A polished portfolio now tells you almost nothing about whether the person behind it can think. That's my argument, and it's the one I'd put to every creative director, hiring manager and candidate reading this: in 2027, the useful part of creative portfolio assessment is no longer the finished image. It's the evidence of decisions, the reasoning that survives when you ask "why this, and not the obvious thing?"
The reason is simple. Generative tools have made competent execution cheap. A lit render, a clean brand system, a punchy launch film: any of them can now be produced by someone who's never made a judgement call in their working life. So the surface has stopped being a signal, and the work of evaluation has moved underneath it.
Why the finish no longer proves anything
I spent twelve years commissioning and judging creative work from the PR side, which means I sat through a great many agency credentials decks and campaign reviews. The rule I learned early was that the glossiest slide in the room usually came with the vaguest story. Nothing about that has changed. It's just gotten easier to fake the gloss.
What I'd look at now is the texture of the thinking. Does a project show a brief, a constraint, a wrong turn? Real work has scar tissue. A logo that was rejected twice before the client's finance director said something that cracked the whole problem open. A social campaign that tested badly with one audience and got reworked. Machine-made portfolios tend to arrive fully formed, without a single abandoned route, and that's strange if you think about it. Nobody's process is that clean.
There's no reliable detector for generated imagery, and I'd distrust anyone selling you one. Obvious tells (warped hands, garbled lettering, a strange sameness of light across unrelated projects) catch the lazy cases only. The better candidates edit those out. So stop hunting for artefacts and start asking for the thing a prompt can't hand over.
Beyond Tick Boxes: Diversity Recruitment Strategies That Actually Transform UK Workplaces
Master the Virtual Hot Seat: 7 Video Interview Techniques Recruiters Don't Tell You
AI Governance Careers Surge in 2026: Why This Role Will Define the Next Decade
What to ask for instead
Ask to see the middle. Rough sketches, early decks, the version that didn't make it, the email where a client pushed back. A candidate who made the work can talk about it for twenty minutes without repeating themselves, and can tell you what they'd change. Someone who curated outputs will keep circling the same three sentences about "clean, modern, human-centred" and then go quiet when you ask what the audience actually did.
Try a live constraint, too. Change the brief in the room: same idea, but now it's a six-second bumper, or a printed leaflet with no colour, or a tone suitable for a regulator's audience. Adaptability across platforms is the real skill hiding behind the buzzword, and it can't be pre-rendered. How someone reshapes an idea when you take half its tools away tells you more than any case study. The design portfolio trends piece on OHUB covers what directors say they're leaning towards, and it's a useful companion to this.
And look for a through-line. Someone with a point of view makes work that rhymes across projects even when the sectors differ. Someone assembling outputs makes work that looks like whatever the last prompt was.
The strongest objection
Here's the fair challenge, and it deserves a straight answer. Plenty of excellent creatives now use generative tools openly, and some of the most interesting work I see is made that way. If I'm marking down anything that touched a model, I'm punishing the people who've adapted fastest.
I'd concede the point completely. Using the tools isn't the red flag. Hiding it is, and so is being unable to say which parts were yours. A candidate who writes "concept, art direction and edit mine; first-pass visuals generated, then heavily reworked" is showing me exactly what I want to see: judgement applied to a fast tool. That's a skill, arguably a scarcer one than hand rendering.
But the concession doesn't dissolve the argument, it sharpens it. If everyone can generate, then the differentiator is the decision-making around the generation. Disclosure plus reasoning beats a flawless reel with no explanation every time.
Red flags that survive any tool
Some warning signs have nothing to do with AI. Every project is presented as a triumph with no constraint named. Credit is vague about who did what on team work (a common problem, and often innocent, but ask). Results are asserted without any account of how anyone knew. If a candidate claims a campaign "went viral" and can't tell you what viral meant for that client, that's a gap worth probing gently.
The flip side matters for job seekers. If you're building a portfolio now, show your working, label your tool use, and include one project that went wrong and what you took from it. It feels risky. It reads as confidence. You can see how recruiters weigh formats in this guide to portfolio formats that get design candidates hired, and if you're hunting for roles where that kind of evidence is valued, browse creative jobs on OHUB.
The CIPD has long argued that structured, evidence-based selection beats gut feel, and its guidance on selection methods applies neatly here: score the reasoning, not the render.
By 2027 the best portfolios will probably be shorter, rougher in places, and far more talkative about why. The candidates who pull ahead won't be the ones with the shiniest reel. They'll be the ones who can sit across from you and defend a choice.
