Timwi

I was actually surprised to see Trinity in the list. Not only does it remind me of The Matrix (1999), I also don't think I've ever encountered someone with that as a real name.

On the one hand, the paper (and especially the headline) overstates and overhypes the results.

On the other hand, the overstated results, if they were true, would be tantamount to saying “we're speedrunning exactly the kind of recursive self-improvement that is a major ingredient in every human extinction scenario”, and yet it is both presented and received as an exciting opportunity.

Brains and neural nets are already quite different things and difficult to compare, but at least you can wave your hands and vaguely say that both are “thinking” by some means of “abstraction or generalization”. DNA is absolutely nothing like that. DNA is more like an encoding for a flowchart for a recipe for building a specific body.

Try using the zero-width non-joiner instead (U+200C).

Except that only works until the ads become designed to deceive the agents.

Well yeah... in a time before audio recordings and films, bards and poets were the entertainment superstars. Today it's singers and actors, and we still give them millions for their output.

A “fiction” implemented as environment variables, which the ffmpeg process would inherit, so there's nothing stopping ffmpeg from “indulging the fiction”. Depending on how frequently users expect this behavior to work, I might even call it a bug in ffmpeg (or whatever file handling library it relies on) that isn't even all that hard to fix.

I notice that the first iteration of a drawing is usually by far the only good one. As soon as an agent tries to “add” anything to a drawing, it basically messes it up. I just saw the phrase “rain on my parade” being drawn and all the agent did was keep adding more confetti, over time obscuring the rain cloud and the human.

> One person commits a violent crime against another and you aren't holding them entirely accountable for their actions

Incorrect. Nowhere did I say or imply that.

> your looking to who they may have listened to and what they may have heard such that some blame belongs there.

That is correct.

You seem to think there's a fixed amount of total blame X and if we assign even the tiniest amount of blame to one person, then the main perpetrator must get less blame than they otherwise would. Nothing I said implies that.

> Notably you didn't engage with my analogy at all.

That's because it's not an analogy, it's a strawman.

> Would how would you legally distinguish the two scenarios?

One of them is wearing certain clothes, the other is intentionally inciting hatred and violence. If you can't make that distinction, I don't know what to tell you.

> Who defines what a "hate speech" is?

Who defines what is murder and what is self-defence? You already know the answer to this. The legislature.

Your first link is about police misconduct, not what the laws are or should be. Your second quote does not contain any of the relevant posts and just spews random unverifiable claims.

Nobody is talking about shirking any blame off. Maybe this is our disconnect. You think there must be one singular person to blame entirely. That if someone is held accountable for a consequence of their action, then no one else can. That is not how the real world works.

I don't disagree with you that Discord is the best option for voice chat now, but there was a godsend before it and that was Skype. It didn't have the equivalent of Discord “servers” but it finally solved the voice chat problem that was previously plagued with technical issues, instability and incompatibility. Discord became the godsend when Skype decided to speedrun enshittification.

In the choice between being allowed to carry a gun (and having people with guns all around me) versus being offered free healthcare (and having healthier people all around me) I know not just which one I'd choose, but also which one I'd call more free.

That is a very American perspective, which I like to think many people in the civilized world find irresponsible. (In the literal sense. You don't want to take responsibility for the consequences of your actions.)

It is a common misconception, partly intentionally perpetuated by the UK right-wing press, that people are arrested and jailed for making jokes or otherwise casually chatting on social media. If you dig deeper you'll generally find that people jailed for Twitter or Facebook posts went vastly beyond “telling a joke” or “saying something inappropriate”. The UK quite meaningfully prosecutes hate speech, death threats and similar felonies, and I for one don't want to live in a place that condones or even actively protects it. Downplaying it as “a joke about immigrants” plays right into the hands of the xenophobic right.

Has anyone ever tracked down SolidGoldMagikarp and interviewed them? It's such a uniquely bizarre claim to fame.

How is it a strawman if I've seen plenty Marvel movies and report on my experiences?

I know that you copied this from the article, but isn't this part of the prompt wrong?

> another instance of the same model should have a reasonable chance of recovering the secret after inspecting the SVG;

The whole premise is that the decoder model can't see the SVG; it can only see a rendered PNG of it. I wonder if changing this part of the prompt changes the results significantly.

That watermarking is not something the LLM chooses to add, neither through training nor in response to any prompt. This article is about the choices LLMs make when explicitly told to hide messages.

Ah, now I see what you mean by distraction. I wholeheartedly agree with you.

I think we're in agreement except in mere terminology. I was disagreeing with the characterization as a “distraction”, but I was also saying that if techbros pay out something for the plebs and call it UBI then it's not really UBI. I wholeheartedly agree with the last sentence of your post, and I think it won't be real UBI without a real democracy.

Depending on how broadly or narrowly you define “profit”, your statement is either too weak to be useful, or false.

If you define it so broadly that the hunter-gatherer inventing a spear for hunting more effectively is “profit”, then your statement is a tautology. The spear innovation was driven by “profit” and the “profit” is having the spear. Great.

If you apply a narrow, capitalist definition of “profit”, the statement is false. Plenty of advancement of civilization was and still is being made without or outside of capitalism, especially if you go as far back as the Stone Age that you mentioned.

You're not wrong, but we should not absolve those who perpetuate the fairy tales from their share of the responsibility.

You're talking “always existed” too literally.

> If these tools could do your work, people would feel existential dread anyway.

That is, if your entire self worth is tied to your work.

I think we should consider moving away from that notion.

I think that is what we should do. It's called UBI. It acknowledges that everyone has inherent value, at least enough to deserve to stay alive and participate in society.

I'm not sure I agree. You make it sound like you think the idea of UBI is for the rich to pay it out to us, the plebs. That's us-vs-them thinking. The idea is for the society as a whole, democratically controlled by everyone, to pay it out to everyone equally.

Obviously that can only work in a real democracy that is really controlled by everybody, not a fake one where you technically vote but it's really controlled by rich people. That's oligarchy.

Surely by then we'll have the technology to simply adjust the Earth's rotation so that it conforms to our standards.

A century later we'll finally be able to fix Earth's orbit to make a nice round 360 days per year, 12 months of 30 days, nice and clean, no more complexity or confusion... ah, that'll be the day.

I'm a little confused; the article text implies an exciting battle for resources where the agents discover they can steal from each other, but looking through the transcript it seems that only one agent discovered that and used it only once before the end of the experiment. Did I miss something?

> I'd love to see data, but my intuition is that the average developer has access to dramatically better security reviews and far lower cost than ever.

Where? If I ask Claude to do a “security review” of my software, it gets blocked as a possible hacking attempt.