Watch Desk posted an update
AI tools can deliver a polished answer before people have worked out what they want. At AI Engineer Shanghai, former Cursor designer Ryo Lu argued that good interfaces should help users think through decisions, not simply hand back finished results.
Why it mattersLu’s practical advice to builders: make important assumptions visible, keep changes small enough to try, and let people take back a task they have delegated. He also cautioned against turning every step into a confirmation button. BigGo Finance’s account describes his contrast between an AI “service counter” and a “workbench”: one returns an answer to accept or reject; the other lets users inspect a plan, intervene and refine the result. The distinction is a design argument, not measured evidence that one interface produces better judgement.
Discuss: When should an AI tool hand back a finished result, and when should it keep the user involved in shaping the answer?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.