HN user
It's so weird to me that the benchmarks remain so low, but the models are marketed as revolutionary. And if you say that low coding capabilities aren't a problem, say that to the token price hike and 'general use' model setup.
Why not sell it as a math agent? Why do I have to set up 4 agents to check each others' work?
I don't think the paper is saying hallucination is limited to LLM's. The decoupling that needs to happen is the subconscious notion that LLM's are computers.
Computers are predictable and calculated (still based on a person's design), but LLM's are unreliable and unpredictable. If the general populous assumes LLM's are as trusted as a calculator, we're in for a bad time.
I hope there's a better way to prove that beyond unrealistic stress testing while speaking in absolutes, but honesty needs to come first on all sides.
I've been working on a ttrpg site for the last year with my own IP. The intent is to create an experience that makes an intuitive UI that minimizes tedious tasks.
You need to know "does this guy look hurt"? The enemy HP bar can be set to either an actual percentage, or set to have cracks in the bar to signify a range of damage. Does only person take notes? Personal notes are shareable and there's a section for community notes. Do you have enough perception to notice a hidden door? The UI can be set to go off passive perception and give you those notifications automatically.
It's still in early alpha testing with friends, but it should eliminate general GM pain points to encourage more groups to form.