Today I was asked by a friend about her kids' concerns whether it was worth spending 4 or more years in college studying computer programming and digital media when it looks like AI (artificial intelligence) might replace them.
I've been following AI development since 1980. Then, and now, this topic constantly comes up with various pundits sharing their opinions. So, I'll share mine.
Although what AI can do now seems impressive, it is still limited. AI will be significant in helping programmers code more efficiently and effectively. I would wager that will be the case in digital media as well.
AI is a tool. The best it can do is create a limited solution, whether a coding routine or a graphic design, for a constrained problem that is already known and previously understood with rules, context and relationships spelled out and “taught”. The solution is limited because it only looks at what is known and can't consider what is unknown. So, if the environment changes, the solution will “break”. AI can be most effective in “constrained” problems because AI can only consider a limited number of parameters. The bigger the problem, the less effective AI is.
Let’s look at music. Music that sounds good follows a known set of ground rules called theory with certain preferences. Each form of music- classical, rock, rap, folk, country, etc., has a set of rules on chord structure, instruments used, melodic lines, voicing, rhythms and other parameters that many listeners, composers and musicians consider “good” music. These are “taught” to an AI program so that it can “create” music in these genres. The result can sound pretty good. But AI is not “creative”. It would never think to a new form of music like electronic music or play classical music with wave form generators, fast Fourier transforms, envelope generators, frequency modulators and amplifiers to create music like early Wendy Carlos recordings. AI wouldn’t know to “break” a rule such as what Peter Schickele has done to make his compositions humorous, fun but still very good musically. Or, to use classical theory in bluegrass compositions like Mark O’Connor. Unless AI was taught and trained to do it.
In other words, AI can only do what it is taught and told to do. It cannot truly create something original.
So, AI has come a long way, but it has a long way to go. And as with any technology or change in culture, what we learn today may not be what we use tomorrow. So good programmers and digital media designers will always be learning new ideas, training on new technologies and keeping abreast of these changes.
But realize things can move very slowly sometimes. I graduated college knowing how to program in Pascal, Fortran and Assembly. Fortran is still used in scientific research, assembly in building drivers for time critical operations (e.g., pacemakers) and there are still jobs available for Pascal programmers. And this is after 40 years!
So, I tell my friend to tell her kids that college is only the start of their learning. It is never the end of it. Go for it!
Photo by Tara Winstead: https://www.pexels.com/photo/person-reaching-out-to-a-robot-8386434/
Published Jan 05, 2023