| 1 person marked this as a favorite. |
Go to [https://2e.aonprd.com/Rules.aspx?ID=2191] unarmed attacks AON[/url].
So yes you can have your hands full with potions and kicks or head-buts or shoulder slams are still available for an unarmed attack or any other bodypart available to strike with.
| 7 people marked this as a favorite. |
Don't ask LLMs for game rules
Or anything else that needs to be factually correct.
| 4 people marked this as a favorite. |
Right on, just straight forward then. Thanks for the answers!
For the record, I didn't ask them, they forced their opinions on me at the top of the search!
I can understand that, every google search result now basically has an AI summary at the top. Just ignore them. They make stuff up and are incorrect a lot of the time, especially when it comes to table top because they seem to have difficulty telling different editions apart from one another, and even sometimes confuse systems (like Pathfinder and D&D) because there are similarly named rules elements.
Don't ask LLMs for game rulesOr anything else that needs to be factually correct.
I just want to triple down on this. "AI" isn't designed to answer your question. Some like chatgpt are just there to mimic speech, others like Google search just recover data from other places and isn't capable of any nuance.
This isn't just for the OP but everyone who might read it, way too many people don't know what these processes do.
That might be confusing what a result might look like for the process that creates the result.
Probably there's more. And likely more will be possible.
And that is a gross overstatement.
They absolutely can work with texts, reformulate, change styles, formats, even make summaries (most often quite decent).
They do this without any understanding of the meaning of such text. Which is why the disclaimer of 'most often quite decent' is necessary. And which is why it is not going to ever get better.
As said earlier, you can never rely on AI if you need something factually accurate. Anything you get from AI has to, at a minimum, be reviewed by a real intelligence who is knowledgeable of the subject.
This is the primary implication of Rice's Theorem. To summarize: No algorithm is able to determine if an algorithm is producing correct information. This is why the computer programming industry - even after half a century of innovation - still has job positions for Quality Assurance Analysts. Someone knowledgeable has to review the results of what is created.
And still, the most ironic to me is that most of what AI can currently do - what is hyped to be new and innovative - has already existed for decades.
Oh, AI can analyze your source code for potential errors? That's cute. It's called a Linter and it has been around since the late 1970s.
AI can help doctors diagnose patients? Yeah, technology for that has been around for a long time too. Web MD was created in 1998. I expect paid proprietary systems were available before that.
Back in the early 2000s I got a handheld electronic toy (I think this one) that would play '20 questions'. The person could think of something and the device would ask questions to determine what the thing is. It was remarkably accurate.
The most effective use case of AI mentioned in this thread so far is to break a person out of writer's block.
