What’s Next For Artificial Intelligence?

Nigerian software start-up launches an Artificial Intelligent Bot
Nigerian software start-up launches an Artificial Intelligent Bot.
We have stopped thinking about AI as a domain for rocket scientists only.
We have stopped thinking about AI as a domain for rocket scientists only.

Last year was a hugely significant year for artificial intelligence. It was a year of spectacular achievement. For the first time a system was created with the ability to defeat a human at the impossibly strategic game of Go (a feat that has been 20 years in development). And it was the year in which deep learning became the buzzword because although neural network techniques have been used to teach machines for more than 30 years, the deep learning variant is the one that gives a machine second life.

More importantly, 2016 saw a mindset change amongst developers. By and large, we stopped thinking about AI as a domain for rocket scientists only. The world’s software giants (Amazon, Google, Facebook, IBM and Microsoft) have joined forces within the “Partnership on AI”. And open source tools like TensorFlow (originally developed by the Google Brain Team) have had an enormous influence making people realise that they don’t need a Ph.D. to build software based on deep learning anymore. Open solutions can be applied here and now, almost on-click.

Finally – and this cannot be underestimated – AI has simply become fashionable. It’s reported on by mass media and, more and more, referred to by politicians and celebrities. It is trendy to know about it. Even more trendy to buy into it – if you tried to buy an Amazon Echo (or Alexa) just before Christmas, you were too late. AI has never been so cool or received so much air time.

So what can we expect this year? Of course we’ll hear a lot about new applications of deep learning techniques. Over and above that, I predict 3 key areas of major development in 2017.

Cloud and social networks
Social media and networks are extremely powerful nowadays. People, in general, are very enthusiastic about sharing their thoughts and private resources with the world. From a user perspective it is usually pure fun and entertainment but from a business perspective it is the richest data collection and marketing opportunity in history. That is why social networks are strongly connected with cloud platforms and big data analysis. Combining them with AI provides amazing results.

For instance, Cloud Vision API allows you to upload your picture and gives you feedback about your location (based on other’s people images and Google map solutions). It also offers strong sentiment analysis (are the people photographed happy?), and more. This trend will definitely continue and the data will be processed by various services in a more “intelligent” way. So we can expect more surprising features from our favourite social media channels (like improved friend and content search engines) and a new generation of ads: more user-friendly (instead of annoying pop-ups) and better focused on our current needs. A more personalised relationship all round.

Taken a stage further, the technology may even help us to improve our love lives. Imagine matchmaker portals which not only analyse the data you have added to your profile, but also your everyday habits (monitored by PC usage and wearables), interests (based on online search) and lifestyle (shopping history) – all making it easier to find the perfect partner.

Everyday applications
Another market that will see massive development and growth is mobile applications. The reason is obvious: they are relatively easy to implement, distribute and promote (all the infrastructure such as stores, APIs are ready) and the potential profits are very high. So we can expect more applications using deep learning and other AI techniques to emerge as everyday consumer software tools.

One example is Prisma. Release in 2016, it is  a mobile app that turns your photos into works of art using the “styles” (neural network learning patterns) of famous artists. 2017 AI-based applications will probably cover more topics: voice analysis, GPS support and travel planning, smart calendar management, etc.

Smart home, smart car… smart world
We live in a world driven by technology, whether we want it or not. Smart homes were rich people’s toys a few years ago. Today you can buy an Amazon Dot for next to nothing. And you can ask these small devices (connected to the Cloud) about the weather, Wikipedia facts, to play the radio or your favorite song, to read an ebook, or to switch off the lights in your dining room. You can enjoy basic conversation: “Alexa, I love my wife”, “Make sure she knows that”. And you can even ask her for a joke! Google are hot on Amazon’s heels with Google Home.

It won’t be long before all of our household good are integrated. Expect new features in Alexa that inform you, for example, if you are running out of milk (fridge scanner), or that it is high time to use the vacuum cleaner (a self-operating one of course!). Features like these may be available in a matter of months. The technology exists, all that is required is the investment .

Our cars are already full of AI as well. Not the self-driving prototypes of tomorrow, but the cars you can buy today at you local dealer. Parking assistance, track control, roadblock alerts, eco-friendly hybrid solutions – AI techniques are everywhere and look set to multiply as more and more customers see these features as differentiators.

For example, with recent improvements in vision analysis, we can expect cars that recognise road signs and alert the driver if necessary. They’ll also read billboards and intelligently cross-reference what they see with other data that they’ve gathered from your various connected devices (you’ve been searching for a new laptop recently -> that ad tells me that a huge selection is on sale in a nearby store  -> “would you like me to update the route to stop there on the way?”). Yesterday’s science fiction is tomorrow’s reality.

Finally, the current development in “smart” devices is rapid and AI-based natural language processing is a key part of this. So expect huge improvements in this area in 2017. Human/computer interaction is set to change dramatically and voice will be the main channel within a few years in our homes, cars and office spaces.

By Jedrzej Osinski, Cognifide