UPDATING · 00:00 BEIJING
SUNDAY · MAY 31, 2026 EN
A daily map of how HCAI
research directions, questions
and findings evolve

HCAI Research Radar

A daily map of Human-Centered AI research · directions · questions · papers
Edited from open scholarly
data sources · OpenAlex ·
arXiv · Semantic Scholar · Crossref
Today
12
vs yesterday + 3
Past 7 Days
86
vs prev 7d + 18%
Past 30 Days
312
vs prev 30d + 11%
Directions
18
primary · 142 sub-questions
New questions
24
last 30 days · + 9
Last Run
00:00
BEIJING · MAY 31
all 4 sources OK
· · ·
From today's pull · 00:00 Beijing

Papers of the Day

12 entries · sorted by HCAI score
01

Calibrating Reliance: How Clinicians Use AI-Generated Explanations in Diagnostic Decisions

Maya R. Kapoor, Lin Wei, A. P. Sundaram, Daniel Foster, Sarah Greene, J. Mukherjee, R. Park, et al.
Trust & Reliance Explainable AI AI-assisted Decision arXiv preprint

A mixed-methods study with 84 radiologists across two hospitals examines how confidence intervals in AI explanations shift diagnostic agreement — finding under-reliance dominates ambiguous cases…

94
HCAI Score
2026·05·31
02

When the Agent Drives: Designing Handoff Moments in LLM-Powered Co-Programming

Jonas Berg, T. Yamamoto, Pei Chen, Eleanor Whitfield, M. de Sá
AI Agents LLM User Study Mixed-Initiative

Through three weeks of diary studies with 24 software engineers, the authors propose four canonical handoff patterns between human and agent — and their failure modes in production environments…

91
HCAI Score
2026·05·31
03

"It Listens, But Does It Hear?" — Affective Mismatch in AI Mental-Health Chatbots

Wen-Hua Zhou, R. Banerjee, K. Iwata, Sofia Allen
Single-school AI in Mental Health Affective Computing LLM User Study

A 6-week field deployment with 142 university students documents systematic gaps between user-perceived empathy and clinician-rated appropriateness, with implications for safety guardrails…

89
HCAI Score
2026·05·30
04

Algorithm Appreciation Reconsidered: When Users Prefer AI Despite Equal Accuracy

Hannah Brooks, J. Tanaka, Ravi Iyengar, S. Ostrowski
Trust & Reliance Algorithm Aversion Human Factors

A pre-registered replication of Logg et al. (2019) across four task domains finds appreciation effects reverse under accountability framing — challenging the algorithm-appreciation generalization…

87
HCAI Score
2026·05·30
05

Steering Multiple Minds: A Control Interface for Human Oversight of AI Agent Teams

Ada Nakamura, Ben Olson, K. Pillai, Y. Zhao, M. Rashid
Multi-Agent Collaboration AI Agents AI UX Evaluation affiliation pending

Introduces a visual control language for monitoring and intervening in agent teams; evaluated through controlled experiments with 96 participants across three task complexities…

82
HCAI Score
2026·05·29
06

Teachers as Co-Designers: Adapting Generative AI for Inclusive Classrooms

L. Ferreira, Mei Lin, J. Hardcastle, P. Quattrocchi
AI in Education Human-AI Collaboration Responsible AI

Participatory design with 18 teachers across two countries surfaces tensions in adapting generative tools — including pedagogical authority, transparency, and equity considerations…

78
HCAI Score
2026·05·29
Hot & Emerging

Active Directions

01
Human-AI Interaction
primary direction
68
↑ 10
02
Trust & Reliance
trust calibration
49
↑ 11
03
AI Agents
high-growth
44
↑ 13
04
Explainable AI
with user study
35
↑ 4
05
AI-assisted Decision
decision support
31
↑ 5
06
LLM User Study
large-model UX
29
↑ 8
07
Multi-Agent Collaboration
team UX
23
↑ 9
08
AI in Mental Health
user-facing
20
↑ 2
09
Responsible AI
user / society lens
17
↑ 2
10
AI in Education
human-in-the-loop
15
↑ 3
Emerging questions

Emerging Questions

Human Control of AI Agents
under AI Agents
18
+ 120%
LLM Feedback in Education
under AI in Education
15
+ 75%
Overreliance on AI Explanations
under Trust / XAI
12
+ 60%
Agentic Workflow UX
under AI Agents
11
+ 58%
Sycophancy in Chatbots
under LLM User Study
9
+ 50%
Topic heat · last 30 days

Topic Heat

Human-AI Interaction
68
Trust & Reliance
49
AI Agents
44
Explainable AI
35
AI-assisted Decision
31
LLM User Study
29
Multi-Agent Collab
23
AI in Mental Health
20
Responsible AI
17
AI in Education
15
· · ·
Latest high-relevance

Latest High-Relevance Papers · HCAI ≥ 85

Sorted by HCAI score · across all directions
94HCAI
Calibrating Reliance: How Clinicians Use AI-Generated Explanations in Diagnostic Decisions
Directions · Trust & Reliance / XAI / Decision Support
Question · whether users over-rely on AI explanations · 84 radiologists · user experiment
2026·05·31
91HCAI
When the Agent Drives: Designing Handoff Moments in LLM-Powered Co-Programming
Directions · AI Agents / Human Control / Mixed-Initiative
Question · how humans control AI agents · 24 engineers · diary study 3 weeks
2026·05·31
89HCAI
"It Listens, But Does It Hear?" — Affective Mismatch in AI Mental-Health Chatbots
Directions · AI in Mental Health / Affective Computing
Question · gap between perceived empathy and clinical appropriateness · 142 students · 6-week field
2026·05·30
87HCAI
Algorithm Appreciation Reconsidered: When Users Prefer AI Despite Equal Accuracy
Directions · Trust & Reliance / Algorithm Aversion
Question · how accountability framing shifts reliance · pre-registered replication · 4 task domains
2026·05·30
86HCAI
Persuasive Defaults: How AI-Generated Recommendations Reshape Consumer Reasoning Online
Directions · AI UX Evaluation / Human Factors
Question · persuasive effects of generative defaults · N=1,247 · three experiments
2026·05·28
END OF FRONT PAGE · TURN TO PAPERS LIST →
Filterable database

All Papers

312 papers · last 30 days
REGION All US UK / EU CN JP SG
312 papers total · showing 1–10
SORT Latest HCAI Score Citations Schools
01

Calibrating Reliance: How Clinicians Use AI-Generated Explanations in Diagnostic Decisions

Maya R. Kapoor, Lin Wei, A. P. Sundaram et al. · Stanford / CMU / U-Washington
Trust & Reliance Explainable AI HI score ≥ 90 arXiv:2406.18234·2026·05·31

A mixed-methods study with 84 radiologists examines how AI confidence shifts diagnostic agreement, finding under-reliance dominates ambiguous cases…

94
HCAI Score
2026·05·31
02

When the Agent Drives: Designing Handoff Moments in LLM-Powered Co-Programming

Jonas Berg, T. Yamamoto, Pei Chen et al. · CMU / Georgia Tech
AI Agents LLM User Study CHI 2026 LBW·2026·05·31

Diary studies with 24 software engineers propose four canonical handoff patterns and their failure modes in production environments.

91
HCAI Score
2026·05·31
03

"It Listens, But Does It Hear?" — Affective Mismatch in AI Mental-Health Chatbots

Wen-Hua Zhou, R. Banerjee, K. Iwata · Zhejiang University
Single-school AI in Mental Health Affective Computing CSCW 2026·2026·05·30

A 6-week field deployment with 142 students documents systematic gaps between user-perceived empathy and clinician-rated appropriateness.

89
HCAI Score
2026·05·30
04

Algorithm Appreciation Reconsidered: When Users Prefer AI Despite Equal Accuracy

Hannah Brooks, J. Tanaka, Ravi Iyengar · Northwestern / U-Tokyo
Trust & Reliance Algorithm Aversion CSCW 2026·2026·05·30

A pre-registered replication of Logg et al. (2019) across four task domains finds appreciation effects reverse under accountability framing.

87
HCAI Score
2026·05·30
05

Persuasive Defaults: How AI-Generated Recommendations Reshape Consumer Reasoning Online

A. Bhattacharya, L. Tremblay, R. Goh · U-Michigan / Berkeley / NUS
AI UX Evaluation Human Factors CHI 2026·2026·05·28

Three experiments (N=1,247) show LLM-generated framings systematically narrow consideration sets even when users believe they are unaffected.

86
HCAI Score
2026·05·28
06

Steering Multiple Minds: A Control Interface for Human Oversight of AI Agent Teams

Ada Nakamura, Ben Olson, K. Pillai · MIT / Cornell / ETH / U-Tokyo
Multi-Agent Collaboration affiliation pending UIST 2026·2026·05·29

A visual control language for monitoring and intervening in agent teams, evaluated with 96 participants across three task complexities.

82
HCAI Score
2026·05·29
07

Mental Models of Generative AI: A Cross-Cultural User Study

Y. Saito, P. Andersson, M. Chowdhury · U-Tokyo / KTH / IIIT-D
Human-AI Interaction LLM User Study arXiv:2405.14021·2026·05·27

Interviews with 60 participants across Japan, Sweden and India reveal divergent mental models for LLM agency, error attribution and trust formation.

81
HCAI Score
2026·05·27
08

Teachers as Co-Designers: Adapting Generative AI for Inclusive Classrooms

L. Ferreira, Mei Lin, J. Hardcastle · UCL / NTU
AI in Education Responsible AI CHI 2026·2026·05·29

Participatory design with 18 teachers across two countries surfaces tensions around pedagogical authority, transparency and equity.

78
HCAI Score
2026·05·29
09

The Quiet Hand-Off: Cognitive Load Patterns in AI-Assisted Surgical Robotics

R. Singh, T. Sato, Y. Liu · Imperial College / U-Tokyo / SJTU
Human Factors AI-assisted Decision FAccT 2026·2026·05·26

Eye-tracking and EEG data from 28 surgeons in simulated procedures characterize cognitive load shifts at agent-takeover moments.

77
HCAI Score
2026·05·26
10

Beyond Accuracy: Trust Calibration in Long-Term AI Companion Apps

S. Park, N. Oliveira, F. Müller · KAIST / EPFL / TUM
Trust & Reliance Affective Computing CSCW 2026·2026·05·25

12-month longitudinal data from 312 users shows trust trajectories diverging at three distinct interaction phases, contradicting accuracy-driven models.

75
HCAI Score
2026·05·25
‹ Prev 1 2 3 32 Next ›
Phase I · 18 primary directions

Research Directions

Default · by 30-day paper count
SCOPE All 18 Strong 22 Mid 13 SORT 30-day 7-day Growth Avg score
# Direction Tier Today 7-day 30-day 1-year Questions Growth Ø Score Top Questions
01 Human-AI Interactionprimary direction strong 51868412 24 + 17% 82 collaboration · role allocation · control · mental models
02 Trust & Reliancetrust calibration strong 31149204 18 + 29% 84 overreliance · calibration · appreciation · explanation effects
03 AI Agentshigh-growth strong 41244186 16 + 42% 81 human control · handoff moments · agentic workflow
04 Explainable AIwith user study strong 2935198 14 + 13% 79 understanding · length effect · confidence display
05 AI-assisted Decisiondecision support strong 2831156 12 + 19% 80 role allocation · acceptance · override behavior
06 LLM User Studylarge-model UX strong 1729142 13 + 38% 78 sycophancy · mental models · cross-cultural use
07 Multi-Agent Collab.team UX strong 162396 9 + 100% 81 oversight · team task · trust propagation
08 AI in Mental Healthuser-facing mid 1520108 10 + 11% 76 affective mismatch · help-seeking · long-term trust
09 Responsible AIuser / society lens strong 141788 8 + 13% 77 accountability · fairness perception · transparency needs
10 AI in Educationhuman-in-the-loop mid 141594 9 + 25% 75 LLM feedback · pacing · teacher re-design
11 Affective Computingaffect / emotion mid 031476 7 + 0% 74 recognition · empathy design · mis-aligned responses
12 Human Factorscognitive ergonomics strong 031282 9 + 33% 78 cognitive load · situation awareness · error recovery
13 Algorithm Aversionaversion / appreciation strong 131154 6 + 83% 81 accountability framing · accuracy perception · acceptance
14 AI UX EvaluationUX of AI strong 031068 7 + 11% 76 evaluation frameworks · heuristics · long-term use
15 Mixed-Initiativeinitiative-switching strong 02952 6 + 13% 79 initiative handoff · control granularity
Showing top 15 of 18 directions · TURN PAGE FOR MORE
Section · Trust & Reliance

Calibrating Reliance: How Clinicians Use AI-Generated Explanations in Diagnostic Decisions

1Stanford HAI · Stanford University
2Human-Computer Interaction Institute · Carnegie Mellon University
3Paul G. Allen School of CSE · University of Washington
Trust & Reliance Explainable AI AI-assisted Decision Making Human Factors arXiv preprint HI Score ≥ 90

We report on a mixed-methods study examining how confidence intervals embedded in AI-generated explanations shift diagnostic decisions in clinical radiology. Across 84 radiologists at two hospitals, we find that the dominant failure mode under ambiguity is not over-reliance but a systematic under-reliance — clinicians discount well-calibrated AI outputs when explanation length exceeds working-memory thresholds. We articulate three design principles for reliance calibration: temporal pacing, evidence framing, and contextual disagreement surfacing. Implications for clinical AI deployment and explanation interface design are discussed.

Why HCAI · Classification Reasoning

This paper satisfies several strong-relevance criteria simultaneously: the study object is real human interaction with an AI decision-support system; it presents a mixed-methods empirical study with 84 clinical users; it focuses on core research questions including trust calibration, explanation understanding, and reliance behavior; and it derives interpretable UX-oriented design principles. The abstract is free of algorithm-benchmark bias and surfaces multi-dimensional content — user experiment, AI system type, and evaluation metrics (trust, reliance, cognitive load). Across 7 scoring dimensions the paper receives 94.

Similar Work in This Direction

Beyond Accuracy: Trust Calibration in Long-Term AI Companion Apps
KAIST / EPFL / TUM · score 75 · 2026·05·25
Algorithm Appreciation Reconsidered: When Users Prefer AI Despite Equal Accuracy
Northwestern / U-Tokyo · score 87 · 2026·05·30
The Quiet Hand-Off: Cognitive Load in AI-Assisted Surgical Robotics
Imperial / U-Tokyo / SJTU · score 77 · 2026·05·26
From Disagreement to Decision: A Framework for Surfacing Clinician–AI Conflict
UCL / Oxford · score 73 · 2026·05·22
PRIMARY DIRECTION · TIER · STRONG · RANK 02 · 30-day

Trust & Reliance

trust calibration in AI
DEFINITION · Research on how users form, calibrate, and adjust trust in AI
outputs — including over-reliance, under-reliance, and how explanation design,
confidence display, and accountability framing shape reliance behavior.
ALSO INDEXED AS · Trust in AI · Reliance · Overreliance · Algorithm Aversion · Algorithm Appreciation
11
7-day
49
30-day
204
1-year
18
Questions
+ 29%
vs prev 30
84
Ø HCAI Score
Question composition

Research Questions

204
papers · 1Y
Over- / under-reliance 57 · 28%
Trust calibration 43 · 21%
Explanation effects 33 · 16%
Aversion / Appreciation 27 · 13%
Accountability framing 22 · 11%
Others (6 questions) 22 · 11%
Content structure

Frequent Content

User experimentmethod68%
Survey + behavior logmethod31%
Healthcare decisioncontext28%
AI writing / coding aidcontext19%
Recommender decisionscontext14%
Cliniciansuser group22%
General usersuser group51%
trustmetric91%
reliancemetric62%
cognitive loadmetric28%
Representative papers

Recent Papers

Last 7 days · 11 papers
TIME 7-day 30-day 90-day 1-year QUESTION All Overreliance Calibration Explanation Accountability
01

Calibrating Reliance: How Clinicians Use AI-Generated Explanations in Diagnostic Decisions

Maya R. Kapoor, A. P. Sundaram et al. · Question · do users over-rely on AI explanations
Trust & Reliance User experiment Healthcare N = 84 clinicians arXiv preprint·2026·05·31
94
HCAI
02

Algorithm Appreciation Reconsidered: When Users Prefer AI Despite Equal Accuracy

Hannah Brooks, J. Tanaka et al. · Question · accountability framing & reliance
Trust & Reliance Pre-registered replication 4 task domains CSCW 2026·2026·05·30
87
HCAI
03

Beyond Accuracy: Trust Calibration in Long-Term AI Companion Apps

S. Park, N. Oliveira et al. · Question · trust divergence over long-term use
Trust & Reliance 12-month longitudinal N = 312 CSCW 2026·2026·05·25
75
HCAI
04

From Disagreement to Decision: A Framework for Surfacing Clinician–AI Conflict

M. Patel, L. Akhter et al. · Question · how to surface disagreement
Trust & Reliance Prototype Healthcare FAccT 2026·2026·05·22
73
HCAI
Adjacent directions · most co-tagged

Adjacent Directions

Explainable AI
32co-occurrence
AI-assisted Decision
28co-occurrence
Human-AI Interaction
24co-occurrence
Algorithm Aversion
18co-occurrence
Human Factors
14co-occurrence
Showing 4 / 49 · LOAD MORE PAPERS →
Phase II · How directions relate

Direction Map

18 primary directions · relationships derived from shared questions, methods, and application contexts

TIME 30-day 90-day 1-year RELATION All Co-occurrence Method borrowing Evolution TIER All Core Method Applied
COREMETHOD / EVALAPPLIEDHuman-AI InteractionTrust & RelianceExplainable AIAI-assisted DecisionAlgorithm AversionAI AgentsMulti-Agent Collab.Mixed-InitiativeHuman FactorsAI UX EvaluationLLM User StudyResponsible AIAI in Mental HealthAI in EducationAffective ComputingSocial RoboticsGenerative AI ToolsAI-mediated Comm.

Legend

Core directions (trust / decision / explanation / interaction)
Method / evaluation / factors
Applied contexts (mental health / education / affect)
Within-cluster
Cross-cluster (evolution / borrowing)
Node size · 1Y paper count

Relationship Insights

Most centralHuman-AI Interaction · degree 11
Strongest pairTrust × XAI · 32 co-occurring papers
Fastest evolutionAI Agents → Multi-Agent Collab. · + 100%
Newest linkLLM User Study × Responsible AI · 9 / 30d
Isolated0 · every direction has neighbors
Top Direction Co-occurrences

Top Co-occurring Pairs

Last 12 months · by joint paper count
Direction pair Main shared content Co-occur (1Y)
Trust & Reliance × Explainable AIcross-clusterexplanation effects on trust32
Trust & Reliance × AI-assisted Decisioncross-clusterdiagnostic / decision reliance28
AI Agents × Multi-Agent Collab.cross-clusterhuman oversight of agents21
AI Agents × Mixed-Initiativecross-clusterhandoff / initiative switching18
Explainable AI × AI-assisted Decisioncross-clusterexplanation + recommendation17
LLM User Study × Responsible AIcross-clustersycophancy / bias / misuse14
Trust & Reliance × Algorithm Aversioncross-clusteraccountability framing13
AI in Mental Health × Affective Computingcross-clusterempathy / mis-alignment12
AI in Education × LLM User Studycross-clusterLLM feedback in learning11
Human Factors × AI Agentscross-clustercognitive load / error recovery10
Phase II · Side-by-side · 2–5 directions

Direction Comparison

Comparing 4 directions · how research questions, methods, contexts, metrics and entry points differ

Select directions to compare (up to 5)
Trust & RelianceAI AgentsExplainable AILLM User StudyAI-assisted DecisionMulti-Agent Collab.AI in Mental HealthAI in EducationHuman FactorsAlgorithm AversionResponsible AIAffective Computing
Selected · 4 / 5 Reset
Trust & Reliance
Strong · Primary
AI Agents
Strong · High-growth
Explainable AI
Strong · with User Study
LLM User Study
Strong · Newer
30-day / 1-year
49
30-day
204
1-year
44
30-day
186
1-year
35
30-day
198
1-year
29
30-day
142
1-year
Growth
+ 29%
vs prev 30
+ 42%
vs prev 30
+ 13%
vs prev 30
+ 38%
vs prev 30
Ø HCAI Score
84
/ 100
81
/ 100
79
/ 100
78
/ 100
Top Research Questions

Over-reliance · trust calibration · explanation effects · appreciation · accountability framing

Human control · handoff moments · agentic workflow · error recovery

Understanding · length effect · confidence display · perceived transparency

Sycophancy · cross-cultural mental models · repair loop · long-term use

Methodology lean

User experiment 68% · survey + log 31% · longitudinal 11%

Prototype 52% · user experiment 43% · diary / field 22%

User experiment 74% · explanation UI eval 32% · heuristics 18%

User experiment 62% · interview 41% · field 27%

Common contexts

Healthcare · AI writing / coding · recommender

Programming · office tasks · multi-agent teams

Healthcare · finance · recommender · legal

Writing · programming · learning · everyday chat

Common metrics

trust · reliance · accuracy · cognitive load · decision time

control · task completion · efficiency · trust handoff

comprehension · satisfaction · trust · reliance

trust · usefulness · authorship · over-trust · perceived accuracy

Psychology entry

Judgment & decision · trust theory · dual-process · working memory

Situation awareness · mental models · human factors · automation theory

Cognitive psychology · attention · information display · individual differences

Conversational pragmatics · cross-cultural psych · metacognition

Representative paper
Calibrating Reliance: How Clinicians Use AI Explanations
Stanford · CMU · UW · HCAI 94
When the Agent Drives: Handoff Moments in Co-Programming
CMU · GA Tech · HCAI 91
Visualizing High-Dimensional Feature Explanations
UW · MIT · HCAI 82
Mental Models of Generative AI: A Cross-Cultural Study
UTokyo · KTH · IIIT-D · HCAI 81
Phase II · 2026 · Week 22 · May 25 – 31

This Week in HCAI Research

Compiled from HCAI papers added or updated between May 25 and May 31. 86 entries this week, spanning 14 of the 18 directions, with 5 new research questions surfacing.

New this week
86
vs 78 last week · + 10%
Collaborations
38
44% · 6-week high
Active directions
14 / 18
4 directions no new entries
Emerging topics
2
Agentic Workflow · LLM Pair Failure

Headlines

01
Calibrating Reliance: How Clinicians Use AI-Generated Explanations
Stanford · CMU · UW · arXiv · HCAI 94
The strongest entry this week — a mixed-methods study with 84 radiologists that lays out three concrete design principles for reliance calibration.
02
When the Agent Drives: Designing Handoff Moments in LLM Co-Programming
CMU · Georgia Tech · CHI 2026 LBW · HCAI 91
Three-week diary study with 24 engineers breaks the often-aspirational "seamless handoff" into four canonical, observable patterns.
03
Steering Multiple Minds: A Control Interface for Human Oversight of Agent Teams
MIT · Cornell · ETH · U-Tokyo · UIST 2026 · HCAI 82
Most ambitious of this week's 4-school papers — supervising agent teams is shaping up to be a defining 2026 HCI theme.

Notable Papers

  • Calibrating Reliance: How Clinicians Use AI ExplanationsStanford × CMU × UW
  • Steering Multiple Minds: Control Interface for Agent TeamsMIT × Cornell × ETH × U-Tokyo
  • When the Agent Drives: Handoff Moments in Co-ProgrammingCMU × Georgia Tech
  • Persuasive Defaults: AI Recommendations Reshape ReasoningU-Michigan × Berkeley × NUS

China · Research Update

Zhejiang University added 9 papers this week, concentrated in trust calibration and affective computing; two of them — collaborations with CMU and Imperial — sit in the high-score band. Tsinghua added 6, mostly in AI Agents and multi-agent work. SJTU, PKU and HKU each contributed 3–5 entries.

The strongest China-overseas pair this week is Imperial × SJTU (AI-assisted surgery), followed by UCL × Tsinghua (responsible AI / ethics).

Overseas · Research Update

CMU and Stanford together account for 23% of this week's new entries (12 + 11). UCL continues strong output in AI mental health with 6 new papers. U-Tokyo participated in this week's 4-school paper, making it the most-collaborated Asian node this period.

One small signal worth watching: both EPFL and KAIST entered the trust-calibration direction this week — worth seeing if it persists next issue.

Most Active Directions

  • Carnegie Mellon12
  • Stanford11
  • Zhejiang9
  • UW8
  • MIT7
  • UCL6
  • Tsinghua6

High-Frequency Directions

  • Trust & Reliance14
  • AI Agents12
  • Human-AI Interaction11
  • LLM User Study9
  • AI-assisted Decision7
  • AI in Mental Health5

Vs Last Week

86 new entries this week, up 10.3% from 78 last week. Direction coverage rose from 12 to 14 of the 18 primary directions, a 6-week high. The clearest direction movement is AI Agents, with 12 new entries this week — 1.6× the trailing 4-week average.
Editor: HCAI Research Radar · auto-compiled + human review
Phase II · Sub-direction Profile · Trust & Reliance

Trust & Reliance

Research on how users form, calibrate, and adjust trust in AI outputs — including over-reliance, under-reliance, and how explanation design, confidence display, and accountability framing shape reliance behavior.

49
30-day
204
1-year
22
Active directions
+ 11
vs prev 30
84
Avg HCAI
61%
Collab rate

12-month trajectory

Monthly paper count in this direction over 12 months
0204060JunJulAugSepOctNovDecJanFebMarAprMay
Most Active Directions in This Direction

Active Schools

Last 1Y · % of direction
01Stanford8.8%18
02Carnegie Mellon7.8%16
03UW6.9%14
04MIT5.4%11
05UCL4.9%10
06Zhejiang4.4%9
07ETH3.9%8
08Cornell3.4%7
09KAIST2.9%6
10U-Tokyo2.5%5
Adjacent Directions

Adjacent Directions

Explainable AI
32co-occurrence
AI-assisted Decision
28co-occurrence
Human-AI Interaction
24co-occurrence
Algorithm Aversion
18co-occurrence
Affective Computing
11co-occurrence
Responsible AI
9co-occurrence
Related Sub-Topics

Sub-Topics

Trust calibration18Reliance behavior14Algorithm aversion11Confidence display9Explanation length7Accountability framing6Under-reliance6Over-reliance5Disagreement surfacing5Longitudinal trust4
Representative Papers in This Direction

Representative Papers

Top 4 by HCAI score

Calibrating Reliance: How Clinicians Use AI-Generated Explanations

Stanford × CMU × UW · arXiv · HCAI 94

Algorithm Appreciation Reconsidered: Preferring AI Despite Equal Accuracy

Northwestern × U-Tokyo · CSCW 2026 · HCAI 87

Beyond Accuracy: Trust Calibration in Long-Term AI Companion Apps

KAIST × EPFL × TUM · CSCW 2026 · HCAI 75

From Disagreement to Decision: A Framework for Clinician–AI Conflict

UCL × Oxford · FAccT 2026 · HCAI 73