@karlclement Agreed. I reflected on this as it relates to designers specifically back in 2022 and find it relevant as ever today:
https://t.co/iL3yh4uNrp
Thanks for sharing, and for engaging in this discussion Anthony! Threads can sometimes come across as more aggressive or defensive than intended. I see this as a meaningful conversation.
There are legitimate concerns in the list, but much of the critique assumes details we don't know, and presents potential problems as established failures.
That's my main concern with using AI this way: asking an LLM to find problems will of course produce a convincing list of objections. But the existence of those objections doesn't tell us anything about their validity, severity, or relevance.
Critique is valuable when it helps us understand tradeoffs, not simply enumerate everything that could go wrong. That's what I found missing from your original post.
There's also some precedent: JetBrains remembers individual tool window sizes, MacOS snaps window edges into alignment, and Figma uses contextual snapping (with an override).
None of these are identical to the original prototype we're discussing, but they demonstrate why assisted manipulation isn't _inherently_ problematic.
A compelling prototype doesn't establish that an interaction works. But a list of possible failures doesn't establish that it fails, either.
@hobdaydesign Agreed, designers should seek critique, especially with novel interactions.
But generating a list of potential problems doesn’t say anything about whether they’re valid, consequential, or worth solving.
And you still haven’t shared a single one of the nine problems yourself.
Things like this are the type of attention to detail most teams overlook. When we say “delightful experience” in software, it’s these details that can make it so.
Of course understanding and debugging matter. But you’re assuming those require manually reading and writing code. AI can help with both.
The real distinction isn’t who writes the code, but who can reason through problems, evaluate solutions, and ship working software. That’s engineering by definition.
@noiterations@karrisaarinen It's often easy to misinterpret _measurable_ progress for _meaningful_ progress.
Building tools or processes can be immensely valuable. But _progress on either_ doesn't guarantee the actual work gets easier or better.
Sometimes the better investment is simply doing the work.
Design hiring has changed. Larger talent pools, shifting expectations, evolving biz priorities, and AI, have made finding the right fit really hard.
On the latest episode of the New Layer podcast, we share what candidates and hiring teams can do better.
https://t.co/aq5p5zxi4K
Touch can be faster to use than a trackpad on a laptop. The reason is surprisingly simple: with a trackpad, you move your hand, aim the cursor, then click. With touch, that process is condensed to reaching and aiming in the same motion.
https://t.co/k4r1BjnWsd
@designertom You can do both. In fact, great design IS both. Too many designers just want to be artists and entirely miss the point of what it means for something to be “designed”.
@benhylak Team, we can invest time in either: A) Improving our user experience so users stay with our business, or B) Make unnecessary but cool things specifically at the point when users have told us they don’t think we provide any value. 😂