I appreciate the historical comparison but think the issues here are different. With looking up of information in the early days of the internet, we eventually decided as a society that we did not care much how the information was retrieved. The goal of the task was simply the retrieval of knowledge and it did not matter whether the knowledge was retrieved using analog or digital means.

The situation is different here because the goal of an exam (in many but not all cases) is to assess whether the student has learned the material. That is, whether they have internalized the information and understand it for themselves. If a student generates the words with AI, that undermines the goal to assess the student’s own knowledge. Hence measures like a blue book exam format can help preserve the integrity of the assessment.

Now it is certainly possible that the goal of assessments will change over time. We may decide as a society that we care less about assessing whether a student understands knowledge for themselves so much as whether they can “get the task done” using AI tools.

My view is that this potential shift from thinking through knowledge to tool-based task performance would be a great loss in many educational contexts. For one thing, there is a strong 0.92 correlation between knowledge and creativity. If we drop the idea of learning things for ourselves, then we risk losing the material that we use to think with. To think flexibly for ourselves.