On reading memes, teaching in a classroom, and some of the things that might save us from a rogue AI someday
One of the true and unvarnished joys in life is to teach another person -- or a group of people -- and to have at least one of them acknowledge that you have managed to solve a problem for them. Classroom teachers get that experience when graduates visit them to say "Thanks". Workplace trainers and presenters get it when someone comes up after a session to say, "I was just dealing with that thing you described". Others get it from having their explanations liked or shared by others on social media.
■ Maybe we should be grateful more often that as human beings, we have an extraordinary capacity to help solve one another's problems. Collaboration itself towards a worthy goal is often quite intrinsically satisfying, and often the greatest satisfaction comes about when there is no remuneration involved. Normal, well-adjusted people often happily stop to give directions to strangers who look lost, or take time out to explain a smarter or safer way to complete a task someone obviously has underway, or write out an online product review without compensation.
■ It has been noted that some of the things that are easiest for computers are hardest for people, and vice-versa. People share memes without a second thought, yet a 2021 paper concluded that "comprehending memes is indeed a challenging task, and hence a major limitation of AI".
■ Perhaps these few ideas which computer science will probably never be able to solve are the very things that will act someday as the kinds of "kill switches" we could need someday to rein in an artificial intelligence tool that has gone rogue. Computers may be able to answer many questions, but it's not credible to believe they can be made to "feel" a sense of satisfaction from solving problems for people.
■ It is widely believed we are living through an unusually epic moment of technological change, but the evidence remains scant that we've truly covered all the bases when pondering the consequences when -- not "if", regrettably -- things go wrong. It may be satisfying to realize that there are some means of communicating and some motivations for working together that are likely to always elude even the "smartest" of artificial general intelligence. In the end, it may be those peculiarities of human existence that save us.