Magic tricks created using artificial intelligence
scienceblog.com
scienceblog.com
This is being reported in many publications as the first time artificial intelligence has been used to create a magic trick. This is false, not only has it been done by others before, it has been done by this team at Queen's Mary University before. Saying it's the first time certainly makes it more news worthy though.
Edit 1: If I were communicating this to someone else, I would say that they used analysis to create a new method, not a new trick. It's a new method, if it's really new, I haven't validated that in magic, and the vast majority of time any method you have isn't new. But perhaps it is, I will take that at face value for now.
Edit 2: "we suspected that audiences would be suspicious of the involvement of technology in the delivery of a trick but we’ve found out that isn’t the case."
Audiences are suspicious of the involvement of a phone during a trick. This is a minor contention, but their quote is flat out wrong. It would be interesting to see why they came to this conclusion.
Edit 3: Here is a video for the main trick discussed, "Phoney". Apologies for the misc intro, this was the best version I could find.
https://www.youtube.com/watch?v=h5qWi3yeu0M
Edit 4: Analysis of the method of the trick -
The captions say the cards are shuffled, but they aren't. They can only be cut. From there you deal 6 cards, and ask the spectator in order to tell if you if they are red and black. From there, you can determine the card.
Edit 5: And for my last trick, I found this method isn't even new, I can see it published by Persi Diaconis - http://en.wikipedia.org/wiki/Persi_Diaconis
Edit 6: As with any method, I might find a significantly earlier providence for the method, but I'm still looking for my own sake.
A version of the concept explained by Brady from Numberphile https://www.youtube.com/watch?v=EWG6e-yBL94
Edit 8: Here's some more detail for the math types. Universal cycles for Combinatorial Structures from 1990 by Chung, Diaconis and Graham
http://www.math.ucsd.edu/~fan/wp/universalcycle.pdf
And Products of Universal Cycles connecting it's application to card tricks by Diaconis and Graham
http://www.math.ucsd.edu/~ronspubs/08_05_products.pdf
Edit 9: So far I think I've traced this trick back to Charles Jordan in 1919, however given enough time I'm sure there would be better detail.
And while we're at it, is there any reason the videos emphasize the flip to show the face of the cards as they are being laid out? If they are going to do this, it seems like a much simpler and direct way to do the trick.
The video is actually the worst way I've seen to perform this method/trick. The flip is actually unnecessary, as is even asking for the color of each individual card.
I really would like to continue further, but I already feel bad about any exposure I've included.
Looks like "AI" is a cool term nowadays and it's hijacking the more basic notion of formalizing a problem and using a computer.