Question the AI Answers
22 September 2026
It’s taken some time, but AI safety and trustworthiness are now capturing the world’s attention.
Six months ago, Co-Founder & CTO, Simon Gregory, published his report on the Six Pillars of Trustworthy Financial AI. The current headlines are now about the unsolved issues of AI alignment and unpredictability.
His first pillar, ‘Auditability’ explains that while you can ‘ask’ an AI to stay within constraints, there’s nothing actually forcing it to. Also, what an AI explains about what it actually did isn’t an audit trail, it’s simply a retrospectively generated plausible narrative of what it ‘could’ have done.
The sixth pillar, ‘Determinism’ explains why AI is fundamentally unpredictable by nature, and what that means in practice.
These are the unsolved issues at the heart of ‘AI alignment’, but they don’t have to prevent you from safely using AI for specific scenarios.
His report explained that AI can be trusted, as long as you are deliberate on its application, and that it is wrapped in a fully deterministic architecture (which isn’t at the mercy of model behaviour).
Read the Full Article: https://lnkd.in/e-rWGzd6
(7 min read)
