AI review 2015: Preparing for take-off
When Mark Zuckerberg posts something on Facebook – the very social network he founded – he is sure to not only have the attention of his 37 million followers, but of an even broader audience of the business, science and media community. It has become some sort of tradition that Zuckerberg posts a New Year’s resolution at the beginning of each year. In past years, he has vowed to learn Mandarin and read two books each month. This year he proclaimed to focus on building a simple artificial intelligence (AI) system to run his house. Quite ambitious for a side project, but Zuckerberg’s post underlines the attention AI will receive in the upcoming year. Reason enough, to review what has happened in AI in 2015.
Heavy weights increased M&A activity
Even though AI has received a lot of media attention, the AI community is, in fact, still rather small. No wonder that the war for talent has been fought with increased intensity during the last months. Big shots like Apple, Amazon, Facebook and Google are lined up to access the smartest AI brains through direct hires from university. Amazon even opened a machine learning development centre in Berlin – a city with a vibrant, attractive lifestyle but little relevance for Amazon’s operations – , solely dedicated to attract the best German (and European) AI specialists. At the same time, 2015 has brought an increasing amount of mergers & acquisitions of rising startups in the AI sector as a method of getting access to superb intellectual capital. Early 2015, Facebook acquired the speech recognition startup wit.ai, which provides a natural language platform for developers to build applications with text- or voice recognition features. Later this year, Apple bought VocalIQ, a company focused on natural language (speech) processing. Just these days, news arose that Apple had bought Emotient, a startup which provides a software for facial recognition. Leading technology companies have been heavily investing into different aspects of artificial intelligence in the last year, prepping their in-house capabilities for AI take-off.
Virtual Assistants on the Rise?
This year has also seen the rise of virtual assistant services. Already in February “Magic”, a virtual concierge backed by Y-Combinator entrepreneurs, started its service and evoked a huge social media stir. The service went viral within days, with a long waiting list assembled for product use. A few months later, GoButler went live in Germany and the US – with almost the same value proposition. Except, users do not even have to pay a fee to use the service. How does it work? Well, as of right now users communicate with a “concierge” through a standard messaging interface (e.g. WhatsApp) and can request and ask almost anything (from food delivery to flight booking) – and the concierge does its best to get it done. Is there any smart AI behind it? Frankly, not yet. It is a product that doesn’t scale. There is real people handling the requests. But the pitch of both companies, is – essentially – to build a smart AI system around it in order to handle most request automatically. Facebook, assumably moves into the same direction, with its Messenger service, trying to build upon a step-by-step development of an AI, starting with large human involvement and subsequent gradual phase-out. This, at least what super-focused start-ups like Claralabs, which provides a virtual scheduling assistant for meetings, apply as a best-practice strategy. However, premature holistic approaches, which Go Butler and Magic exemplified in 2015, might overstretch AI promises and lead to user disappointment in the end – or bankruptcy for the companies, which will continue to depend on heavy manpower.
Building the AI infrastructure to succeed
However, 2015 has also brought some very much needed depth into the AI discussions. At the Joint Conferences on Artificial Intelligence, notable characters such as Elon Musk (Tesla, SpaceX), Stephen Hawking and Steve Wozniak (Apple) have signed an open letter, which warns of the terrible outcomes AI advances can have if applied in the military sector. Autonomous weaponry does not only make war an utterly unhuman endeavour, but also increases risk of a mad dictator, turning against humanity as such. What sounds like a science fiction movie, was brought to public attention to increase awareness. Later this year, Musk and Sam Altman, who heads the accelerator program Y Combinator, announced a massive one-billion-dollar non-profit company, which aims to maximize the power of artificial intelligence for the public good. Called OpenAI, the project should live up to its name: aggregate data and AI talent to provide a platform accessible for anyone willing to develop new applications that leverage technology. As AI research will always remain a function of talent and data volume, companies like Google or Facebook were in a prime position. OpenAI might level the playing field and encourage more innovation in AI.
2015, one could argue, might have thus paved the road for 2016. Open AI could form the infrastructure necessary for small AI start-ups to flourish and not fall capture to aggressive acquisitions of the leading tech giants like Facebook and Google. Their behaviour in 2015 underlines that AI will continue to play an increasingly important role in 2016. And most probably, Mark Zuckerberg’s private AI system will not remain the only endeavour taken into a direction, which fully captures the value of AI for business and society.