Steve Law

Environment

Undeniably, AI is bad for the environment — there’s no definitive measure of how bad. Two reflections. First, if we’re going to use it, how do we offset the negative with positive? In the same way that the fuel burnt in a lifeboat is a net good if we save a life, how can we get enough good out of AI to make it a net good? Second, learning how to use AI in ways that reduce impact — right tools, right scale, right contexts.

Data

This is a personal project with no guarantees of data privacy or security. Consult your organisation’s policies, and never enter anything you wouldn’t be comfortable making public. Treat every input as potentially visible.

Part of this work is exploring the changing regulatory and technology landscape — and potentially helping the sector understand and lobby for what it needs from policy.

Bias

Bias exists everywhere in the sector, and AI is no exception. There are specific known issues I’m actively working to understand and mitigate — through prompt design, tool choice, and honest evaluation of outputs. New approaches are emerging constantly. I’m always open to discussion on this.

Slop

AI is generating enormous volumes of noise — generic, hollow content that fills space without adding value. I want PurposeKit to actually be useful: tools that produce outputs worth reading, grounded in real charity context, not outputs that look right but say nothing.