Search results

I think the boycott attempt in the rpg community had some success in getting Wizards to declare they weren't going to use AI art. But, I also think those tactics are going to be less effective now that AI is genuinely useful. RPGs are small and networked enough that there's still hope of...

That's...a very uncharitable reading of my post, which explicitly notes years of disruption as a consequence. By 'disruption' I don't mean 'minor harms only'.

I noted both of these scenarios in my post.

There is something paradoxical here. The folks pushing against AI are largely doing it by downplaying the capabilities of the models. "They aren't reliable. They hallucinate. They can't do task X". If all of that is true, then there isn't as much to worry about. There will be some disruption...

I think artists have good reason to fear, just as technical people have reason to fear. I'm not sure how close is 'close'--but people who use this stuff think we are within a couple of years of driving out entry level technical jobs. That seems merited, imo.

I think this is a reasonable ask, but I'm also not surprised the Arcane Library doesn't have it yet. Too small an outfit.

There have been some statements suggesting this idea in the long run, but I can't imagine anyone doing it seriously prior to profitability . An interesting recent result is Anthropic's hackathon, where many winning entries were from folks with little programming experience but a lot of domain...

I've been out to the REH museum and the volunteers working there are so generous with their time and very dedicated to the history. It's sad, but not surprising, that the house is in need of repair. We are lucky to still have it; they told me that it was within a very short distance of a tornado...

It has been a while since I was at the FLGS where I ran AL. But my impression is they are still running. I think it is being pushed less actively by wizards; afaik, they haven't published any AL specific content in years. So folks are either running the old content (the "season 1, 2" etc mods)...

Yes, they do. Here is the documentation for Claude Code. In "Start a session", the instructions are: If you check the models you can pick from, you get the following options For example

Claude code uses models like Sonnet or Opus, so issues you have with the training apply.

I think this is wrong on the merits. Most immediately, Claude Code played a key role in the rover study. There is much more to say, but I'm not sure it is worth going into the details.

Maybe it came across wrong--I'm not trying to convince people their concerns are not valid because I think those concerns are valid. AI will hammer the labor market and make jobs redundant and displace workers. It is going to make mistakes, consequential ones, that will get people killed. My...

Yeah, no disagreement from me

Right. The point is that they're aware of the reliability issues and think there are ways to address them even for high-stakes work. Therefore, the fact that someone saw a factual error in a chatbot doesn't tell us much about the path of the technology as a whole. From the anthropic article:

You don't see why people using models for tasks where reliability is very important is relevant to your point about reliability?

Respectfully, if your main engagement with AI reliability is checking the answers you get from chatbots or search summaries, I don't think that generalizes appropriately to AI more broadly. There is a lot of work being done by very clever people who are using AI productively in science and...