Microsoft Doubling Down on AI

This week, Microsoft announced improvements to Office 365 and a new camera app for iOS called Microsoft Pix. While these two products are very different, in nature they clearly demonstrate how Artificial Intelligence for Microsoft is much broader than Cortana and bots.

Keeping Word competitive by adding smartness

As Matt illustrates in today’s article, Word, the unquestioned word processing tool of choice, has been challenged by new tools especially around cloud and collaboration. Even users like millennials, who clearly favor it for some of their work, look elsewhere when it comes to collaboration. Keeping up with competition is essential and, while there are areas where I think Office should invest more such as real-time collaboration, this week’s additions – Researcher and Editor – are certainly welcome.

Researcher lets you incorporate reliable sources and content into your paper. Within the document you are writing, you can look at material that is relevant and add as a proper citation. Researcher uses the Bing Knowledge Graph to pull content from the web. Researcher responds to Google’s introduction of Google search into Google Docs. This tool should be particularly well liked by students for their research papers.

Editor is really where AI comes in. In the blog announcing the enhancements, Kirk Koenigsbauer, corporate vice president for the Office team, described Editor as follows:

“Leveraging machine learning and natural language processing—mixed with input from our own team of linguists—Editor makes suggestions to help you improve your writing. Initially, it will help you simplify and streamline written communications by flagging unclear phrases or complex words, such as recommending “most” in place of “the majority of.”

At first, Editor will help with writing style suggestions to improve clarity. In the Fall, it will add the ability to explain why words or sentences may not be correct, becoming more like a teacher versus just a correction tool.

As I mentioned, Editor is a “smart” editor so it will improve over time; not just as it gets more features but as it learns about how you write. The fact Microsoft chose to give these new features a name reflective of a person’s job certainly speaks to wanting to develop a relationship with the users to increase dependence. As I’ve discussed before, humanizing virtual assistants helps deepen the relationship with the user. In our interviews with millennials, it was clear the high-level of dependence on Google Docs for their collaborative work has developed an emotional connection of sorts that drew feelings of sadness and loss at the thought of losing access to it. These connections are much harder to break.

As someone who has worked with human editors over many years and, as much as it was always a love-and-hate relationship, I came to depend on them and appreciate that my work was better because of them. Editor has the potential to do the same for many, especially as it becomes a teaching tool versus just a quick fix.

Microsoft Pix is a smart camera but also a smart business move

On Wednesday, Microsoft introduced a new iOS camera app called Pix that uses artificial intelligence to eliminate human error by capturing several pictures to find the best color, exposure and focus. When Pix detects a person in the shot, it optimizes for him/her. It assumes if you are pointing the camera at a person, that’s who you want to get a picture of. Pix also has an option to mimic Apple’s Live Photos by using the burst set of pictures captured and combines them into a loop video. This functionality is only triggered when movement is detected and cannot be manually set.

Pix is not just a smart app for users. It is a smart solution for Microsoft to get around the fact there are not millions of Windows phone in the wild. Microsoft is clearly following the “If you can’t beat them, join them” mantra and creating apps for iOS to increase user engagement in different ways than through owning their hardware. If I look at my iPhone and iPad, I now have about ten Microsoft apps I use regularly and I’ve come to depend on while being very aware they are Microsoft apps. This is ultimately what will help Microsoft’s AI aspirations. The more I use those apps, the more Microsoft will know about me and, in return, the more useful my interactions with those apps will be. Of course, this is not the ideal scenario as not being embedded in the hardware as much as possible has limitations. With Pix, for instance, the biggest hurdle to adoption will be the fact most iPhone users access their camera from the lock screen. Getting to Pix will require an extra step users will take only if the product delivered is considerably better. I need to use the app more to decide if that is the case for me.

Microsoft will have to continue to be smart, balancing the deeper integration on its own devices and devices running Windows as well as leveraging as much as possible the interactions with users on other platforms. While not as deep, there is still learning and data to be had from those platforms too so why not?

Published by

Carolina Milanesi

Carolina is a Principal Analyst at Creative Strategies, Inc, a market intelligence and strategy consulting firm based in Silicon Valley and recognized as one of the premier sources of quantitative and qualitative research and insights in tech. At Creative Strategies, Carolina focuses on consumer tech across the board. From hardware to services, she analyzes today to help predict and shape tomorrow. In her prior role as Chief of Research at Kantar Worldpanel ComTech, she drove thought leadership research by marrying her deep understanding of global market dynamics with the wealth of data coming from ComTech’s longitudinal studies on smartphones and tablets. Prior to her ComTech role, Carolina spent 14 years at Gartner, most recently as their Consumer Devices Research VP and Agenda Manager. In this role, she led the forecast and market share teams on smartphones, tablets, and PCs. She spent most of her time advising clients from VC firms, to technology providers, to traditional enterprise clients. Carolina is often quoted as an industry expert and commentator in publications such as The Financial Times, Bloomberg, The New York Times and The Wall Street Journal. She regularly appears on BBC, Bloomberg TV, Fox, NBC News and other networks. Her Twitter account was recently listed in the “101 accounts to follow to make Twitter more interesting” by Wired Italy.

5 thoughts on “Microsoft Doubling Down on AI”

  1. Great article. I have one issue though.

    Why do we have to keep on repeating the mantra “the more personal data you get, the better”, when the AI features that Microsoft has just shown us strongly suggest that you can create market-leading AI solutions without any prior personal data?

    For the Office AI features, for example, wouldn’t it be better to train your AI using professionally edited and published newspaper articles and books. Do you really want to train your Editor AI using poorly written Gmail messages?

    For photos, wouldn’t you prefer to train your camera AI using published archives of professional photos rather than random selfies on Instagram or Google Photos?

    The way I see it, Microsoft’s AI solutions are proving that this is indeed the case. As Ben mentioned in an earlier post, the distinction between personal vs. communal AI is important, and I strongly suspect that personal data collection is unimportant for the majority of communal AI.

    Google and Facebook would like us to believe that personal data collection is essential for providing sophisticated AI solutions. They want to justify their intrusions of privacy on the grounds that it will make our lives better, even if in reality they only need it for advertising purposes. I have yet to see proof of Google’s and Facebook’s argument, and the examples that Microsoft has shown actually disprove it. You don’t always need personal data.

    Let us be open to the possibility that personal data might not be necesary for good AI after all.

    1. Don’t disagree with you but as Ben also pointed out there is a personal side to AI and that is what I was referring to. If I think of Editor for instance the human editors I work with correct my grammar and improve my sentenced but remain true to my style – mainly rambling on long sentences 🙂 the same with pictures. So it is not so much data but personal preferences

      1. Interesting.

        The current autocorrect in word has a lot of settings. Maybe Editor AI will have settings to leave sentence length alone. In that case, you’ll be able to preserve your style with a flip of a setting, without the AI learning about you at all.

Leave a Reply

Your email address will not be published. Required fields are marked *