The ‘A’ of AI
I came across this article on Google blogs: https://blog.google/inside-google/googlers/hip-hop-artificial-intelligence/
The mentioned person used his existing music tracks to train a model which creates something new in his own genre. I had been dwelling on a similar cue from a long while. Well this just gave me a nudge.
Premise
We all have, at some point, sort of looked at AI as more of a replacement (read threat) to the human cult rather than something that may assist us (as well). I will try being assertive but at the same time wary of not hurting anyone’s sentiments here.
Here are 2 questions:
- Is creativity something one is born with or can it be acquired with time?
If you answered that it’s something one is born with, it is well quoted now by numerous studies that Creativity is “technically” inherited, but by everyone. It’s more of a common human trait than a gift. Hence, more than being a seemingly innate feature, it can actually be honed.
Now if you answered that it can actually be acquired, aligned with the experimental results as well, answer this:
2. Does it matter who is acquiring it: a human or a machine?
Here’s a fun experiment: http://boilthefrog.playlistmachinery.com/
Does physicality always matter?
In the above link, as well as in many other real life scenarios, data is being used to ‘discover’ things. We all have landed on a similar artist’s music we love on YouTube by following just the related videos from the one we already like. Now since this is data driven and automated, we call it an algorithm. But in some very niche cases, physicality might actually take a back seat.
Why?
If something (eg a model trained on music embeddings) has the capability to produce a unit of work (eg the similar genre of hip hop beats produced by the Google engineer), which could have been a human output in a parallel universe anyway- is an indicator that this ‘BLACK- BOX’ nature of the very ML might actually ‘impart’ creativity, which is deemed as a human only trait.
Nuance: AI and the Art- blurring boundaries?
- Artists can use this to acquire a perspective they didn’t have otherwise to create something new. Here, algorithms are not meant to follow a set of rules, but to “learn” a specific aesthetic by analyzing thousands of data points.
- Augmenting human creativity might not be a ‘wrong’ ambition. When assisting artists, a computer is already an instrument, a brush and a painting canvas. Computational creativity is something we have achieved by shaking hands with AI.
- The Turing test might not be the best testimony at the end of the day. A joke will still be funny, whether or not it was produced by a human or a machine. But in visual arts, this might not be the case.
- If we understand creativity like the result of establishing the relations between the pieces of information we already have, this ML aspect of us might reduce the bite which AI seems to give to some.
The Convergence
Some (artificial) creativity (soothe your soul if you may) can hence be a by product of an ML simulation (there are many music albums which have used entirely machine- made background beats).
I whole heartedly agree that anything and everything AI might or might not always be of some use to us. But the units which are useful, are just something one did not have and has now. Yes it is artificial, and might not totally ‘feel’ yours. But ‘discovering’ an ‘invention’, is creativity as well.
Now how artificial you think AI really is?
PS: Thanks to the friends who brainstormed this with me, you know who you are :)
