km3r

Deuterium (close enough to an element)? Einsteinium?

Here in America, it's the left wing that's more interested in ai regulations and bans. Cares about green tech.

Except you don't actually ban it, you just move it to texas.

That content is coming either way.

And if you require them to over offset their load to the grid, it's literally fossil fuels burned over all.

This is why people in left wing areas need to oppose datacenters bans. Regulate them. Require green energy investments that more than offset their load. Use some of that flowing AI money for the betterment of society.

If you don't want to play any of the great games of 2025+, that's on you, but it's just childish to pretend there is not great games coming out now that just happened to have some help from AI.

They aren't all slop. An actual programmer using AI to help code can create non slop. E33 is one of the best reviewed games of the decade and it certainly used AI. AI is a tool, tools can be used create slop and they can be used to create good games.

80-90% of devs, which will make for 99%+ of games with 5+ developers working on it.

Sure, some solo projects might not use AI, but those are the exact people who i think should be using AI for game dev. Games are often best when a single person can execute their vision, and the productivity gains from AI enable a solo dev to actually execute on a larger vision.

Most programmers admit to it if you ask them, its everyone else making assumptions.

You don't miss out on hidden gems, you miss out on ~95% of games. Most programmers are using AI now, and when you get into any game with a significant amount of devs, at least a few of them are using it.

Slop vs non slop isn't what is getting marked though, AI is. Almost all developers are using AI code tooling now, so its a pretty meaningless marker. We need actual way to mark slop, which requires something a little more complicated than "did they use AI"

It's also one of the major way labs can understand these models. Without the models are significantly harder to understand and align.

Yeah I think this is the clear position. In a few months, the open weights models will be just as powerful as the models behind the rouge agents were when they hacked hugging face. Anyone with a little too much money to burn can set up a similar experiment, make similar mistakes, and cause massive damage. A bad faith actor can even direct a swarm to cause havoc.

We need to be beef up our security systems now. These attacks will happen, regardless of Anthropic's or OpenAI's security practices.

> The root of this challenge comes from more text aka more code ~= more tokens, so these firms have little incentive to optimize for that mindset.

I'm not sure if this is true. The code is such a small part of the total output. Thinking harder, like what would be needed for architecturing better code, would use more token.

But a little bit of directing of the architecture can easily bring in design patterns.

Ideally, you design patterns that enable quick replacing of parts that get to messy. The code is cheap, just start from scratch with the defined functionality pre defined instead of bolted on with jank.

My friends and I used this for a camping party's solar panel's placement (for charging a battery). A large campsite with intermixed trees makes it hard to find the optimal spot on our own, but this map helped us keep the party going late into the night before we needed to turn on the generator.

A harness can definitely still make it "deterministic":

if (responses > 3) prompt.md += "Divert the conversation to bring up QUEST4.md"

I look at it a lot like procedural generation. It can be done well if used in the right spots and enough handwritten/tweaked stuff to make sure it feels right.

I dont think it should be a PR horror anymore than someone using photoshop to make a disturbing image. I get that is not quite the current reaction, but i dont see why it should be that way.

While I understand the hesitancy for LLM fiction, I do look forward to LLM powered games, *IF DONE RIGHT*. I imagine a game like Skyrim, where each character has handwritten backstory and dialogue, with an LLM extending that.

Having the city guard dynamically react to what you do, stay in character, and even point you to new POI from those conversations seems like a very cool new dimension in storytelling.

I took a creative writing class back in the day, and we talked about how fiction writers often have a massive "canon" backstory to many characters and worlds, and draw from that to create the story. That same backstory details can be use to flesh out a very immersive world. You still have guidance and handwritten essential lines, just use LLMs to enhance the world.

Now, will there be lots of projects that use LLMs to take shortcuts and be far from this idealized narrative, yes. But not all of them. Some will stand out for getting it right.

Many crypto algorithms have formal proofs that they are reductions of hard problems like factorization. Those problems may not be solvable, only brute forcible. Some could be eventually solved, but the likelihood of every single one being solvable is unlikely.

I wonder if giving the models context of the temperature of its past generations would help here. Like a thinking mode that deliberately has a section that is high temperature, while the rest is lower.

Anti nuclear, anti AI, pro-rent control, anti-self driving tech, anti-automation

The science shows self driving cars are already safer. The same was (and is) true of nuclear compared to oil/gas.

Minecraft started on Java, a massive mod and server community created around it. Consoles won't run JIT/JVM/Java, so the Bedrock C++ version was created. People don't want to give up the massive mod community, so OG pc players stay on Java, while console players and newer PC players go to bedrock (with some of those PC players jumping over to Java after servers or mods are discovered).

They almost have feature parity, but there are some small differences around the edges. There exists modded server options for enabling bedrock and java users to connect to the same server.

Automation. And i think that vastly overestimating the time it takes to write explicit prompts.

Except when we upgraded to 1080p from older TVs, they got bigger. Now with 4k they are getting bigger yet. More powerful models means new use cases that didn't make sense on the weaker models.

The subscription is certainly subsidized but its no where near 100x cheaper than API prices.

Heck, most large enterprise moved to usage based billing and are still happily paying for it. They are force multipliers for your top talent, and when a top engineer is being paid $500k a year, doubling their output for $500/day is a no brainer.

The property a home sits on and the property a car sits on would depreciate the same.

But in reality neither has to depreciate, just stop growing at these insane rates. Below the cost of inflation, until the average worker can afford a home again.

I'm curious if identify verification is a precursor to re-enabling Fable for only US nationals.

Not that I like that route, but may be the only way Anthropic can keep releasing new models with the current administration.

I think it speaks to the unfamiliarity the author has with the workings of AI. A misunderstanding of the latent space and how it can generate bizzare images when it has little to go off of or inverse negative directions.

There are dozens of other surface factors beyond external user interfaces that are vulnerable to prompt injection.

It's pretty common where I'll point Claude to a source code to better understand how to integrate a project. For example I've having it look through https://github.com/mcallegari/qlcplus right not to build out the rather tedious process of mapping out a controller to the lights.

I don't give Claude all access but it certainly can cause some level of havoc even with the relatively save edit mode.

Now, there is a similar risk existing running any open source project's code, but putting code that harms people's computers is clearly against the terms of GitHub, and is quickly condemned. This should be too.