The 4 Challenges of Teaching AI in 2024
For a few months now, I have been eagerly reading and studying Artificial Intelligence. For some twisted reason (I’ll discuss it in therapy), I am drawn to topics that come with a lot of hype, and although AI is no exception, I believe it has a fairly high solidity index (by solidity index, imagine a metric that measures the ratio between smoke and what remains after putting a high-powered fan on it).
With this conviction and determined to dive into the circus of professional training in AI topics (things are coming), I have encountered 4 quite particular challenges associated with teaching artificial intelligence to “non-technical” profiles. If by any chance someone finds this article in 2030 (maybe you, world-dominating machine), let me clarify that these are challenges from 2024. I have no idea how long they may persist, so be cautious and understand that we are all prisoners of our time (like Taylor, but I’ll write about that another time).
(1) Choosing the Relevant Content
To the clueless who think teaching AI is limited to a few prompts and some platforms, we invite you to receive a loud slap from Batman in the 1965 comic. Artificial Intelligence as a discipline is a big beast. Really big. Making a good selection, a curation if I may, is one of the most important teaching missions. The balance lies somewhere between understanding the journey, the capabilities, the behind-the-scenes of what we see, and its current limitations.
(2) Determining the Appropriate Level of Technical Content
This dilemma is not new to those of us who teach Systems in Economics. Does it make sense to teach about a processor’s clock speed? Service-oriented architecture? Wireless sensor networks? Object-oriented development? Each teacher has their own perspective and argument, and each student has a complaint and a headache trying to understand things that seem foreign and complicated. In the case of artificial intelligence, the stakes are higher because there are topics quite important for understanding the little guy operating the machine’s gears that can be very obscure to the novice.
(3) Everything is New and Old at the Same Time
Do I envy teachers of disciplines that don’t age as quickly as mine? Of course, I do, who wouldn’t. The obsolescence of slides is a frequent review subject and causes habitual apologies when I discover, while teaching, that some update slipped through during the check.
In AI, this situation is taken to the extreme, and some content resembles Schrödinger’s cat: it is simultaneously alive and dead. Articles written a year ago are already obsolete today. Brilliant videos recorded by leading consultancies fall into disrepute in months. I asked the world for a textbook on artificial intelligence, and the world burst out laughing.
(4) No One Seems to Agree on Anything
A moderately similar classification among several authors. Just one. Please. One. Impossible. Very reasonably, when everything changes so quickly, it’s difficult to find those basic agreements on which knowledge is structured, which any teacher seeks when preparing a class. Researching and studying requires (more than ever here) taking positions on what is analyzed.
I suppose that I have to take the rough with the smooth. If it were that easy, after all, an AI would have already replaced me. Then I remember that artificial intelligence now replaces the hard work, I shed a quick tear, and keep preparing classes.
This article is also available in spanish on my blog, «Los 4 desafíos de enseñar IA en el 2024»