AI FIELD NOTES

AI Tested

Explore how to evaluate AI tools and workflows hands-on, report observed results, and state limitations and capabilities that were not tested.

Approved experiment concepts

Experiment concept

Can AI Really Plan a Better Working Day?

Compare ChatGPT and Claude using the same realistic workload, including prioritisation, scheduling and interruptions, with recommendations assessed against actual usefulness.

Experiment concept

Can You Build a Useful Website Using AI Without Writing Code?

Examine the real experience of building a website with Replit, including successes, failures, human direction and work required.

Experiment concept

Can AI Research a Major Decision Better Than Traditional Search?

Compare AI-assisted research with conventional search for a real decision, evaluating accuracy, sources, omissions and recommendations.