GPT-3 is the beginning of the end for academic essays

One of the most puzzling features of the assessment strategy in the UK's Higher Education sector is its over reliance on essays. Students are given a topic or questions and then have a few weeks to write a few thousand words about said topic. On my law degree, this type of assessment was the exception, not the rule but that was in Portugal and 20 years ago.

All assessment strategies have drawbacks, but essays for me have always been exceptionally easy to game and prone to academic unfair practice. Essay mills have mushroomed in the last decade for a reason after all. But fighting essay mills is  fighting the last war and even then, the tactics were all wrong. The coming war on essay reliability will not be around which human wrote a given essay but if a human wrote it at all.

In that sense, the GPT-3 AI is a game changer. It is shocking how logical and coherent it sounds and it is only a question of (little) time until it is deployed by some entrepreneurial souls for their own benefit (if they're students) or their clients (essay mills for example).

From now onwards, it is no longer reasonable to assume any given essay is written by a human until proved otherwise.

Of course, there is plenty that universities can do to avoid the onslaught but that requires quick feet and a willingness to change an ingrained approach. Inertia will ensure the change will not happen until there is blood in the streets under the guise of articles in the newspapers questioning the reliability of essays. That will take time.

Until then all will be fine because the procedures are "rigorous" and "best in class", plus students already sign a statement swearing it is their piece of work that is being submitted. And we know how those have been working so far.