A behavioural intelligence framework for understanding human action in AI systems — proposing ABSD: Action, Behaviour, State, Drive.
When a student pauses for thirty seconds before a math problem, the system records the pause and moves on. It cannot tell confusion from careful thinking, anxiety from strategic deliberation.
When a customer abandons a shopping cart, the platform logs the event but has no idea whether the person got interrupted, felt sticker shock, or simply could not decide.
AI systems track what users do. They miss why users do it.
AI is the primary interface between organisations and users. Chatbots handle 85% of customer interactions at leading companies. Recommendation engines shape what 2 billion people see daily. Adaptive learning platforms guide millions of students.
Yet across these systems, three structural reasons keep psychology out of the picture:
Clicks and timestamps capture automatically. Emotions and motivation do not show up in server logs — they must be inferred from patterns, which most AI development skips.
Psychologists study motivation but lack computational tools. Computer scientists build pattern recognition but lack theoretical grounding. Cross-citation between the fields is minimal.
When engagement metrics become KPIs, systems optimize for time-on-platform rather than user goal achievement. Personalisation stays on the surface.
The result: AI that reacts instead of anticipates. Systems that optimize for clicks instead of outcomes. Tools that personalize surfaces while missing depths.
The framework synthesises four rigorous but siloed research streams. Each explains one piece of the puzzle. None has been integrated into a model practitioners can actually use.
Gomez et al. (2024) found minimal cross-citation between these fields. Each tradition rediscovers similar phenomena using different vocabulary. ABSD is the integration.
Each layer is generated by the one beneath it from the user's perspective — and must be inferred from the one beneath it from the AI's perspective. Tap a layer to expand.
Why the user does it. The fundamental psychological needs that shape how users respond when states are challenged — autonomy (volitional control), competence (feeling effective), relatedness (connection).
A user with high competence drive may persist through confusion; one with frustrated competence needs may disengage.
What the user feels. Not directly observable — inferred from behavioural patterns. Confusion can resolve into engagement or spiral into frustration depending on how the system responds inside that window.
How the user does it. Patterns emerging across actions over time — not single events. The bridge between what AI can see and what it needs to understand.
↓ The behavioural signal taxonomy:
What the user does. Discrete, surface-level events with high fidelity capture today. This is what almost every AI system already has — and where most stop.
Current AI sits at Layer 1. The framework is a roadmap for reaching Layers 2, 3, and 4 — using research that already exists.
A 30-second pause before a complex financial trade is prudent deliberation. The same pause before a simple checkout is probably confusion. ABSD treats signal-to-state mapping as probabilistic and context-weighted, not deterministic. Domain, user history, and temporal context all modulate inference.
This is a conceptual framework study — not an empirical model build. Validation focuses on three dimensions: conceptual coherence, theoretical grounding, and practitioner utility.
Integrative review across SDT, affective computing, JITAI, and learning analytics. Substantially complete.
Semi-structured, 15–20 min. Primarily CX, secondarily EdTech. Thematic analysis (Braun & Clarke).
Descriptive breadth on signal recognition, state inference, and framework utility. Indicative, not generalisable.
One CX, one EdTech. Illustrative applications of the framework using public documentation.
Three artefacts: (1) a validated conceptual framework with defined constructs and signal mappings; (2) design heuristics for state-aware AI systems; (3) a practitioner-oriented implementation guide. Customer experience is the empirical anchor; finance, healthcare, and others are theoretical applicability domains for future work.
15 semi-structured questions for expert practitioners. Mapped to RQ1, RQ2, RQ3.
15 questions across 4 sections — benchmarking, recognition, mapping, utility.
Click-through version with progressive save. Responses persisted locally, ready for API hookup.
Polished long-form version with all sections, ethics, contributions, and references.
This work needs practitioner input. If you build AI products — in customer experience, education, or anywhere AI meets users — your perspective sharpens the framework.
predictwhy.com · Arunamirtharaj Chandrasekar · SP Jain EMBA 26