Something worth noticing in the current panic: the International Baccalaureate has required teachers to confirm the authenticity of internal assessment work for a very long time, and the mechanism it specifies is not software. It is the supervisor knowing the student's work well enough to vouch for it, through scheduled check-ins during the process.
The Extended Essay makes this explicit with the reflection sessions and the viva voce at the end. A supervisor is expected to talk to the student about their own research and form a judgment.
That design is not a response to generative AI. It predates it by decades. It happens to be the right shape.
Why the shape holds up
The IB model assumes the artefact alone is insufficient evidence, and that the relationship between a supervisor and the work over time is what authenticates it. Both assumptions survived generative models untouched, which cannot be said for most assessment design of the same era.
The parts of assessment that assumed the document proves the learning are the parts in trouble. The parts that assumed a person has to vouch for it are fine.
Where it strains
It strains on load. The model works because a supervisor carries a small number of students through a long process. Scale that to every subject, every internal assessment, in a school where teachers already supervise a full timetable, and the check-ins compress into a signature.
It also strains on consistency. Two supervisors in the same school can hold noticeably different bars, and the framework relies on professional judgment rather than a common script. That is a strength when supervisors have time and a weakness when they do not.
What schools are doing about it
The IB schools we work with are not replacing supervisor judgment. They are trying to make the routine part cheaper so the judgment part still happens.
- A short structured check on each draft milestone, grounded in what the student submitted that week, so the supervisor arrives at the reflection session already knowing where the student is shaky.
- A record of those checks, which matters when a school has to demonstrate its authentication process to the IB rather than assert it.
- Consistent question coverage across supervisors, so the bar is closer to the same across a cohort.
None of that replaces the viva. It makes the viva better informed, which is the direction any change here should run.
The uncomfortable observation for everyone else: the framework that is coping best right now is the one that never fully believed the document in the first place.