Design perspective: Microsoft Build 2023
Leer “Perspectiva de diseño: Microsoft Build 2023” en Español.
Although other technology companies usually attract more attention, Microsoft is still a very important player. Despite having lost the race for mobile operating systems, its presence in the corporate world is dominant and has been building for years a solid offer of cloud services, even flirting with intelligence and vision systems for quite some time.
In this event we continue to talk about Artificial Intelligence (AI), but with a different approach than Google. Let’s look at the highlights:
Bing + GPT
ChatGPT broke molds when it was launched, and Microsoft was quick to license this technology from OpenIA and integrate it into its search engine. They managed to increase the use of his search engine six-fold, and in the process they gave Edge, a very competent browser, a significant boost. In the Microsoft Build event it was announced a deeper integration of the language model, together with the ability to connect with search engine plugins, and even with Edge extensions.
Unlike Google, which continues to take over the content and experience of its users, Microsoft has chosen an open approach, which presents an opportunity for many products and services to integrate. If you work on a project with a community around, this can be an additional point of contact where you can add value in a very organic way.
Copilot, son of Cortana and grandson of Clippy
Since Office’s “beloved” mascot, Microsoft has been offering assistants since 1995. After killing Clippy six years later, they tried again when Siri arrived on the Mac. This time they rescued the assistant from the Halo videogame franchise named Cortana, and integrated it into Windows with little success.
In other news, a collaboration between GitHub (belonging to Microsoft) and OpenAI (creators of ChatGPT) launched in 2021 an assistant for developers that had learned various programming languages from the millions of public projects published on the platform. The result was surprising, and quickly generated curiosity. Also very interesting was the name: Copilot. The value proposition was to accompany and expand the capabilities of programmers, never to replace them.
The announcement of Windows Copilot brings that same concept to Redmond’s operating system in their third attempt to find the perfect assistant. Windows Copilot is integrated across the operating system and all the programs installed, takes into account the context and integrates both with local information and with Bing. Perhaps the most interesting thing is that, after the fever for voice interfaces, Copilot can only be used typing. What could that mean for the next generation of digital assistants?
Azure
Azure is the name that bundles all of Microsoft cloud services. It is one of his main sources of income (almost as much as all Office licenses) and this year they used the event to tell us the principles upon they have built all these capabilities. Like Google, Microsoft knows that credibility is key shaping public perception, and has once again shared a resource about Responsible AI that had been hidden on its website since last year.
Microsoft has also shared how they are integrating everything related to data. These are very interesting announcements, but tremendously focused on the technical. Microsoft Fabric (the name for this unified data processing ecosystem) has serious implications for what your customers will be able to do in a corporate environment. When designing sophisticated B2E solutions this can be an interesting reference, although quite niche.
Azure AI Studio
Apart from all the technical innovations presented about Azure, there is something especially relevant for the design of products and services. At Google I/O 2023 we talked about how much work we have left to do designing interaction patterns with AI. Here Microsoft opened another related debate, such as interaction patterns for the management, administration and testing of AI.
Azure AI Studio is the back office for generative AI. It facilitates the application of trained models to specific scenarios, and uses the interconnection of the different systems that can be integrated into Azure to offer a tool where an administrator selects an AI model, the data sources with which to give information, and the rest of the parameters. It even includes a space where you can test the result of the integration and solve possible bugs.
Azure AI Studio may have a similar impact to the one Wordpress had in 2003, democratizing access to configuring and managing AI models to less technical profiles. For us designers it gives us additional visibility of the challenges related to the management of these models, with the corresponding opportunity to contribute in this space. Google and Amazon already have similar tools in their clouds, and I assume that a multitude of other companies and startups will contemplate similar projects.
What will happen to Xbox?
The gaming segment reports to Microsoft 9% of all its revenues, the fourth most important of the firm. Xbox became a cloud service when the Xbox Cloud Gaming beta was launched, in a market that suffers from an entry barrier for new users due to the purchase of the console and new games, the search for a subscription model is of great interest to Microsoft. Nothing was mentioned in the event, but seeing the transformation that all cloud services are undergoing, and in the amount of game data with which Microsoft could be training models with which to generate levels, content or virtual opponents, I would not be surprised if we saw any gaming related proposal at some point in the future.
Conclusions
The expansion of Copilot to Windows will have a much greater impact than AI integrations in Google Workplace. Windows and the Office suite cap over 74% of the business market, and their cloud solutions serve 23% of companies, the second provider behind Amazon. Besides the fact that it remains to be seen how assistants will evolve with language models such as GPT, Copilot ditches voice interaction to return to text-based input and boasts an unprecedented degree of integration with the system. This means that in a few years the market penetration of Copilot will be absolute, and the familiarity of users with these assistants will allow us to offer increasingly sophisticated AI-based proposals.
The need to evolve interaction models does no longer apply only for users, but also for the entire management and administration of AI-based services. These are very complex tools, and in the race to offer these services to more and more companies usability will become a very relevant differentiating factor, in which designers have a lot to contribute.
Finally, let’s consider the impact that these announcements have on the real availability of AI in the development of new products and services. First Google, and now Microsoft, offer solutions that shorten the time and investments necessary to do so, which makes it more realistic to accept their use in real projects. Amazon already has a catalog of cognitive services that will surely be infected by the improvements in usability that we are seeing and will end up taking off the prevalence of AI and, therefore, the demand for design profiles with these capabilities.
