As some of you may know my daughter recently graduated from university with a degree in UX design. Watching her work is fascinating for me. The tools she has at her disposal today enable her to get from thought to clickable prototype in minutes. The AI revolution has made design faster and easier than ever before. The output from these tools is polished. It feels like it’s done. For a mid or late career designer the inevitable iteration on these initial drafts is obvious. But for entry-level folks, how will they know what good looks like? Especially, when what comes out of the tools looks pretty damn good already.
The answer is not “more AI’
I recently asked this same question in a group chat with a few dozen very smart designers, product people and other leaders. To my surprise, many of the answers that came back could essentially be boiled down to, “more AI” will help validate the output of the “other” AI tools. I know that this is a technique that is often used as part of building evaluations of AI tools. I can appreciate that it has merit there. But when it comes to evaluating design work, I’m not convinced that more AI is the answer.
For one thing, AI tools like Lovable, V0, Bolt and others don’t have context. The designers who use them — qualified or not — are prompting with these tools. They aren’t designing with them. They’re not designing with them because, aside from their education, they don’t really know how to adjust the output to fit the problem they’re solving.
For example, an early-career designer may prompt the app to produce a mobile shopping experience for retail clothing. The app will happily oblige (with a virtual smile to boot!). Now what? How will this designer know what to prompt next? Without the benefit of experience and expertise they will continue to blindly prompt “improvements” until they believe they’ve created something of value.
A senior designer would likely start with a similar prompt, perhaps with a bit more detail. Once the initial output is rendered though, their ongoing interaction with the tool, often through prompts as well, will focus on refining the interface specifically based on their experience and expertise. They will be designing with the tools, asking for specific UI elements, interactions and workflows rather than leaving it up to the tool.
The answer is mentorship and professional development
The heading here says it all. The way we teach entry-level designers to understand what good looks like is through mentorship. Senior designers and leaders in the organization need to formally take these freshmen under their wing and teach them to recognize good design vs bad. This takes commitment, time and money. In many organizations it requires some formal program that pairs mentors and mentees together. And it requires these senior leaders to take time away from their day-to-day work to provide this mentorship.
There’s another path — professional development. Providing a budget for a designer to attend a conference or workshop is a great start. An even better approach is a designed development track that is based on the understanding that our entry-level folks need support to become mid and senior level designers. The mentorship component mentioned above is part of this. On top of that, there needs to be a sustained cadence of training, community and coaching. Together these tactics start to build a foundation for career growth and with it, better design work.
Don’t outsource the development of your staff to AI
I know that, as a design leader, you’re busy. If your organization is bringing on entry-level staff there’s a risk of them producing subpar work without mentorship and professional development. Make the time to do it yourself and to partner with providers who support your vision for great design work. Don’t assume that the AI tools do good work on their own or that they can be policed with other AI tools. The responsibility is yours.
