Microsoft Showcases Its AI Leadership

Strategy Addresses Ethics and Human Aspects of Artificial Intelligence

Last week, I attended a Microsoft event in London to celebrate the 20th anniversary of its research centre in Cambridge. The event focussed on the “intersection of artificial intelligence with people and society”. As part of CCS Insight’s growing research and upcoming buyer surveys in this exciting field, here I’ll summarise and assess the main announcements. For a more comprehensive analysis, please see Instant Insight: Microsoft Stresses the Human Aspect at Artificial Intelligence Event.

Artificial intelligence has become central to Microsoft’s overall strategy over the past 12 months. In September 2016, the company set up a dedicated research group in this area under Harry Shum. Consisting of 7,500 scientists, researchers and engineers from its product teams, the group aims to speed up the creation of artificial intelligence products and to better promote its efforts against those of Google, IBM and Amazon as well as against the growing influence of Chinese players.

Back in May 2016, at Microsoft’s Build developer conference, artificial intelligence took centre stage as the company announced it had more than 500,000 developers using its Cognitive Services, a set of developer tools and APIs in areas such as sentiment detection, vision, speech, knowledge, search and language understanding.

Dr Shum opened the event by highlighting Microsoft’s vision, which centred on people and amplifying humans with “intelligent technology that will reason with, understand and interact with people as well as solve some of society’s fundamental challenges”. Microsoft announced several projects backing up this vision. AI for Earth is a programme that hopes to tackle environmental challenges related to water, agriculture, biodiversity and climate change. The company said it would increase access to its tools, services and support to research bodies and non-governmental organisations. It will also spend $2 million on grants for the programme in its next fiscal year.

The firm also announced a new research and incubation hub, called Microsoft Research AI, within its research labs unit. This is a more focussed division of about 100 researchers dedicated to solving specific problems associated with the technology. The division aims to unite research in currently disparate areas of artificial intelligence such as deep learning, machine perception and natural language processing. Microsoft believes the integration of these disciplines will cause the next technology breakthroughs, such as machine reading. It announced the release of Seeing AI, an iOS app with voice processing and image recognition to help the visually impaired get information by using a smartphone camera.

Ethics and Design

A major focus of the event was on ethics and the design of artificial intelligence. These areas have become contentious issues for critics concerned about bias and negative uses of the technology. Microsoft will create an ethics oversight panel called Aether, which will act in an advisory capacity. It will produce a design guide for product development based on CEO Satya Nadella’s 10 principles for artificial intelligence development announced in June 2016. These announcements build on the principles laid down in Microsoft’s founding role in the Partnership on AI, a cross-industry initiative with Amazon, Facebook, DeepMind, Google, Apple and IBM looking at ethical uses of artificial intelligence.

Microsoft also promoted its research on user experience design for artificial intelligence. This aims to help engineers build systems that strike the right balance between intelligence and empathy and put human augmentation at the centre of the design process. Although Microsoft has more than 100 million users of its chat bot services such as Rinna in Japan and Xiaoice in China, it admits that very little is understood about user experience preferences given the immaturity of the field.

The event looked at promoting ethical artificial intelligence and emphasised the importance of trust, transparency and accountability in the use of the technology, as well as the benefits it brings to society in healthcare and the environment, for example. These have become focal points in educating the public about the technology. In my opinion, ethics aren’t yet a leading indicator with early adopting enterprises when selecting technology suppliers, but it’s essential that Microsoft promotes the technology in this way at this very early stage of development given the real fear, and “the elephant in the room” in most cases, that it will destroy jobs.

Infusing Microsoft’s Products

Another important aspect in Microsoft’s strategy is to promote how its artificial intelligence innovations are being integrated into its products. There were several announcements in this area, including the launch of the Bing Entity Search API, updates to its Project Prague gesture kit and the general availability of Presentation Translator, a PowerPoint add-in that can subtitle presentations in over 60 languages.

The main challenge has been the lack of awareness of these capabilities, which have seen slow adoption so far. Microsoft is hoping that its research into design will enable it to better understand where and when users want more, or less, artificial intelligence in their daily workflows. I believe that understanding this will be crucial to the success of not only Microsoft’s productivity, collaboration and CRM applications, but to the company itself.

Conclusions

Overall, with several customers on display including Dixons Carphone, Cochrane and Prism Skylabs, the event successfully showcased Microsoft’s early leadership in this field. Its support among developers, its data graph assets and its cloud capabilities are major strengths as the battle for supremacy escalates.

However, the enterprise market for artificial intelligence that will manifest over the next decade won’t be a winner-takes-all battle. I expect that many enterprises will pursue flexible architectures for artificial intelligence, using multiple suppliers as they do for cloud services today. They’ll also follow open approaches and avoid supplier lock-in, instead integrating specialist technology for specific business applications.

Microsoft’s open strategy is aligned to capture this trend, but it will also need to bring artificial intelligence out of the developer ivory tower and into the real world. To do this, it must begin to educate businesses on the specific uses that create the most business value and to help with implementing best practices. In the case of deep learning technology, Microsoft will also need to address the emerging requirement that systems must be explainable and transparent for users. These are the first steps along a long road of opportunity for the company.