Why taste is the only thing left that AI can't replicate
things AI can generate in the time it takes you to read this slide.
We solved the making problem. AI can generate a website, a brand identity, a marketing campaign, a product prototype in minutes. The bottleneck isn't creation anymore. The bottleneck is selection. When everything can be made, the only question that matters is: should it be?
Generate ten logos with AI. They're all competent. They're all forgettable. That's the problem with infinite supply and zero opinion. AI output converges on the mean. It gives you the statistical average of everything it's seen. Taste is what pulls output away from the average toward something specific, intentional, and memorable.
People confuse taste with personal preference. They're not the same thing. Preference is what you like. Taste is knowing why something works, who it's for, and whether it belongs. Taste is an informed opinion backed by a thousand references you've absorbed without realising it.
Bad taste is recoverable. It means you have opinions, even if they're wrong. You can refine wrong opinions. No taste is the real problem. No taste is indifference. It's the person who says 'I don't mind' to every design decision. The person who can't tell you why they chose that typeface or that colour. Indifference to quality is the one thing you can't train away. Hire someone with bad taste over someone with no taste every single time.
Your taste is visible in what you choose to keep. Not what you post or what you say you like. What you actually save, bookmark, screenshot, and send to a friend at 11pm. That collection is your taste fingerprint. It's unique to you, and it's more honest than any brand statement.
You can tell in five minutes of conversation whether someone has taste. Not from what they've made. From what they reference. The person who mentions a specific restaurant, a specific typeface, a specific album track rather than the hit single. Taste shows up in specificity. Vague admiration is preference. Precise enthusiasm is taste. Hire for the specifics.
hours of paying attention. That's what taste costs. Not a course. Not a framework. Just years of noticing, saving, and developing an opinion about why things work.
seconds. That's how long it takes to spot whether someone has taste. Look at their phone home screen.
A curator's job isn't to include everything. It's to exclude almost everything. The act of saying "not this, this" is the most valuable skill in an era of abundance. Editors have always known this. Magazines, galleries, record labels. They all sell the same thing: someone else's taste applied to your choices.
Magazine editors were the original taste professionals. They understood that a publication isn't defined by what it includes. It's defined by what it leaves out. Every page that doesn't make the cut sharpens the ones that do. The best digital products work the same way. What you remove is more important than what you add.
If you want to know what someone's strategic instinct actually is, don't read their LinkedIn bio. Look at what they've saved over the last six months. Their bookmarks, their Spotify, their reading list, their photo library. That's the unedited version of their taste. That's what they actually care about.
The algorithm shows you what it thinks you want. Your bookmarks show you what you actually are.
This is the entire argument in one sentence. Generative AI is the most powerful creative tool ever invented. And it has no opinion. It doesn't know what good looks like. It doesn't know when to stop. It doesn't know that less is usually more. You do. That's your job now.
Taste is politically expensive. It requires looking at work the team spent weeks on and saying this isn't good enough. It requires telling a client their favourite option is the wrong one. It requires killing your own ideas when they don't meet the standard. Most people avoid these conversations. That's why most output is mediocre. The willingness to be unpopular in service of quality is what separates taste from politeness.
Taste isn't saying 'I like it.' Taste is saying 'here's why it works, who it's for, and what it's doing that you haven't noticed yet.'
The best taste references don't come from screens. They come from a restaurant that got the lighting right, a bookshop that curates by mood not genre, a hotel that put thought into the weight of the room key. Physical experiences train your eye in ways that scrolling never will. Get away from the screen.
I've worked with Nike, Gucci, BMW, Google. All from London. Taste is not universal. What reads as refined in Tokyo reads as cold in New York. What feels bold in London feels aggressive in Zurich. Understanding how taste translates across cultures is its own skill. The global strategist doesn't impose one taste. They develop fluency in many.
Taste isn't one thing. It's a stack. It's your references (what you've consumed), your filters (what you keep), your standards (what you demand), and your instinct (what you feel before you can explain it). You build this stack over years. It compounds. It can't be automated.
In an era of infinite content, the person who finds it, filters it, and frames it is more valuable than the person who makes more of it. Curation is creative work. It requires taste, point of view, and the confidence to say this matters and this doesn't. The next generation of influential voices won't be the loudest. They'll be the most selective.
People treat taste as a soft skill. Something nice to have. But taste drives commercial outcomes. The product with better taste gets shared more, returned less, and remembered longer. Apple's margin isn't from better chips. It's from better taste applied to every decision from packaging to typography. Taste is the most under-measured driver of revenue in business.
Growth hacking. Move fast and break things. Performance marketing. Conversion rate optimisation. For a decade, the dominant business philosophy was actively hostile to taste. Speed over craft. Data over instinct. And it worked, for a while. Until everything started looking the same. The pendulum is swinging back. Brands that invested in taste during the efficiency era are the ones people remember now. Taste is patient capital.
Every product I've built is a taste decision. What to make. How it looks. What to leave out. The Relevance Index scores cultural salience. Modern Retro curates what deserves a second look. Oishii London filters thousands of restaurants to the ones that matter. Twenty-plus products, each one a taste argument made tangible.
products built with AI where taste was the only differentiator. Same tools as everyone else. Different output.
Read widely outside your industry. Save everything that stops you scrolling. Visit physical spaces that make you feel something. Study the people whose work you admire and understand why you admire it. Develop opinions and defend them. Taste isn't innate. It's a practice. And like any practice, it improves with repetition.
Tools get cheaper. AI gets smarter. Templates get better. Distribution gets democratised. The only thing that doesn't get commoditised is the ability to look at ten options and pick the right one. Or better yet, to know the right one before you've seen it. That's taste. And it's the last unfair advantage.
decision. That's what taste comes down to. Not a framework or a scorecard. One decision at a time. This typeface, not that one. This word, not that one. This feature stays, that one goes. A thousand small taste decisions is what separates forgettable from remarkable.
Taste as competitive moat assumes a stable culture. But taste is contextual, contingent, and often exclusionary. What reads as discernment in one community reads as gatekeeping in another. And if taste is learned, not inherited, then AI tutors could democratise it faster than any human mentor.
If everyone has taste, it stops being an advantage. It becomes table stakes.
This deck argues taste is the last unfair advantage. Andjelic goes further: taste isn't just an advantage, it's the entire basis of the new status economy. If taste wins, her work shows exactly what the winners look like.
Everyone has the same tools now. The question is no longer can you make it. It's should you.
In a world where anyone can make anything, the only question that matters is: does this deserve to exist?