By Dr Robert O’Toole
How many features, small or large, things that users can change, are there in the entire Microsoft 365 platform? Who knows! I bet Microsoft aren’t keeping track of its rapidly escalating complexity. Maybe it’s 10,000. Possibly 100,000? It could even be in the millions, if we include all of the combinations that modify the meaning of each option. I recently tried to write down all of the options for configuring meetings in Microsoft Teams. It was overwhelming. I know Teams quite well, but there was still far too much for me to comprehend.
This is madness. And Microsoft know it is. So why do they do it? Partly it results from their product design approach (let’s not call it a strategy). Compare Microsoft to Apple, whose products are far simpler, focussed, but also much less feature rich. Apple does less better. Much less, much better. They start with a strong idea of a small set of experiences and outcomes that they think many people will like (and which they themselves care about), and then they build streamlined software to enable those outcomes in a slick and always enjoyable way. That explains why pro apps like Final Cut have always received less of a focus than apps like iMovie. Pro users are far more demanding, far more discerning about detailed features. They are a hard bunch to please. Apple certainly don’t want to get stuck in an unhealthy relationship with a vanishingly small pro user base (always the threat Adobe have struggled with). Microsoft, on the other hand, are responding to the demands of many users (especially powerful business clients) in many fields across the globe, and adding in as many features as possible in a desperate pursuit of advantage – sometimes their approach is like throwing mud at a wall to see what sticks.
But there’s another factor enabling Microsoft to believe that it will all work out OK: an almost fanatical technological optimism. Their embrace of AI (with a very close relationship with the OpenAI company behind ChatGPT) is the latest version of their corporate religion. The belief that holds their world-view together is that conversational AI will make the complexity manageable for all users. And that might just work. Although we have to remember Microsoft’s first attempt at this – anyone alive and using a PC between 1997 and 2000 will remember the cursed Office Assistant Clippy, an anthropomorphic paper clip character who would pop-up uninvited, claiming to understand what the user is hoping to achieve, and offering wizards and templates to speed them on their way. It was terrible.
Now imagine Clippy today, reborn as Microsoft Copilot – an intelligent assistant based on ChatGPT (and some other tools). Copilot will have learned about your needs and your methods by inspecting every little detail of your work and the work of your organisation. It knows what you are likely to need before you even ask for it, and starts putting arrangements in place to anticipate your requests. In a chat in Teams you and a colleague decide that you need to arrange a meeting to make some decisions. You start a conversation with Copilot, iteratively choosing the type of meeting, and configuring all of the many settings. Copilot guides you through the complexity by making informed guesses as to what might be useful. Dates and times are suggested, not simply based on availability, but also on preferences, styles, context, gleaned from the vast pool of data it has access to – including measures of performance in the meetings. As this is a decision making meeting, we need to examine the data and the options. They need to be compiled and presented. This is all honed through a friendly, human-like, conversation with the intelligent assistant. At some points in the planning process, other people need to be consulted. Not a problem in this hyper-connected world. And to make it absolutely perfect, one of the many thousands of new features is applied: it even orders the drinks and snacks for the meeting, tailored to everyone’s needs and likes.
If this gamble works, Microsoft’s ploy of drawing everything into its ecosystem and adding gazillions of features will have paid off. If. Big if. I’ve played with Copilot a bit, and the results so far are not impressive. No better than ChatGPT, which is pretty narrow in the scope of its understanding. Considering the case above, what kind of intelligence is needed for it to be useful? – for it to be better than just annoying? What kind of mental model not just of the Microsoft platform but also of the business and its people, changing all the time?
There might be an alternative. Maybe we need to simplify what we are doing and how we do it. Maybe we just need fewer better experiences and products? Maybe Microsoft is leading the world down the wrong path?
Read more…
Want to know more about the bizarre life of Clippy the Office Assistant? Read this great story by Benjamin Cassidy in the Seattle Met.
The Real Reason Voice Assistants Are Female (and Why it Matters) by Chandra Steele
How AI might kill you today, but not through the cold metal hands of a killer robot by Robert O’Toole
Leave a Reply