Skip to content

Experiments

AI Lab

Studies in how generative systems behave — with the verification, refusals and corrections kept in view.

Projects

In the lab

Research archive / process evidence

Method

How the work moves

One cycle runs through every AI-related project here. Verification and rejection are steps, not afterthoughts.

  1. 01

    Question

    What does this work have to answer?

  2. 02

    Frame

    Scope, sources, format, and what a wrong answer looks like.

  3. 03

    AI assistance

    Research, drafting and structuring inside the brief.

  4. 04

    Verify

    Check every claim against a traceable source.

  5. 05

    Reject / refine

    Remove what cannot be supported; rewrite what drifted.

  6. 06

    Human decision

    The judgement stays with the person.

  7. 07

    Output

    Produce the evidence, not just the summary.

  8. 08

    Reflect

    Record what failed and adjust the constraints.

Question → Frame → AI assistance → Verify → Reject / refine → Human decision → Output → Reflect

Strands

Areas of enquiry

  • Creative AI

    Generative systems as material.

  • Human–AI Workflows

    Where assistance helps, and where it degrades the work.

  • Research & Evaluation

    Criteria design, comparison, verification.

  • AI-Assisted Building

    From requirement to working interface.

  • Responsible AI

    Provenance, authorship, limitation, refusal.

  • Experiments

    Short studies that may not resolve.