AI can't trespass on all human domains

I asked it all kinds of things, from serious questions to jokes, to see for myself its learning capabilities. When it comes to answering serious questions, ChatGPT does really well. But when it comes to jokes, it's often very off the mark.

When it comes to the more technical inquiries, given the right questions, the AI can give out answers as good as any expert.

My children were also fascinated by the chat bot. My third grader asked it to write a story about our family members, while my eighth grader wanted it to write an article on the importance of confidence. ChatGPT passed both tests with flying colors. In fact, its writing in English far surpasses anything my eight-grader child could ever conjure at his age.

What if there are more users? What if ChatGPT could improve its learning algorithms even more over time? Will it become an all-knowing machine that cannot crack a joke? Such a thought sends a shudder through me.

I wondered if I should allow my kids to use it for their homework.

The birth of ChatGPT, and many other AIs, is something undeniable. To ban their use or restrict access to them is something impossible. As such, I will only allow my children to use it when they feel that it's necessary.

One day, ChatGPT and its brethren will become just as popular and essential as Google is now. So, I ask myself, how should our education system adapt to this inevitability?

Rote-memorization would be rendered virtually useless. Forcing students to learn "templates" for the sake of achieving high scores would be of no practical help, seeing as an all-knowing machine already knows all the answers.

So instead of focusing on memorization, what we should teach our children is how to ask the right questions.

It is a novel challenge. Such a shift would mean that our 2018 education pedagogy, which is being adjusted to be put into application, might be outdated.

But of course, there is no need to scrap everything and build the pedagogy from scratch. What we need to do is to make adjustments and adapt. The question is, how exactly should we adjust it?

The first, and also the best and easiest adjustment, is a change to our ways of testing and evaluating children. Education should not be about scores. Our current way of education depends too much on tests: however tests are carried out, so should our children study to pass them. A change to testing and evaluation methods should create a change to how our students learn and are taught as well.

Second is allowing schools to shed certain educational materials that are too cumbersome, as well as allowing them to deploy their own education programs that satisfy learners’ demands and are in line with the advancement of technology.

Third is an encouragement of education programs that are liberal in nature, which should help students unlock their full potential. These programs would focus on emotional regulation and the cultivation of an inner world, helping children form personalities and individualities, among other values.

If humans’ natural intelligence will one day pale to the machines’, then it would be our capacity for emotions and the awareness of the self that demark a territory that no machine could ever hope to trespass.

Besides intelligence, humans also possess awareness regarding ethics and aesthetics. An AI, at its best, could only advance itself when it comes to learning about objective knowledge. But the appreciation of beauty would remain a human domain.

In the end, it is important to remember that AIs are created to assist us, not to gain power over us. Humans can only be humans if we retain our awareness of ourselves, and if we don’t, we just might become mere extensions of our machines.

*Giap Van Duong is a Vietnamese educator and former researcher at the Temasek Laboratories of the National University of Singapore.