A group of university administrators has made the following foolish argument (1).
OBSERVATIONS
O1: The introduction of the calculator in the classroom was opposed by many who predicted a decline in academic standards.
O2: The introduction of the calculator in the classroom DID NOT result in a decline in academic standards.
O3: The introduction of A.I. in the classroom is opposed by many who predict a decline in academic standards.
ASSUMPTION
A1: The introduction of A.I. in the classroom IS equivalent to the introduction of the calculator in the classroom.
CONCLUSION
C1: The introduction of A.I. in the classroom WILL NOT result in a decline in academic standards.
Here are my objections.
ON O2
The introduction of the calculator DID result in a decline in academic standards.
Perform the following long division by hand: 123456 ÷ 789.
Did you find it easy? No? Why then should students be exposed to this avoidable tedium?
Because long division does train the mind NOT ONLY in long division. It also teaches students to break a big unsolvable problem into smaller solvable ones. And it exercises the mind in the consistent application of a simple set of rules, a procedure, an algorithm. And it teaches students perseverance and resilience.
Not exactly irrelevant transferable skills and traits in life and in the workplace.
ON A1
The introduction of A.I. in the classroom IS NOT equivalent to the introduction of the calculator.
Calculators are primitive tools. They are about arithmetic, the processing of numbers. Calculators are not smarter than humans, they are just faster.
Of course, mathematics also deals in algebra and calculus, the processing of variables and equations, respectively. We have invented tools to make us faster and more accurate in those areas as well (2). But using these tools requires knowledge about what you are doing.
Even using the calculator: What exactly is it that you want to calculate?
Conversely, much of today's "A.I." is based on Large Language Models, artificial neural networks that process complex symbols. These provide easy access to products of analysis and synthesis. Using them doesn't require the discipline of acquiring knowledge. An idiot can use ChatGPT and look smart (3).
ON C1
Prompted with the observations (O1, O2, O3), the assumption (A1), and the conclusion (C1), the ChatGTP artificial neural network responded (4):
"The key to this conclusion lies in the assumption that the introduction of A.I. is fundamentally similar to the introduction of calculators in terms of the context and scope of use. If this premise holds true, then the positive outcome from calculators can be reasonably extended to A.I., supporting the conclusion that the impact on academic standards will not be adverse."
Not bad. A machine displaying more intellectual rigour than the authors of the article. (And yes, I see the irony here: Had the authors used ChatGTP, they could have made a stronger argument.) Still, I am somewhat disappointed. I expected the machine to have more of an agenda.
NOTES AND REFERENCES
(1) M. M. Crow, N. K. Mayberry, T. Mitchell, and D. Anderson (2024), AI Can Transform the Classroom Just Like the Calculator. https://www.scientificamerican.com/article/ai-can-transform-the-classroom-just-like-the-calculator/ (Accessed: 24 Apr 2024).
(2) E.g. Mathematica, Maple.
(3) You want to give a presentation on say 17th century French literature? Just sign up to ChatGPT and enter the following prompt: "Can you provide me with the text for ten PowerPoint slides on 17th century French literature?" Hey presto!
(4) ChatGTP 3.5 (Prompted: 25 Apr 2024). ChatGTP gave a variety of responses to the same prompt. All consistent with each other.