Microsoft’s Plans for AI
The adoption of AI in business and society is being spurred on by tech giants with resources to design, build and roll out services affordable and simple enough for everyday use. Microsoft is one of those at the forefront. This year, the words “artificial intelligence” appeared in a vision statement for the first time, reaffirming that smart, learning machines are considered central to everything they do.
While it may only just be beginning to shout about it, Microsoft has been building intelligent functionality into many of its products and services for some time. If you regularly use Skype, Office 365, Cortana or Bing, you have probably come across them.
Machine learning – which is what most people mean at the moment, when they use the term AI – helps to match searches with useful results, and gives the Cortana virtual assistant the ability to improve and become more helpful over time. Through Skype, it enables chatbots to run on its communications platform, where they can be used for customer services or accessing services such as weather or travel information.
Within its Office enterprise productivity suite, as well as assistance from Cortana, Microsoft has been rolling out AI-assisted features designed to offer help with everyday tasks, such as live translation of recorded speech.
These are great examples of how specialized AI – designed to carry out one single task and become increasingly good at it – has already become embedded in our lives. Microsoft however has made it clear that their ambition goes further. It is working towards the goal of generalized AI – intelligent machines which can turn their talents to any task.
Harry Shum, executive vice president of its AI and Research group, has said “Computers today can perform specific tasks very well, but when it comes to general tasks, AI cannot compete with a human child.”
The Research and AI group was founded in 2016 as Microsoft’s fourth engineering division, alongside the Office, Windows and Cloud teams. In under one year it has grown to 8,000 employees. It is safe to say that after losing out on the last seismic change in the technological landscape – the jump to mobile – it does not want to be slow off the starting block again.
Aside from the services it makes available directly to end users, Microsoft’s backbone services are made available to organizations wanting to build their own intelligent tools. The overarching framework is known as Microsoft AI Platform, which ties together a number of more specialized packages such as Microsoft Cognitive Services, Microsoft Cognitive Toolkit and Microsoft Bot Framework. These work by allowing AI (machine learning) algorithms to be deployed on the powerful and popular Azure cloud computing platform, with users simply paying for processing and storage as it is required. Removing the need for organizations to host their own expensive and fast-ageing infrastructure is key to CEO Satya Nadella’s vision of “democratizing AI”.
The tech giant also has an interest in development of autonomous vehicle technology. This year it announced that it is partnering with the leading Chinese search engine Baidu to develop a platform for self-driving cars. These vehicles will rely heavily on artificial intelligence to interpret data from onboard sensors and react appropriately to driving hazards.
Most recently, Microsoft announced new technology designed to accelerate machine learning algorithms to real time. Known as Project Brainwave, it uses programmable processors known as FPGAs to run the sophisticated and compute-hungry algorithms. Essentially this is software which can be programmed directly onto a programmable chip, enabling commodity hardware to function as specialized deep neural network processing units.Microsoft is positioned to capitalize on this due to the investment it has made in installing FPGAs in its data centers worldwide over recent years.
Microsoft is also developing industry specific AI applications. The company has just announced a new healthcare division based on artificial intelligence with the aim of developing predictive analytic tools that can alert people about medical problems, help diagnose diseases, and recommend the right treatments and interventions.
Put together, these projects represent the culmination of an aim first stated by the company’s most famous figurehead, founder Bill Gates, back in 1991, when he said that he believed computers would one day see, hear and learn just as humans do.