From the comments, in particular those made by members of the examination committees here (and elsewhere) and from what is known publicly about the way the EQE papers are drafted I conclude that there is a very high standard when it comes to the drafting of the EQE papers. They must be impeccable in all three official languages.
(Apologies to all involved in creating the papers for using the derogative term "perfectionism" in the title to attract attention -- the criticism therein relates to the process/system that has evolved over time, and expressly not to the superb job you are all doing!)
As a result, the process of creating a single paper is tedious and expensive. And the result is a paper that is often criticized for being a jig-saw like puzzle that can be solved by mastering a methodology rather than patent attorney skills.
Since we are discussing nothing less than the future of the EQE, why not have a debate on that point.
Let me start the debate with saying: the papers do not need to be perfect. If there is a flaw in the paper, give the markers the freedom to assess how the candidates dealt with the flaw -- did the candidates even notice and if yes, is their response in line with "fit to practice"?
In other words, I think there needs to be a "culture of error" that provides room for imperfect papers which will then be assessed with sense of proportion, always keeping in mind: is the candidate's answer/proposal a reasonable choice in view of the EPC and the client's wishes, or does it fail to comply with either and if yes, is that failure of compliance corrigible or final?
If we were to accept errors in the exam papers there will most likely be more appeals, in particular in the first years after the change. But once candidates realize that the errors in the papers are part of the process and that they are not losing marks for not finding the "envisaged" answer because of some ambiguity in the problem (and that instead they lost the marks for providing an answer which is unacceptable under any sensible interpretation of the ambiguity) I expect the appeals to abate.
And yes, this requires that the EQE overseers and the appeals body also accept that future exam papers are less perfect.
But in my eyes this is a better approach to reducing cost than reducing the number of papers and/or relying on an AI that in my eyes will not exist for many years, possibly even decades, to come.