Human-in-the-Loop Grantmaking: Why It Matters

Human‑in‑the‑loop means AI assists with speed and consistency while people remain responsible for decisions.

• Protects against bias and mistakes.
• Ensures transparency and appeal routes.
• Aligns with UK data‑protection principles.

Practical model for teams

Define where AI helps and where humans must decide.

• AI: triage, summarisation, drafting, anomaly detection.
• People: approvals, exceptions, ethics and proportionality.
• Logs: prompts, outputs and edits recorded for audit.

Key takeaway: clear roles make automation safe and effective.

Accountability in practice

Keep decision records short, factual and explainable.

• Record what evidence was considered and why.
• Offer applicants clear reasons and next steps.
• Review samples for quality and fairness.

Key takeaway: Plinth supports strong, explainable decisions.

Building trust with applicants

Explain how tools are used and how people can challenge results.

• Plain‑English privacy and AI use statements.
• Contact routes for clarifications and appeals.
• Accessible guidance and examples.

Key takeaway: openness reduces concern and improves engagement.

FAQs

Is AI decision‑making banned?

Automated solely‑algorithmic decisions are restricted; maintain human review.

Do reviewers need training?

Yes—a short session on prompts, checks and escalation routes.

Can we disable AI per programme?

Yes. Plinth lets you toggle features to match policy and risk.