gettyAs digital products become more streamlined and automated, design choices that once felt helpful can sometimes create confusion, reduce transparency or limit user choice. Features that minimize effort can make workflows faster, but they can also obscure how decisions are made, what actions are happening behind the scenes, and how much control users actually have.For development teams, the challenge is finding the right balance between convenience and user agency. Below, members of Forbes Technology Council share UX and UI conventions designers should reconsider, along with the design priorities that can help users stay informed, confident and in control while interacting with modern digital products.User Intent In AI RewritesDesigners should reconsider AI rewrite buttons that replace the user’s original wording without showing what changed. It may feel helpful, but it can quietly remove the user’s intent, tone or responsibility. A better pattern is side-by-side review: original, suggested version, key changes, and a clear accept or reject choice. - Sibasis Padhi, Walmart Inc.Contextual Guidance Over Forced TutorialsForced onboarding tutorials have become ubiquitous and even recur when new updates are made. While the intention is often good, many tools are straightforward enough for the target market that they are unnecessary. Inline help on specific features would be more impactful, instead of a pop-up that ruins a user’s first impression and subverts their agency before they have a chance to use it. - Luke Wallace, Bottle RocketForbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?Human Choice Over Smart DefaultsOne UX convention designers should reconsider is the growing reliance on smart defaults being presented as the primary path. Make the human decision the primary surface and the AI recommendation an explicitly labeled assistive layer the user can opt into, inspect or ignore. The goal is not to remove AI from the workflow but to preserve informed intent instead of manufacturing passive consent. - Jeffrey Highman, TruaVisibility Into Automated ActionsOne UI convention designers should reconsider is overrelying on progressive disclosure—specifically, hiding too many controls, steps or options in the name of simplicity. As products become more automated, designers often strip away visible controls to make the interface feel clean, but this can backfire. Users don’t know what the system is doing behind the scenes, which is important. - Gaurav Vashisht, KrakenContext For Confidence ScoresConfidence scores aren’t new; they’ve been in interfaces for years. What’s changed is the stakes. When AI moves those scores into lending, claims or clinical decisions, users need more than a number. They need enough context to exercise real judgment. In regulated industries, that gap between score and understanding is where risk lives. - Jenny Larsson, Intact Insurance Specialty SolutionsTransparency Before One-Click AutomationDesigners should reconsider one-click automation without explanation. Streamlined flows can be useful, but when systems act before users understand what will happen, agency is reduced. Interfaces should make consequences visible: what will change, what data will be used and how users can reverse the action. Good UX should accelerate decisions, not obscure them. - Craig Davies, GathidClear Boundaries For AI AssistanceReconsider AI assistants that replace browsing and reading. With AI summaries, users get an answer but lose the ability to judge what was omitted or how contested it is. AI buttons often appear unannounced—users don’t know what they search, what data they use or what they favor. A tooltip explaining the AI button’s purpose and scope earns more trust than visual prominence. - Konstantin Klyagin, RedwerkDeliberate Review Of Prefilled DataModern design pushes toward zero friction, but that comes at a cost. When forms prepopulate fields, users skim instead of reading and submit incorrect data without noticing. The interface handles the thinking, so users gradually stop doing it themselves. Designers should break prefill into visible, confirmable steps or leave key fields blank. A slower input process here is often the right call. - Mateusz Mucha, Omni CalculatorAuditable AI RecommendationsStop hiding the AI’s reasoning. The clean “AI suggested this” card feels magical, but your users can’t audit it, override it or learn from it. They’re losing agency without noticing. The fix? Make every AI recommendation expandable: confidence, data, alternatives. Trust isn’t built by magic. It’s built by showing your work. - Varun J. Vincent, FalconFirst AIUser Control Over Inferred DataDigital products can become overly predictive and retain more context than users realize. Users should be able to see what the product has inferred about them, remove irrelevant signals and disable specific assumptions. For example, a one-time search for garden supplies on behalf of a neighbor should not turn into weeks of unwanted recommendations in that category. - Kostiantyn Gitko, Devox SoftwareExplanations For Restricted ActionsUX design that undermines understanding includes disabled form fields with no explanation. Every enterprise app is full of grayed-out buttons and locked inputs that tell users “no” without saying why. We build SaaS products, and this pattern generates more support tickets than almost any other UI choice. A grayed-out button that says nothing trains people to stop trying. One that says “requires admin approval” trains them to solve it themselves. - Denys Vorobyov, EltexSoftIntentional Pauses Before High-Stakes ActionsOne UX convention designers should rethink is the disappearing confirmation step. Modern products auto-save, auto-send and auto-execute in the name of speed. It feels seamless until a user realizes they can’t undo what just happened. Friction isn’t always bad design. For destructive or financial actions, a half-second of intentional pause restores agency. Speed without reversibility is hostile. - Nidhi Jain, CloudEagle.aiClear Paths For Failure ScenariosThe thing we should think about again is making things that only work when everything goes right. When something goes wrong, people have no idea what happened or what they should do next. Designing for when things go wrong is actually the way that people start to trust systems that they do not fully understand, like automation systems. - Maitrik Patel, AppleUser Jobs Over Screen OptimizationMost UX work optimizes the visible artifact: the screen, the flow, the layout. Instead, think about jobs-to-be-done. JTBD starts with the actual job that the user hired the product to do. Good design makes the next move obvious without explanation. In an AI-mediated interface, that obviousness must extend to certainty, uncertainty and risk. Design for the job, and the screen takes care of itself. - Michael Quoc, Product.aiVisible Usage CostsFlat “unlimited” pricing for AI products is worth reconsidering. It sounds generous, but it hides the real cost of computing and trains users to treat generation as disposable. Designers and founders should reconsider pricing UX that obscures usage—clear credits, visible consumption and honest tiers give users agency over what they spend and respect for what they create. - Sourabh Pateriya, Soundverse Inc.Adaptive Interfaces For Individual NeedsLet’s reconsider the one-size-fits-all design itself. We still build interfaces as if every user arrives with the same context, expertise and intent, and then we layer in personalization through cosmetic settings. The real next step is interfaces that adapt to the person in front of them: what they’re trying to do, what they already know, and what gets in their way. Not 1:N, but (1:1) × N. - Anna Drobakha, Groupe SEBUser-Controlled Content MovementAuto-playing carousels that move without the user clicking are a UX design that should be reconsidered. They take away control and often make people miss what they wanted to see. Designers should let users set the pace. That small change respects agency and understanding. - Rohan Pinto, 1Kosmos BlockIDSystemwide UX ThinkingDesigners should start thinking in terms of systems instead of pages. With more and more companies releasing headless solutions, UI design stops being tied to the platform. The backend becomes invisible, and the frontend can live anywhere. This means UX designers need to start thinking across multiple systems and focus on what data is needed, not what screens need to be designed. - Tal Frankfurt, Cloud for GoodClear Auto-Save Status SignalsOne design feature to reconsider is auto-save without visibility. It feels like a convenience feature until something goes wrong and users have no idea what version they’re in, what was changed or how to get back. Control isn’t just about having options—it’s about knowing the system’s state. Removing that feedback loop quietly erodes trust. - Harsh Jangid, Coozmoo Digital SolutionsNatural Stopping Points In Digital ExperiencesInfinite scroll deserves a rethink. It keeps people engaged, but it also strips away any sense of where they are or how much they’ve consumed. Without natural stopping points, users lose orientation and, with it, a sense of control. Pagination felt clunky, but it gave people a conscious moment to decide: Keep going or stop. Engagement without awareness isn’t good design. It’s a trap. - Marc Fischer, Dogtown Media LLC