Search by topic, publication type, year, open access, author, journal, and DOI.
Free-text queries are sent to OpenRouter and VoyageAI only after you submit them. Privacy
Conversation quality
Recent research describes a recurring problem in personal AI conversations: models sound empathetic but fall back into the same supportive role across very different situations. The missing quality is not more warmth, but a conversational posture that fits the moment.
First publishedRelationship & impact
A study published in Science shows why uncritical affirmation in personal conflicts is not merely a matter of style: it can shift judgments, weaken accountability, and still be perceived as especially helpful.
First published RevisedTests & data
Conversation quality does not emerge from a single sentence. An overview of 125 studies, along with longitudinal and safety benchmarks, shows why technology, fit, relationship impression, memory, responsibility, and trajectory must be assessed separately.
First published RevisedLaw & ethics
A USENIX study shows that understandable deletion options for sensitive AI conversations can achieve more than technically sophisticated privacy promises. This makes data control part of conversation quality.
First publishedRelationship & impact
Between helpful use, trust, and emotional dependence, there is no clean dividing line. Long-term data and new research frameworks show why immediate relief, autonomy, fabricated user profiles, and social consequences must be examined separately.
First published RevisedModels & technology
System prompts define role, boundaries, and conversational style before a human writes the first word. New research shows what transparency and control users want—and why a published prompt alone does not yet yield a reliable AI.
First publishedConversation quality
Eva Gengler’s book shifts the AI debate from isolated biased outputs to power, purpose, and participation. Independent research supports this critique—but not the notion that feminist AI is already a finished technical method.
First publishedConversation quality
A new preprint separates helpful regulation from unwanted escalation in 9,000 model responses. During venting, both increased simultaneously: a response could appear attentive and still reinforce certainty, partisanship, or emotional intensity.
First publishedConversation quality
A mental-health chatbot can seem cautious in its first response and still overstep its role later. Two studies show why conversational boundaries must be examined across entire conversation trajectories.
First publishedTherapy & mental health
A representative US survey shows how widespread AI chatbots for mental health questions already are among 12- to 21-year-olds – and how rarely others learn about it.
First publishedConversation quality
A new experimental study shows: Even the mere availability of mostly incorrect AI hints can deter people from admitting uncertainty. What this means for personal conversational systems—and what the study does not prove.
First publishedTherapy & mental health
A new study finds dangerous dietary advice even behind warning sentences and seemingly caring language. Why rejection alone is not a sufficient safety measure.
First publishedTests & data
A CHI 2026 study compares GPT-4o, GPT-5, and highly rated Reddit advice. The results are strong – but narrower than the headline suggests.
First publishedTherapy & mental health
AI conversations can facilitate access and appear convincing. However, the lead publication does not provide an efficacy test, but rather a broad problem map. For mental health care, this implies a strict standard: linguistic quality, perceived empathy, and technical reach must not be confused with proven benefit.
First publishedTherapy & mental health
AI-based conversational systems can reduce depressive symptoms and distress. However, the research does not thereby demonstrate broad psychological benefit or equivalence with human treatment. Those who nevertheless speak of replacement are confusing a limited result with a judgment about care.
First publishedTherapy & mental health
A randomized study of 148 young adults found lower depression scores after one week for the CBT-based conversational system XiaoE than for an e-book or a general assistant. What matters, however, is what does not follow from this: measured working alliance is neither a human relationship nor evidence of long-term care.
First published RevisedTherapy & mental health
Empathetic language can create the impression of reliable help. But what is ethically decisive is not only how convincingly an AI system responds, but who takes responsibility for its use, sets boundaries, protects data, and answers for harm. It is precisely this responsibility that often remains vague in the debate.
First publishedTherapy & mental health
A large review maps the ethical conflicts of AI-assisted conversations in mental health care. Its most important finding is not a ranking of technical deficiencies. Rather, safety, confidentiality, efficacy, and fairness depend on an unresolved preliminary question: What role may a system assume at all?
First published