Education in the Age of Generative AI
7 Predictions
A few years ago, we were told that software was eating the world. Now, we are told that Generative AI is eating software; and hence, the world.
What will this world look like? How will Generative AI, for example, affect Education?
Prediction 1: All exams will be subjective
Almost all our exams are objective exams. By “objective” I mean, an assessment that consists of questions with predetermined and clearly defined correct answers.
For example, a quadratic equation in a Mathematics Exam has one or two deterministic solutions. It can’t have any other solutions.
Even exam questions that, in theory, seem subjective, are in practice objective.
For example, “What were the main influences on Buddhism during the reign of King Mahasen?” might seem open and subjective, but any student adequately describing the five “main points” mentioned in the history textbook will get full marks.
Generative AIs can already correctly answer quite a lot of objective questions. With better training data, larger models, and integrations with analytical engines they will get better and better. And before long, then will be able to answer almost all objective questions.
In a world where anyone would be able to correctly answer any objective questions with a few taps on their smart phones, what would be the point of asking such questions in exams?
Thus, in such a world, objective questions would disappear from exams.
Prediction 2: Exam Results will be more polarized
Even today, there are a few exams which include mostly subjective questions. These tend to be restricted to the arts: writing, music, and painting. Examiners in these areas tend to focus less on objective “correctness”, and more on creativity and novelty.
To be fair Computer Science professors have always been interested in the “beauty” of algorithms, and Mathematics Dons look for “elegance” in proofs. But these tended to be restricted to the best teachers and their best students. It was a sort of secret indulgence of geniuses.
As far as the mainstream was concerned, all one had to do was to get something “objectively” right. Beauty or art was neither here nor there.
However, once exams are completely subjective, all we are left with is this world of creative geniuses. Hence, as opposed to A+, A-, B+, B-, C+, C- etc., we move to a world where exams have only two grades G and nG. Either one is a genius, or one is not a genius. We will move from the Bell Curve to two modes: one tiny, the other enormous.
If this seems far-fetched, consider a phenomenon we are already seeing.
Programming tools like GitHub’s Copilot are making Excellent Programmers more efficient, but Mediocre Programmers obsolete. Technology Companies have begun laying off or plan to lay off large chunks of their programming staff — but at the same time plan to pay much more for those who will not make the cull.
Prediction 3: Examining Bodies will have to “own” their Results
The problem with subjective exam criteria is that it is impossible to prove that they are correct. Because if there was some objective means of proving this, they would no longer be subjective.
Hence, there will be more pressure on examination bodies to prove that candidates who do better at their exams, do better in employment, or higher studies.
As it is, the correlation between many examinations and performance in later life is at best weak, and sometimes even negative. This phenomenon will be more stark and aggravated.
Prediction 4: Examination Monopolies will Break down.
A few software companies hire candidates with no formal academic education. Instead, they look at interview performance, independent projects, and internships. This is because these companies have understood the lack of correlation between exams and future performance.
As of now, the above type of company is a small minority. But as Generative AI expands, they will gradually become a majority. Increasingly, people will be evaluated, not by centralized exams, but by the organization that wants to employ them.
Prediction 5: Many Exams will be Scrapped
Instead, education will focus on acquiring skills and giving candidates an innately intrinsic and subjective sense of what novelty and creativity means.
Many institutions that depended on exams (e.g., Universities for Admission) will revert to other means. Some of these means will be extremely novel — like random admissions.
Prediction 6: Fields that seemed immune to Generative AI will seem less immune
Those who know I am a Computer Scientist will think my views are biased. Particularly, since Generative AI is “our baby” and ours is one of the fields that will be most affected by Generative AI. Hence, they might conclude that my predictions would not hold for (most?) other fields.
However, consider the following.
Surgeons and Dancers have been often quoted as two professions who are fairly immune to AI. This is because these professions are innately physical, and not informational. Generative AIs have no hands or feet, and can only operate on information.
But what happens when there is a machine that can perform all the intricate manual operations a surgeon can perform? Thus, giving Generative AIs the same physical interface as a Surgeon?
Most of us probably witness dances, not in the flesh, but through TV, YouTube, and other electronic means. What happens when it is possible to generate perfect films, documentaries and other videos without any interference or interaction of humans?
Prediction 7: The future will prove that the above 6 predictions are neither completely accurate nor completely exhaustive.
Generative AI is in its earliest stages. Hence, making broad predictions about its implications is a bit like trying to predict what first class air travel would feel like, just as Orville and Wilbur take off the ground.
Take these predictions with the prescribed pinch of salt.
Update: Video of this Article
