Affectosphere Group/CAI
For the Affectosphere era, building the world's most advanced Computational Affective Intelligence
Emotion has become the medium society runs on. We measure that medium, keep its ambiguity intact, and turn AI toward raising human emotional intelligence.
Vision
About Affectosphere
Emotion envelops society like an atmosphere, quietly shaping our judgments, our relationships, and the very texture of our inner life. We read this ubiquitous affective sphere with AI, attune AI itself toward the shape that best serves people, and pursue research that connects this work to a society where, even in the age of AI, everyone can keep a margin in the heart.
Definition
What CAI is
Computational Affective Intelligence is the study of how affect is represented, inferred, generated, learned and utilised by artificial intelligence systems.
Its subject is not only human emotion. How does an AI represent affect, interpret it, reason over it, produce it? How does it hold the uncertainty and the value judgements that come with it? Once a language model reasons about emotion on its own, and answers differently depending on persona and culture, recognition alone no longer covers the question. Emotional intelligence itself has to become the object of computation.
Research themes
Research domains of the lab.
Nine domains that make up CAI, and AI for Science as the tenth.
Ethics and Philosophy of CAI
We keep asking who emotion-measuring technologies are for, and how they should be used. We reexamine, in the language of philosophy, the norms that lie between technology and the human.
Learn more →Understanding Human Emotion
How do humans actually experience and express emotion in the first place? We re-read the accumulated insights of psychology, cognitive science, and neuroscience as the foundation for AI research.
Learn more →Augmenting Emotional Data
We explore methods that augment and complete ambiguous, polysemous emotion labels while preserving their uncertainty — enabling faithful learning even from sparse annotation.
Learn more →Understanding the Inside of CAI
We unravel what models actually base their emotion judgments on and make the underlying behavior visible.
Learn more →Emotion Recognition by AI
Foundational recognition research that estimates emotion distributions from text, speech, and physiological signals — aiming for designs that handle uncertainty rather than point estimates.
Learn more →CAI and Human Interaction
How should we design the place where people and models face each other? Self-reflection, dialogue, and the problem of co-presence.
Learn more →CAI and Business
We re-frame the meeting point of uncertainty, acceptance, and ethics in industrial deployment as practitioner knowledge.
Learn more →CAI and Art
Art as the place where emotion and expression meet — a domain in which AI moves among three positions: 'making,' 'reading,' and 'inspiring.'
Learn more →Development Based on CAI
Development research that implements emotion-reading AI as applied systems and delivers them to society.
Learn more →AI for Science
Applying AI to scientific fields beyond emotion. This is where we bundle cross-disciplinary collaborations in biology, educational safety, and related domains.
Learn more →Looking for collaborators
We are recruiting collaborators in CAI
We welcome students, collaborators, and industry partners interested in building and applying CAI, across a broad range of backgrounds.
Recent publications
- Domestic Journal Representation
Crossing the “LLMs Give Labels but Not Uncertainty” Wall: Implementation and Operational Practices for Generative AI Deployment in Subjective Judgment Tasks
Keito Inoshita
IPSJ Transactions on Digital Practices · Apr 2027
- International Conference Beyond affect
Re-Defining Vanishing Municipalities in Japan: A Multidimensional Clustering and SLM-Based Policy Insight Framework
Toma Okugawa, Keito Inoshita
IEEE GCCE 2026 · Oct 2026
- International Conference Interaction
Behavioral Fidelity and Philosophy-Grounded Design for Self-Reflective Conversational AI
Takumi Matsuo, Keito Inoshita
IEEE GCCE 2026 · Oct 2026
- International Conference Human Affect
Keito Inoshita, Takato Ueno
IEEE ICDM 2026 · Aug 2026 · Rank A*
- International Journal Society
Keito Inoshita
AI & Society · Aug 2026 · IF 6.1 / SJR Q1
Principal Investigator
Keito Inoshita
Computational Affective Intelligence researcher
He works on giving AI emotional intelligence. Measuring emotion as a probability and returning it with its ambiguity intact — perception, representation, reasoning, generation, interaction and ethics treated as one continuous system. That system is what he calls Computational Affective Intelligence (CAI), and building it is the through-line of the work.
Affiliated with the Graduate School of Business Administration at Kansai University, the Data Science and AI Innovation Research Center at Shiga University, and the Japan Safety Society Research Center, he also pursues deployment through collaborations with a wide range of companies. The horizon is an AI endowed with an EQ that surpasses humans.
Latest News
- Paper
- Award
- Award
- Paper
- Paper