05/12/2012 06:44 GMT | Updated 04/02/2013 05:12 GMT

'Random Shopper' Darius Kazemi Pays Robot $50 To Buy Him Unexpected Amazon Items

What happens when you let a robot randomly buy stuff from Amazon using your money?

You get into Hungarian avant-garde composer Ákos Rózmann, obviously.

Darius Kazemi has built a software bot which buys him $50 worth of paperback books, CDs and DVDs every month - just as an experiment.

Kazemi details his purchases via his website, Random Shopper, and from the sounds of things it's been a pretty illuminating experience so far.

Speaking to the Verge, as part of a wider piece about randomised shopping, he said that he likes "being forced to consume things outside of what these large corporations like Amazon think".

And writing on his website, he goes on to say that he had first thought to get the robot to buys things he'd probably like - but adds this more risky way is more fun.

Well, I thought: what if I just wrote a program to buy stuff for me? The first iteration of this was going to be a program that bought me stuff that I probably would like.

But then I decided that was too boring. How about I build something that buys me things completely at random? Something that just... fills my life with crap? How would these purchases make me feel? Would they actually be any less meaningful than the crap I buy myself on a regular basis anyway?

So far the experiment seems to be going well - he enjoyed the Hungarian CD ("It's fantastic, like nothing I've ever heard") and also received a book, Noam Chomsky's Cartesian Linguistics. which he said he might have wanted to read anyway.

Kazemi also disputes criticism that his project is a waste of money - arguing any form of art is potentially extravagant.

"The way I look at it is this: I'm spending $50 a month on art supplies. Some people might spend $50 a month on painting supplies: canvas, paint, brushes, etc. For me it's a bit more abstract than that, but that's what I'm doing."

We love this idea - but it's worth reading The Verge's piece for a wider understanding of how it fits into the concept of recommendation engines.