swaying other countries to their benefit).

Votes in a similar fashion are also backed by your ability to deny or inflict your personal power on the system you are a a part of. The clearest manifestation of that power being your ability to contribute to the institution as a whole. If we significantly reduced that capacity, or made it unnecessary for the continuation of the current system, then the power of that vote is reduced in-kind.

Democracy only exists as a suppressor of violence, the only reason why democratic institutions are upheld is because all participants could enact violence as a response to perceived threats to their group that democracy is seemingly failing.

Historically, if democratic institutions failed, refusing labor to a ruling class has been a first violent step for the laborer class and a preemption against actual physical violence if the ruling class overstepped. However, we are seeing that possibility being taken away step by step. This leaves only physically violent uprising as a means of protesting overstep, and it is not obvious how effective that will be given the massive power imbalance between the labor class and ruling class.

So to put it succinctly, no I don’t have faith that democracies will solve this issue, because democracies only work when there is some semblance of equilibrium.

I see it as “if this can be represented in tokens it can be trained in and ‘solved’”. I don’t think there will be a plateau, but there might be issues with how effectively we can represent some things in a tokenized form and still be efficient.

> If you ignore every instance of this happening it's really easy to see no instances of it.

Yeah, if this is true then I’m wrong.

I fully agree that it’s plausible for an AI to do things that are outside of the purview of the users intentions and be very dangerous and capable while doing so, however that fact alone doesn’t absolve me of responsibility for instantiating the AI that did a bad thing if that AI would never have done the bad thing if it was never instantiated.

You could possibly move the blame higher to the manufacturer of the product, for example, the weapons grade plutonium you provided, it doesn’t exist unless you take intentional actions to make it so, and even when it does exist it doesn’t nuke a city unless negligence or intention is applied, in both those cases the fault lies in the initial operator. We don’t blame split atoms for the chain reaction caused.

Now, if an agent decided to spontaneously and maliciously act in a way to cause harm that is in direct contradiction to the initial intent, then yeah it would totally be the AI’s fault, however I don’t think we have seen that yet (I’ll change my opinion if I’m wrong here) and until we do I can’t place blame on the machine.

If I tell an AI to “do what it thinks is best” and it just starts randomly hacking things with no particular goal in mind, how different is that from me telling a campfire the same and then leaving while it burns a forest down.

In both of those scenarios I am responsible for my negligence, even if in the latter I happen to die in the forest fire. Neither scenario existed without my instigation.

The question now is HOW responsible am I? That depends on the intentionality I put into instantiating the campfire/LLM.

I just don’t know enough to agree with you, but I would argue, probably poorly, that an AI or collection of agents could “invent things” simply by virtue of trying essentially everything in a reasonable bound and stumbling into a solution.

Call that brute force perhaps, but I would consider it technically inventing something on the merit that it would at least be an abstraction above naively throwing everything against a wall to only throwing things that would most likely be sticky.

That’s fair, I have no idea if this simulator is actually accurate but I appreciate the response.

For the benefit of a layman, can you explain why this is so much different than a human doing it?

Like sure it didn’t have the inclination to make the sim and hardware designs, but it did make them though yes?

Primarily forgetting how to be a half-respectable programmer.

For a lawyer yes, if a task takes less time they bill less hours and therefore it costs less.

So… if AI is reducing time taken for billable actions, then that reduces the cost no? Unless it’s not actually reducing the time spent on said actions.

If I were to guess, being pleasantly surprised is just a learned appropriate social response from the expectation of receiving a reward and as such, that social norm is codified sufficiently enough in our writings that it appears in LLMs output.

It’s sort of like all the people who will ask Claude or GPT to validate their complete nonsense and receive unyielding praise for it, the models just learned that this is the best received response based on training data and RL.

I bet these same sorts of expressions can be found in practically every failed attempt as well.

I suppose you could say, hypothetically, that if the field were “solved” then the protection of the feelings and sense of need for human understanding of the now obsolete practitioners might hinder any benefits that the now solve field could provide.

The answer to this is probably just “get out of the way” and then your “work” would be to develop a better appreciation of what has been produced.

Stop trying to pump your Tarot reader IPO

“Yes, there’s still enormous uncertainty about what the rest of our lives will look like, but as far as I can tell, there’s no longer any real uncertainty that it’ll all mostly revolve around AI, and the extent to which we succeed or fail at directing its power toward human flourishing.”

Good to know that in the throes of coping with apparent existential unknowns, even smart people can still wiggle in some hubris. Imagine, using his own analogy, upon the resurrection you try to steer the son of the Hebrew god toward the perceived benefits of western moral philosophy.

Suppose it matters what order you do it in.

The best good thing possible for humans who are all suffering from an extremely painful, incurable disease might just be ending that misery.

The worst bad thing possible in a perfect utopia might be a minor inconvenience comparatively.

Honestly I think it’s an apt punishment for ever inventing YAML in the first place.

I wonder if multiple attempts at the opossum would produce better results.

If we didn’t have the previous example I would interpret this as pretty solid evidence that labs were training on the Pelican “benchmark”.

I just can’t imagine a model dropping so significantly from one version to the next on such a silly task.

Yeah if this is his “information he wanted for years” it’s pretty abysmal in terms of crashing the “ai bubble”.