Identifying and defining intelligence is a challenge since there are so many ways in which humans, animals, and artificial creations display intelligence. Although some academics have made the case that standardized tests like the SAT measure intelligence reasonably well, they notably avoid trying to define intelligence at all. Certainly, the SAT measures a specific sort of intelligence, but even the most enthusiastic promoters of standardized tests would be hard-pressed to argue that their tests reveal an all-encompassing intelligence. Indeed, five years ago, artificial intelligence (AI) could solve geometry problems on the SAT as well as an average test taker. No doubt AI today could outperform the average, but that would certainly not suggest that AI has achieved human intelligence.
In education, Howard Gardner’s theory of multiple intelligences is widely, and often incorrectly, used to assign students learning styles, but the central insight that human intelligence is multifaceted is a key one in a world dominated by the specific sorts of intelligence measured by the SAT. Gardner would likely say the SAT measures specific sorts of verbal-linguistic intelligence and logical-mathematical intelligence, but that these are only two of nine forms of intelligence.
Technologists have a wide range of views on the likelihood and desirability of true artificial intelligence, it is an open question as to what, for example, interpersonal or intrapersonal intelligence might look like in an artificial being. However, as Brian Christian discusses in The Alignment Problem: Machine Learning and Human Values, increasingly computer scientists are developing artificial neutral networks explicitly developed on the basis of insights from neuroscience about the human brain. Indeed, temporal difference learning is modeled on the dopamine system, which leads Christian to hypothesize that AI modeled on human brains is likely to lead to some sort of subjective experience (consciousness).
For the technologist and science fiction writer Ted Chiang, this possibility means we are likely to inflict vast amounts of suffering on AI far before AI is properly close to human intelligence. Although most of the AI currently developed is unlikely to be able to experience suffering, Christian describes an AI designed to play/explore Super Mario Bros on the basis of identifying novelty within the game structure that, when there is no more novelty to discover, does nothing or simply kills itself. As a feature of the program, that is effectively an intelligent response to there being nothing left to do
In this short of space, it is quite impossible to develop any deep conclusions about human and artificial intelligence, but in my view, these ethical and moral issues are critical ones to consider placing in front of students. There’s a quote that is often misattributed to Albert Einstein: “Everyone is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.” I’m not sure that everyone is a genius, but certainly, everyone is intelligent to varying degrees in a variety of capacities. The same might currently be said for artificial intelligence, which can now beat not only the best human chess players but also the best Go players. As an educator, I hope my students can reflect and identify areas of their intelligence to build on them, and also that they consider some of these unresolved questions about human and artifical intelligence which will certainly have some fundamental impacts on their lives.