Designing in the age of AI: why foundational skills still matter
Steve Johnson runs a product and service design studio that uses AI in every stage of its work. He and Aziz Musa explain why junior designers still need the foundational skills AI is quietly taking away.
Contents
AI now does much of the slow work in a design studio, and the risk is that junior designers never learn the craft that lets them spot when it gets things wrong. That is the problem Steve Johnson and Aziz Musa keep returning to in this episode of The Impact Lab: AI is a superpower for people who already have the foundations, and a trap for people who never build them.
Steve founded Furthermore, which he describes as an evidence-led product and service design studio, in London in 2013, and also runs the walking app Go Jauntly. He and Aziz worked together about ten years ago.
Where AI already does the work
Steve's studio uses AI in every part of the process, and research is where it has changed most. Interview transcripts that took days to analyse are now produced, analysed and grouped into themes for them. The research itself is still done by people, with users and stakeholders, often before any prototype exists.
The bigger shift came when the team put all of a project's research into Google's NotebookLM. For the first time, he says, they could have a conversation with their research. The tool's Audio Overview turns sources into a discussion between two AI hosts, and instead of the large PowerPoint summary they would once have built, they sent the client that podcast. The client listened on a flight to Davos and loved it.
Aziz describes a parallel at Cush Digital: interviews at every level of a new client go into a private GPT, so anyone on the team can ask it what the client's primary objective is, and why.
Six creatives in a room
In ideation, AI gives the team a first pass at priorities. Because NotebookLM knows all the research, it can suggest an order of work and link each hypothesis to product features. That used to take a couple of days of sticky notes on a wall. Steve still prefers the old start for ideas, though: "there's nothing better than, you know, like six creatives in a room", with AI used afterwards to look for holes.
Aziz calls this the hybrid approach: AI takes the work nobody wants and becomes a sparring partner for the rest.
What AI gets wrong about design
The difficulty is that AI learns from what already exists, and the internet is full of badly designed websites. One of Steve's designers asked for a gym aimed at women. It came back pink and full of stereotypes, and the main image showed a man at a desk. Ask for a dashboard twice and you get the same charts in the same style, when a single big number might tell the story better.
"If the prompt engineering is taking longer than just kind of moving things around, then it's a waste of your time."
So at Furthermore, AI fills mock-ups with images and handles admin, while the design itself stays with designers. Steve admits that may be old-fashioned. Aziz does not think it is.
The craft juniors no longer practise
Steve's worry is the small pieces of craft that juniors used to learn on. Cutting a person out of a photo and being told by your boss you had missed a bit of the hair. Naming layers properly so others could follow the file. One of his designers told him they no longer name layers: AI does it. But naming them gave you time to think about how elements nest and work, and names such as the hero and the burger menu probably came from people naming layers. Without that foundation, he says, "you're not really going to be able to spot when it's made the mistake."
Aziz splits it in two. Foundational knowledge is recognising that something needs cutting out. Foundational skill is cutting it out cleanly. Skip both, and AI seems to make the job easier while it shrinks your ability to grow. He has a designer he calls a wonderful artist who he worries will become only a good prompt writer, a "prompt monkey", as he puts it affectionately.
Learn the basics before the shortcuts
Aziz has acted on it. New joiners at Cush Digital used to ideate with AI from day one, and he realised it was not helping them. Now they cannot use AI for their first three months. They write posts themselves, learn how to layer a story in copy, what a call to action should look like, how to write hooks and how to research competitors.
He gives a recent example. A team member was asked to make a mood board with AI, and with a good prompt it produced a perfectly good one. Aziz still felt something was lost from not scrolling through Pinterest to build it, though he struggles to say what. Steve names it: moving things around, grouping them and telling a story with them, a skill his art directors taught him to value. A friend in fashion has been told to generate her inspiration with AI instead of visiting markets, a cost cut he thinks risks making everything look the same.
Keeping a hybrid team human
The answer, Steve says, is to connect on a human level and stay curious. His team is together only a couple of times a year, so the daily stand-up is kept deliberately light: the weekend, the cinema, what people watched. Those chats, he says, are where Go Jauntly came from, out of conversations about people wanting to get outside more. There are also design parties with odd briefs, such as drawing yourself as a sandwich, and a training budget for events.
Aziz adds taste. His wife says she needs to travel to "clean her eyes": to be exposed to beauty, scenery and architecture that feed her creativity. Steve asks new people what they did at the weekend and which exhibition they saw. Many sources of inspiration, not one.
Designing for people who fear the prompt box
AI has also brought Furthermore new work. The studio is building a tool that helps nonprofits use the data they hold. Most people, Steve says, find an empty prompt box intimidating, so the tool lets them choose from menus, describing their region and problem, and then builds the prompts for them to take to whichever AI they use. That is the human side of AI design. If you want design and video that start from the people you are trying to reach, that is how our design and video work begins, inside the marketing department we run for clients. For related conversations, see our episodes on no-code automation and whether AI will replace jobs.
Questions people ask
How are design studios using AI?
Mostly in research and early ideation. Steve Johnson's studio uses AI to transcribe and theme user interviews, to question its research in NotebookLM and to suggest priorities. The design itself stays with designers, because AI's own designs tend to look the same.
Will AI stop junior designers learning their craft?
It can. Steve Johnson worries that tasks such as cutting out images and naming layers, which juniors learned on, are now done by AI. Without that foundation, he says, designers cannot spot when AI has made a mistake.
Should new designers use AI from day one?
Aziz Musa thinks not. At Cush Digital, new joiners cannot use AI for their first three months. They write posts, hooks and calls to action themselves and research competitors by hand, so they build the foundational knowledge and skills first.
The two-minute diagnostic shows which pattern your marketing is in
The diagnostic asks nine questions and names the pattern your marketing is stuck in. No call, no pitch, and you get the paper written for that pattern.