How trust leaders read AI claims
Selective evidence, AI washing, and the human-in-the-loop entry requirement.
The evidence problem
Leaders do not lack vendor evidence. They lack evidence they can trust. Vendors point to studies that show exactly what they want shown. Peer recommendations carry weight, but those peers may have a beneficial relationship with the supplier.
One leader described the edtech evidence landscape as a minefield. Independent standards exist, but they are not comprehensive enough to rely on for every purchase. Government and NGO evaluations help, yet they are slow and sometimes carry their own conflicts of interest.
AI washing is visible from the classroom
- Products advertise AI functionality that is little more than a label. One leader pointed to AI features bolted onto management information systems as superficial.
- Schools are rightly reluctant to feed pupil data into external AI tools. Past experience has made them cautious about claims of data security.
- The result is a trust gap: leaders can see the potential, but they have seen too many over-promises and under-deliveries to move quickly.
What earns a hearing
Three features were received positively in every call. A confidence score on any output. The ability to edit that output. And an explicit flag where the data behind it is thin or missing.
Without these, one leader said, there is no way to hold the tool to account. Human-in-the-loop is not a hedge. It is the entry requirement.
The deeper worry
Several leaders raised a concern beyond procurement: the gap between what state schools are measured on and what students actually need. The current system is results-driven and fact-heavy. Project-based learning, leadership skills and self-presentation are treated as harder to assess, so they are deprioritised.
One leader framed this as an inequality issue. Independent and international schools build these skills as a core offer. State schools, responsible for the vast majority of students, often cannot.
What follows
The opportunity is not to sell more AI to schools. It is to give leaders a way to evaluate AI, and any other tool, against their own framework, with their own data, and with the limitations visible.
Method and anonymisation
Coded from the same anonymised trust-leader discovery calls. Themes that appeared across multiple conversations are reported; one-off observations are excluded. No trust, school or individual is identified.
educave.ai · Free to copy, forward and use inside your institution, with attribution.