AI Abundance Needs Curation, Not More Output
2026-05-24 · 4 min read · Janaina Maia
AI is making it much easier to create more content. That is not the same as creating a better experience.
TechCrunch criticised Spotify’s AI direction as a move toward more automated playlists, more generated recommendations, and potentially more synthetic audio in a product that many users already experience as crowded. The critique is not only about Spotify. It is about a broader pattern in AI products: once generation gets cheap, companies are tempted to fill every surface with more.
That is where product teams need to be careful.
More is not a strategy.
When AI can produce endless summaries, playlists, images, drafts, emails, insights, reports, or recommendations, the bottleneck changes. The hard part is no longer making something appear. The hard part is deciding what deserves the user’s attention.
This is especially important in products built around taste, trust, or expertise. A music app is not valuable because it can generate an infinite queue. It is valuable because it helps someone find the right thing for this moment. An enterprise AI product is not valuable because it can produce twenty analyses. It is valuable because it helps a team understand which analysis is credible, relevant, and worth acting on.
Cheap output raises the bar for curation.
The product risk is content inflation.
Content inflation happens when the amount of generated material grows faster than the user’s ability to judge it. The interface starts to feel productive, but the human experience gets heavier. There is more to scan, more to distrust, more to dismiss, and more to clean up.
This can look harmless in consumer products: another playlist, another recommendation row, another auto-made feed. In workplace products, it becomes more serious. AI can flood teams with meeting summaries, duplicated research notes, synthetic insights, risk flags, draft decisions, and generated tasks. If those outputs are not clearly ranked, sourced, governed, and owned, AI does not reduce work. It relocates the work to human filtering.
That is not intelligence. That is mess with better typography.
Users need agency, not just personalisation.
Personalisation often gets framed as “the system knows what you want.” But in AI-heavy products, users also need the ability to shape what the system should optimise for. Do I want novelty or familiarity? Efficiency or exploration? Official sources or broader inspiration? A quick draft or a careful answer? Automation or review?
Those preferences should not be hidden inside a model’s guess about the user. They should become part of the experience. The user should be able to steer the system’s taste, narrow its scope, reject patterns, and understand why something was shown.
For designers, this is the difference between a feed that acts on people and a product that collaborates with them.
Design implication: build curation surfaces.
If AI generation is part of the product, I would look for the curation surface before I look at the generation surface.
- Why this? Show the signal behind a recommendation, summary, or generated artifact.
- What was used? Make sources, inputs, and constraints visible enough to inspect.
- What can I tune? Let users shape intent, strictness, freshness, risk tolerance, and review depth.
- What should disappear? Give users meaningful ways to remove noise, not just consume more.
- Who owns the final call? Make human approval clear when generated output affects real decisions.
Without these surfaces, AI products risk turning abundance into a tax on attention.
My take.
The next product advantage will not come from generating more things. Everyone will be able to do that.
The advantage will come from knowing what not to show, what not to automate, what not to trust yet, and where the user needs control instead of another clever suggestion.
AI makes output cheap. Product design has to make judgement visible.