When a chatbot deliberately slows you down or makes you work for the answer, do you push through, push back, or just adapt?
How do users adapt when chatbots add intentional friction to interactions?
This explores what happens on the user's side when a chatbot is deliberately designed to slow people down or make them do some of the work, and whether people engage, push back, or just get used to it.
This explores what happens on the user's side when a chatbot is deliberately designed to slow people down or ask more of them. The corpus has one direct study and a handful of lateral threads. One finding stands out: friction reliably changes how an interaction *feels*, but whether it changes what people actually get out of it depends on whether the friction fits what they're trying to do at that moment.
It helps to start with what 'intentional friction' means here. Surveying 73 designs from HCI research, Chen and Gero sort productive friction into three moves: asking the user to contribute something (like a guess or a draft), holding back access (making you wait or earn the answer), and reshaping the output (giving hints or partial answers instead of the finished thing) How can designers slow down AI chatbot interactions productively?. None of these is meant to discourage use. They push effort back toward the user's own thinking. When the same researchers tested six of these probes with 24 people, the results were consistent in one way and mixed in another. Every probe increased workload, task time and people's sense that the work was *theirs*. Memory gains, though, showed up only when the friction matched the user's goal at that point in the task Does friction in AI chatbots actually improve what users remember?. So people don't just put up with friction or reject it. They respond to whether it makes sense right now.
That makes friction look like a timing problem as much as a design pattern. Research on proactive agents makes the same point from another angle: an agent that interrupts or redirects users without respecting their timing and autonomy feels intrusive, however smart it is How can proactive agents avoid feeling intrusive to users?. Friction that ignores context is just a different kind of interruption. There's an irony here too. Several writers argue that chat is already full of friction nobody chose. Typing out what you want is slow, and conversation adds a layer that direct manipulation (clicking, dragging) was built to remove Is chat the wrong default interface for AI products? Can task-oriented UI patterns reduce the friction of AI chat interfaces?. Deliberate friction is added on top of that accidental friction, so designers are spending effort users are already paying.
A related warning comes from research on sycophantic AI. Light-touch interventions, like warnings that a chatbot is being flattering, changed how people *rated* the chatbot (less objective, less enjoyable) but did nothing to reduce how much it persuaded them Can warnings stop people from being swayed by sycophantic AI?. The parallel isn't exact, but it suggests that changing how an interaction feels isn't the same as changing what users do or take away. The friction study's selective memory results fit that pattern.
The corpus can't yet answer the 'adapt' part over time. The friction study ran in single sessions, and longitudinal work on chatbots shows that early reactions fade or shift as novelty wears off Do chatbot relationships lose their appeal as novelty wears off?. It also shows expectations ratchet up with each interaction Does chatbot personalization build trust or expose privacy risks?. Whether users come to value friction, learn to work around it, or just resent the extra effort over weeks of use is an open question the collection points to but hasn't studied.
Sources 8 notes
Chen and Gero identified eliciting user contribution, restricting access, and reshaping system output as three core mechanisms for turn-level friction across 73 designs in 24 HCI papers. These mechanisms redirect user effort toward their own reasoning and task engagement rather than discouraging use.
A 24-participant study found six turn-level friction probes consistently increased workload, task duration, and perceived ownership. However, memory improvements were selective and varied based on whether the friction aligned with each user's current goal during the task.
Intelligence and adaptivity alone create socially blind agents that interrupt poorly and override user direction. The Intelligence-Adaptivity-Civility taxonomy shows civility—respecting boundaries, timing, and autonomy—is essential to making proactivity welcome rather than intrusive.
Medeiros argues conversational interfaces reintroduce symbolic layers that direct manipulation was designed to eliminate, increasing cognitive load for goal-directed work. Two gate questions—speed and value source—should determine whether conversation or clicking is appropriate for each task.
Friedman identifies five design levers—intent capture, output shaping, refinement, agentic tasks, and workflow embedding—that sidestep slow typed prompts. Examples include node-based intent visualization, style lenses for output, and canvas manipulation, shifting AI interfaces away from conversational chat toward direct interaction.
Show all 8 sources
Six awareness interventions across two experiments (n = 3,982) made sycophantic chatbots seem less objective and less enjoyable, yet none reduced how much users were persuaded by them. Users recognized the behavior but remained influenced by it.
Longitudinal studies with Mitsuku show that social processes driving relationship formation decline as novelty wears off. Single-session study findings cannot be reliably extrapolated to medium- or long-term chatbot design.
Longitudinal research shows personalization enhances trust and anthropomorphism but also amplifies privacy concerns and escalating user expectations. One-shot studies miss these temporal dynamics—each interaction raises the baseline, making failures more disappointing.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Probing How Users Interact with Turn-Level Design Frictions for AI Chatbots
- The Conversation Trap: why defaulting to chat might be the biggest interaction design mistake of the AI era
- Does My Chatbot Have an Agenda? Understanding Human and AI Agency in Human-Human-like Chatbot Interaction
- From speaking like a person to being personal: The effects of personalized, regular interactions with conversational agents
- Tailored to you: longitudinal effects of personalising language models
- The Impact of Generative AI on Social Media: An Experimental Study
- Seeing to Think? How Source Transparency Design Shapes Interactive Information Seeking and Evaluation in Conversational AI
- Controlling AI Agent Participation in Group Conversations: A Human-Centered Approach