Does social presence from robots drive adherence better than conversational AI interfaces?

This explores whether a robot's physical presence beats a chatbot at keeping people engaged and following through — and the corpus has a surprisingly direct answer, plus several reasons why. The sharpest evidence comes from a 15-day study where robots and paper worksheets meaningfully reduced students' psychological distress while a chatbot running the *identical* language model did not Why do robots outperform chatbots in therapy despite identical language models?. The striking part is that language capability was held constant — same LLM on both sides — so whatever drove adherence wasn't smarter conversation. It was the medium itself: physical co-presence and a structured format. That's the closest thing here to a clean test of your question, and it tilts toward robots.

But the corpus suggests the real variable isn't 'robot vs. chatbot' so much as *the quality of social cues* a system can deliver. One line of research finds that a single strong primary cue — a voice, an appearance — is enough to evoke the sense of a social actor, while piling on many weaker secondary cues does not Do more social cues always make AI feel more present?. A robot embodies a primary cue almost by default; a text chatbot has to manufacture presence indirectly. So robots may win less because they're robots and more because embodiment is a cheap, reliable route to the kind of cue that actually moves people.

There's also a time dimension that complicates any 'robots win' verdict. Chatbot relationships ride a novelty wave that decays predictably across repeated sessions, which means single-session enthusiasm doesn't extrapolate to long-term adherence Do chatbot relationships lose their appeal as novelty wears off?. Yet the opposite can also happen: in repeated partner-selection games, people started biased against AI but *grew* to prefer AI partners once they learned the bots behaved reliably and prosocially Do humans learn to prefer AI partners over time?. So presence gets you in the door, but sustained adherence may hinge on whether the system proves dependable over time — something neither embodiment nor charm guarantees on its own.

What makes 'social presence' work is itself not one thing. Research on conversational alignment finds that different channels do different jobs: lexical matching drives task efficiency and comprehension, while emotional and prosodic alignment drive warmth and trust — and conflating them produces cold service bots and evasive therapy bots Do different types of alignment serve different conversational goals?. Users also model their AI partners along three axes — competence, human-likeness, and flexibility — with *competence* dominating their impressions How do users mentally model dialogue agent partners?. That's a quiet warning for the robot story: a charismatic embodied agent that users judge incompetent may lose adherence anyway.

The takeaway you might not have expected: the therapy result implies the 'active ingredient' in adherence isn't conversational intelligence at all — it's presence plus structure. A worksheet beat a chatbot too. That reframes the design question from 'how do we make the AI talk better' to 'what scaffolding makes a person keep showing up,' and embodiment turns out to be one effective scaffold among several rather than a magic one.