Pixels of the Week – February 22, 2026

Pixels of the Week is my weekly-ish curated newsletter for designers, UX folks, devs, and anyone building accessible, inclusive, usable (and let’s be honest, awesome) digital products. This edition looks at AI hype pressure in design and dev work and why chatbot-first UX often fails. Also: an amazing pink moth, embroidery insect art inspiration and a reminder to use prefers-reduced-motion for your animations.

Subscribe to my newsletter to get this directly in your mailbox!

What I’m up to: testing AI alt text generators so you don’t have to

There’s this tool that claims to write AI generated alternative texts for you, so, I gave it a try. And, it was a mess. Accessibility tip of the day: a tool that claims to generate AI alt text, but starts the alt text with “the image is”, is not a tool you should use.
Why? Because screen readers already announce images, so adding “image of” is annoying and useless.  (If you want to react to this, check the LinkedIn post, Mastodon and Bluesky). And if you want to learn how to write proper text alternatives, W3org has some tutorials for different types of image.

Want to learn UX research and product design?

My masterclass “Designing better products” is back, from March 23 to April 6. It’s 5 × 2.5h live sessions, on Mondays & Tuesdays, 17:30–20:00 CET / 9:30 AM – 12:00 PM PT. We will cover:

• user research & interviews (with some nice interview cards to help you build the scripts)
• user flows & content model
• information architecture, to help you build reusable components and pages
• wireframes, prototypes & usability testing
• documenting design decisions

This is NOT just theory. We will build a fake project together. In every session, YOU work, in small teams. We share, and learn, together. And you get my Miro framework: the same one I’ve been using for years, on real client projects, big and small.

Grab your ticket with a 15% discount

The pressure of using AI for our work

I read a couple of interesting articles, from people discussing the pressure of using AI in designer and developer work: follow the hype, or lose your job to people who will be the current threat. And it’s making a lot of folks quite anxious. The answer: wait 6 months, if a hyped new tool doesn’t survive those, it is not worth the hype. Also remember: speed is not automatically improvement.

Stop generating, start thinking – localghost (9min) We do not need more AI code, we need more human thinking about the code we ship. Spicy autocomplete is fine but outsourcing our judgment is not. Especially since LMM are trained on bad code to start with. Do we really want to have them repeat our human mistakes? This part resonates a lot with me: “As I see more and more people generating code instead of writing it, I find myself wondering why engineers are so ready and willing to do away with one of the good bits of our jobs (coding) and leave themselves with the boring bit (reviews).” Yes, yes yes, I fully agree with Sophie Koonin. I feel like it’s the same for design: why do we offload the fun, creative part of thinking about different UI solutions to AI? We are not anti-progress, we are anti-hype. Keep the human in the loop!

We mourn our craft (3min) Nolan Lawson also goes into this direction. He explains that he didn’t sign up to become a “a glorified TSA agent, reviewing code to make sure the AI didn’t smuggle something dangerous into production”. Developers must adopt these tools to stay employable, while privately mourning the lost joy, identity, and shared humanity of crafting software line by line. I can’t help to wonder if this was not the case for most innovations: people morning the hand craft.

How I’m dealing with the pressure to adopt AI as a designer (12min) We all feel pressured to keep up with AI as a designer. In an ocean of marketing-induced anxiety about AI, Martin Wright brings some refreshing perspective: “You wouldn’t adopt a new approach or tool without being sure it had a positive outcome for your work. AI is no different, and it doesn’t get a free pass just because the discourse is louder.”. Stay away from the hype, go slower, have an intentional approach: experiment widely, deploy narrowly, protect your judgement and the messy middle of the work.

What, then, are we paying for? (5min) Even if AI lets anyone build tools, it cannot replace long term problem ownership, judgement, and focus: “The intrinsic complexity and cost of building a solution was high. Much higher than what it costs to pay someone else to deliver a solution.” So, when you pay for a software, you still pay for teams, to deeply understand the complexity, for you, and carry that responsibility. And this can’t be outsourced to an AI. By Quinn Keast

On pushing AI chatbots…

Are we doing UX for AI the right way? (12min) Chatbot-first UX is a trap. Many tasks need structure, rich, multi-modal interaction patterns that conversational interfaces simply cannot support. And predictable steps. A big empty box where you ask people to prompt something is not going to work for most of them. Such interfaces create accessibility, discoverability, and reliability issues. A great article by Katya Korovkina that hits home, especially in enterprise UX. Many users are not tech savvy, and don’t remember exact terms, so a chat box alone is risky. Filters, facets, and guided options reduce memory load and make work tools more usable.

Curiosity cabinet

Photos of Rosy Maple Moth (Dryocampa rubicunda) if you are having a bad week, I would like to remind you that nature is beautiful and that the rosy maple moth is an icon, a legend, and she is the moment.

Beautiful art inspiration

Atelier Noboru is a French art embroiderer who creates the most beautiful embroidery insects. You can also buy them on his Etsy page.

CSS properties you should know about

Dear developers, please respect the reduced-motion preferences of your users.
I’m tired of your websites making me nauseous. Motion is fun until it makes your users sick. Check the Prefers-reduced-motion  media query!