Good AI Task launches a diagnostic tool to tell you whether AI can actually solve your problem before you waste time trying

A new web tool called Good AI Task (shown on Hacker News) lets you paste any task or project description and get back an honest assessment of whether AI can handle it, plus what would need to change if it can't.

Instead of generic yes/no answers, the tool rates your task across five specific dimensions: whether it produces the same output every time (repeatability), how much wiggle room exists in the definition (ambiguity tolerance), whether you have the data and tools available, the cost if the AI gets it wrong, and how much human decision-making is actually needed — no marketing hype, just the gaps.

For anyone considering handing work to an AI — whether you're a manager deciding which workflow to automate, a student figuring out what you can ask ChatGPT to do, or an engineer prototyping an agent — this removes the guesswork: you can now test a task before committing resources to it, rather than discovering halfway through that AI fundamentally can't do what you need.