Designing together with an algorithm, creating a subjective personal algorithm

We have always thought that it would take us humanity for ever before artificial intelligence could help us in our daily lives. Suddenly it was the opposite, everything we deemed impossible, was possible.

I believe in embracing new technology and always being open and critical to these new developments. I want to research how A.I. can aid me, as a human, designer and artist. What it’s limitations are now and how it could aid and limit me in the future.

Currently the level of recognition in algorithms is quite strong. But it is all literal recognition. A human sees a wide open field with fragile clouds. And the program reports back: road, sky. While a human could also see in this image: freedom and joy. You currently also see a lot of cold chat bots with their own A.I. on the Facebook messenger platform for an example. Where everything in the conversation is binary, you either want it or you don’t. This is quite strange because we humans are not binary we are emotional subjective creatures. We should try to capture this emotional aspect in our own personal algorithms.

Image recognition test by standard algorithm, image by Stephen Shore
Personal human description of subjective emotions, image by Stephen Shore

Our first research question is, can we teach an algorithm a specific persons taste and make it subjective in recognition.

How would this work? We make use of the visual recognition technology of Clarifai .ai (thanks Ross)

We diversify the content we feed into the A.I. in to several subjects: design, photography and art. Each of the subjects are fed with 1000’s of images. If we take a look at the subject art. We scrape several art sites and gather their artworks. We feed this into the system. Now we add the human element. I as a person need to teach the A.I. what type of these images I like. If I don’t like the image I don’t have to do anything. I just dismiss it, If I do like the image, I need to describe to the A.I. what I as a human see and experience so it can learn from that.

Personal human description of subjective emotions, painting by Edward Hopper
Personal human description of subjective emotions, painting by Gerhard Richter
Personal human description of subjective emotions, photo by Robert Overweg

In our daily lives this could aid us in several goals. When you have taught your own A.I. to recognise what you like.

Your A.I. will help you design together with an algorithm. The tool can scrape app designs, It can find inspiration for User Interface problems. If you feed the algorithm with several distinctive flows apps normally have. It can learn what type of designs and solutions you like. Possible giving you recommendations during your design proces based on your own preferences.

Your A.I. will help you in finding images you find fitting for your art pieces. The algorithm, knows what kind of art you like. Or when you go on a vacation, it can scrape all the images from local museums and it could recommend you what museum to visit. Because it knows, you just might like that off the track obscure museum. Because you have described some art pieces and the computer recognises a pattern in those.

Your A.I. knows what kind of design and surroundings you like. Now this is a bit more hypothetical, but if it could work together with booking.com or hotel.com it should never recommend me hotels with design or interior I would prefer to miss out on. This saves me in time.

The A.I. could even help you in shopping online, you get the idea.

One final advantage, the algorithm should run local on your own computer, so no one but you knows your specific preferences. So these can never be used against you (in advertising for an example). This is similar to Apple’s approach.

To start this research track and because art has the most freedom, art will be the first subject to start with. In a paralel track the design subject will be started.

Come back soon for the initial results of our experiment and to see the next steps.

Best,
 
Robert / Berend / Mini_You

Robert Overweg guides some of the most leading international companies in making the digital transformation, using design and technology as a catalyst for this change. Robert gives lectures, creates art, offers companies guidance in concept development, strategy and the execution of this.

Berend Jutte is a researcher, studied applied neuroscience, holds a MSc and wants to integrate neuroscience knowledge into technology to create optimal experiences and for the overall betterment of humankind.

Mini_You will be your own personal algorithm, who knows you because you taught him.