Machine Learning might be the most promising future of User Interface Design
Despite UI/UX thriving in these recent years because of the importance laid onto it by the top players of the industry and its obvious effects that separate a good product from a bad one, it still feels static to me. Most user interface only looks dynamic because it is bound to follow a restricted path that is driven by market trends, type of data and consumer behaviour. Take those factors out and you will have something that would never change while adding another single factor can change the entire face of the industry as we know it. The work is on and I see a promising future in it.
When you sign into your facebook account, what differs my newsfeed from yours is the content and the friends we have. Along with the usage patterns, every individual interaction and the interests, we have had shape up the content which makes our profile and the newsfeed unique and different. Give the service enough time and chat with enough people and it shall handpick what interests and engages you so that you interact more with it. It has a snowball effect the discussion on the ethics of which could be left for later. Machine Learning is already being used to shape our experience but the important question here is - can it also shape the interface with enough information about the user? Can your facebook interface differ from that which I see?
I am not talking about the sections you can move or hide but the features that you never use or the ones you use all the time.
When social networking was new, sites like MySpace, Orkut and Hi5 allowed their users to customise their profiles and pages to a large extent. This projection of individuality was directed from the user’s end. The ‘trend’ quickly died in the modern social networking ecosystem where the most you can do to make your profile special is to creatively utilise the space you are provided with and work within the framework. In the presence of this rigidity, there is a very fluid flow of information that is happening on our walls. This limits the variations to a certain area and the management of other things becomes easier with this uniformity which might have its economical benefits. Almost all present social networks have shifted to this approach where personalization and individuality are separated. If this were not so and an artificial intelligence driven user interface could be the popular method to design a user experience, the projection of a user’s individuality would become bidirectional and almost cyclic.
I am sure that we have the resources and the tech to build a system where the user’s behaviour and data shape his/her experience of a service. An existing example could be the Google Now service which gathers information on all fronts but asks you to select what it should show and what it should not through cards. Still, they have not gone fully with the concept the reasons of which I would really be interested in knowing.
Applying Machine Learning to UI is where it is at and the latter would not stand immune to the former for long. The effects of this will bleed out and completely change how we interact with the software running these services and in turn the objects that run this software. It would not only hide what is not important to the user but also bring forward the things he/she might like or find important at that point of time. Say, if facebook has my cookies and knows I was reading up on privacy for hours last night, the next time I log in, more than the ads or the posts from Internet Activism groups, it should simply bring forth the privacy settings/policies option forward and it be highlighted or salient in such a way that it gets my attention. The intangible personalization will also bring in a degree of semi-permeable security where only you would be able to use and understand your facebook timeline and no one else. Only you would know what shows up on your timeline and where.
The automation of customization has been present for long but it needs to be put into the real place where it ought to be. This is the small step it needs to take before its first stride into the real world where information, possibilities and hence the customizations explode.
Interesting reads:
https://en.wikipedia.org/wiki/Intelligent_user_interface
http://www.aaai.org/ojs/index.php/aimagazine/article/viewArticle/434
http://dl.acm.org/citation.cfm?id=344959
http://jucs.org/jucs_12_6/the_berlin_brain_computer/jucs_12_06_0581_0607_blankertz.pdf