Sunday, January 22, 2017

Humans vs. Computers

We hear a lot these days about how computers (and robots) are taking over all the mundane tasks we humans do on a regular basis; how we have to change the way we're teaching in order to prepare students for a very different-looking workforce.

Robots/computers can vacuum our floors (hello, little Roomba!), mix fancy drinks for us, and can drive our cars. On assembly lines, robots complete tasks more efficiently and more consistently than we ever could as a species. They can even create abstract works of art and write newspaper articles.

I've often wondered what the limit is: 
Where is the threshold for what humans can do pretty well, but computers still can't do at all?

Last month, I discovered one possible answer: composing music.

Just before Christmas, a sound clip of the first computer-composed Christmas carol was released. This computer was fed hundreds of hours of traditional Christmas carols which it analyzed, decomposed, pulled the most common elements from, and then used to synthesize something completely original. 

Here it is - take a listen:

It's AWFUL. I'm pretty sure some of the youngest students I teach, who have very little experience in music, could come up with something better. Especially those lyrics. Yikes.

A computer can try and combine the most popular elements of existing songs - basically pulling from a huge resource bank, larger than any human would have access to - but can it really push the boundaries of music? Can a computer be daring? Can a computer take creative risks? At this point, I would venture that they cannot.

Humans, however, can. My husband (@christheij) is a music teacher. Recently, while looking for new choir music for the spring season, he found this gem by Katerina Gimon. Take a listen - it's well worth it.

Perhaps my favourite part of this piece is the score, which contains, among other unique things, the following as notation (seriously, this is actually written into the music) (yes, that's forte fire):

Humans are able to take knowledge of music and not just synthesize from it, but expand on it, creating completely unique music that sounds GOOD, even from seemingly random noises. Computers? Not yet.

So what are we preparing our students for? If the mundane and routine jobs will become automated, but computers still can't be THAT creative, then it's probably a good thing to focus on those 6 C's of 21st century learning: Critical thinking, Collaboration, Creativity, Communication, Citizenship and Character.

But that's not to say we should give up rigour. We humans are still pushing: deepening our knowledge of how our brain works, and translating this into robots and artificial intelligence. The "hard" skills of the scientific method, understanding mathematical processes, and logic sequences that come with activities such as coding need to still be at the forefront. 

With these as skills, our students may one day be able to write a program that allows a computer to compose music that actually makes sense to our ears. 


  1. Good morning Heather!

    Thanks for this awesome share. There's nothing like listening to that composition at 5 a.m. on a Monday to get your week in gear!

    I think we are right to be very mindful of how digital technologies and AI are rapidly shifting our world, and how as educators, we need to be very conscious of matching our work to this exponential change. We know that the digital economy will be very different.

    While I agree that the 6 C's are important, alone they are not enough. Without digital literacies, our kids will not be able to thrive in the new economy. The 6 C's are a good basis from which to tackle digital literacies, but I worry that many seem to think that "doing the 6 C's" without technology is just fine.

    Canada's Digital Talent Strategy talks about the importance of creativity within the digital economy and how important it is to ensure our kids are learning what they need to take part in this rapidly shifting world.

    Thanks for continuing to bring beauty to it!


  2. I saw a science fair project last year where a student wrote a program to do exactly what you described here. It was really interesting! AI can optimize, but can't add feeling.

    Another cool example is AI cannot recognize shapes at different vantage points. Humans can recognize a chair regardless of the angle we see it from, but AI cannot. That's one big argument for why we want to send humans to the Moon or Mars!


Note: Only a member of this blog may post a comment.