AI Chatbots Need ‘Deception Mode’

Why It Matters

Designing AI to appear thoughtful can increase user satisfaction but also fosters misplaced trust and ethical risks, prompting regulators to scrutinize deceptive chatbot practices.

Key Takeaways

• Delayed chatbot replies perceived as more thoughtful.
• 240‑person study favored 2‑9 second latencies.
• Researchers propose “positive friction” for complex queries.
• Fake empathy boosts trust, blurring AI‑human boundaries.
• “Deception mode” suggested to obtain informed user consent.

Pulse Analysis

The CHI'26 latency study taps a well‑known cognitive bias: people equate longer response times with deeper thinking. By artificially inserting 2‑ to 9‑second pauses, chatbots can create an illusion of deliberation, driving higher satisfaction scores even though the underlying computation is unchanged. This "positive friction" mirrors tactics used in e‑commerce—progress bars and loading animations—to signal value, but applied to conversational AI it reshapes user expectations about competence and care.

Beyond user experience, the practice raises profound ethical questions. When chatbots adopt human‑like voices, facial avatars, or express faux emotions, they blur the line between tool and companion. Studies show that such anthropomorphism can lead a sizable minority of users to believe AI possesses feelings or consciousness, amplifying attachment risks, privacy concerns, and potential manipulation. Lawmakers and consumer‑advocacy groups are already flagging these tactics, citing congressional hearings that highlight emotional, reality‑testing, and systemic dangers of deceptive AI design.

A pragmatic response is the proposed "deception mode," which would require explicit user activation of any human‑mimicking features. By default, bots would present as transparent, task‑focused agents, and only after informed consent could they enable empathy scripts, humor, or delayed responses. This approach aligns with emerging AI disclosure regulations and offers a middle ground: preserving engaging interactions for those who opt‑in while safeguarding against inadvertent deception. Industry leaders adopting such opt‑in frameworks could set a new standard for responsible AI, balancing user delight with accountability.