On AI: Where I am halfway through 2026

Here’s where I land on generative AI, halfway through 2026. This is strictly about the work — none of it accounts for the other problems with modern AI: the ecological cost, the damage to the PC market (gaming especially) as RAM and solid-state prices skyrocket, and the moral weight of handing over some of our personal autonomy to an algorithm built by out-of-touch billionaires insulated from the consequences.

• Brainstorming and ideation: Definitely pro. Especially as the sole product/UX person on my team, having a tool to think out loud with is a valuable aid. It doesn’t replace a good conversation with expert people, but it’s a workable substitute with the right prompt and mindset.
  • Recommended
• Generating wireframes and mockups: I’m a bit more hesitant here. I’m a fan for ideation and coming up with variants, but nearly every time I switch to doing it myself pretty quickly. Generative AI just isn’t there yet for creating solid, usable, useful mockups for the more complex web applications my teams build.
  • Cautiously recommended
• Creating attractive interfaces: I’m on the fence with this one. It does a pretty good job decorating a design, but it tends to look same-y (likely due to the reliance on Tailwind and that framework’s rapid growth). With the right design system, though, it can save real time.
  • Cautiously recommended (without a design system: wary)
• Generating front-end code: 100% in favor. Why write the code myself when I can describe what I want to Claude and let it do the heavy lifting? Admittedly, it helps to have a solid background in computer science, web application development, and user experience design to inform how prompts are structured. That said, the developers on my teams are less eager to embrace it, and it definitely increases their workload to review the code. As with UI design, it helps to give the AI something to guide its technical decisions.
  • Recommended
• Media generation for fun: Generating images and music and video is kind of fun, but it gets old. It’s useful to know the limitations of the technology, and I think it makes me better at spotting generative AI out in the wild.
  • Meh
• Media generation for professional work: Don’t. Pay your artists. Let creative expression thrive. If it’s for a personal use, such as for a portfolio site, then sure; my avatar on the homepage was generated by Google Gemini from a photo of me. But in the work I do, I prefer to partner with the artists in our Communications and Marketing group to create bespoke art. That goes for images, video, all of it. There’s some wiggle room around presentation templates, but if a nice template is already available, just use that.
  • Avoid
• Writing emails: If somebody is relying on AI to write an email, what’s the point? Write your own email. But I do use the available tools (primarily Copilot, since my work is a Microsoft shop) to refine it (trim words, sharpen the key points) as long as it still sounds like me.
  • Cautiously recommended
• Drafting documents and longer-form content: My approach varies by document, but at minimum I start with a human-written, detailed outline of what I’m going for and then use generative AI to expand it. Even then, it’s often faster to just write the whole thing myself.
  • Cautiously recommended

I’ve also started paying closer attention to my habits around generative AI out in the real world. If I see a YouTube or TikTok video that’s obviously AI, I block that creator. If I read an article that sounds generated, I avoid that website. I’ve yet to run across an AI-written book, but I’m confident I’d toss it in the nearest fire.

TL;DR: I’m fine with AI for ideation and brainstorming, content editing, and code generation. I’m ambivalent about UI design and personal media generation. I’m against it for content writing and professional media generation.