Ben Caldecott, a sustainable-finance researcher, says investors should put their AI beliefs on paper before the technology reshapes the value and risks of their portfolios. A new tool from Silicon Boundaries turns that argument into 15 questions and seven broad ways of thinking about AI’s future.
Watch Desk analysis
What happened
Caldecott’s commentary in Responsible Investor introduces What do you believe about AI?, a tool from Silicon Boundaries, a joint initiative of the Oxford Sustainable Finance Group and Theia Finance Labs. It asks investors to take positions on questions ranging from whether frontier AI progress should be slowed to how AI should be used in military work and whether training on creators’ work requires consent.
The questions cover three broad areas: the frontier, industry power and AI’s uses. The initiative has grouped common patterns of answers into seven archetypes, including “safety first”, “growth first” and “capital discipline”. Caldecott says the tool takes about five minutes. He also argues that written beliefs could help asset owners choose investment managers, set mandates and identify where a portfolio depends on a particular version of AI’s future.
Why it matters
Investors already make choices that imply a view on AI, Caldecott argues. A portfolio heavy in data centres, for example, effectively bets that AI development will continue without being slowed. Writing down the assumptions makes them easier to examine, explain and revisit, rather than leaving them to hide in the plumbing of an investment strategy.
This is a proposed framework, not evidence that the seven archetypes capture every investor or that using the tool improves returns. Its practical value is more modest, and potentially more useful: it gives institutions a starting point for discussing which AI risks they accept and what evidence would change their minds.
Our read
The five-minute quiz is the easy part. The worthwhile idea is to connect an AI position to actual decisions about holdings, managers and mandates, then record what might make that position change. Otherwise, “we have an AI policy” can be the financial-world equivalent of putting a tasteful label on a very large shrug.
Caldecott makes the case; investors will have to decide whether the framework helps them ask sharper questions or just gives existing preferences tidier names.
What to watch
- Whether asset owners start asking investment managers to state their AI beliefs.
- Whether the seven archetypes prove useful in real portfolio and mandate decisions.
- What evidence investors say would cause them to revise their positions.
- Whether beneficiaries and boards use written AI beliefs to challenge investment choices.
Discussion spark: Should investors be expected to publish their AI beliefs and use them to set investment mandates, or would that turn uncertain forecasts into constraints on their portfolios?
Sources and evidence
- Source update (29 October 2026, 11:35 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.