Microsoft Corp. remains tightly focused on driving artificial intelligence to the cloud’s edge. At this year’s Build developer conference, it made a wide range of announcements that respond to the shift of algorithmic intelligence to edge devices, especially mobile, embedded and “internet of things” devices.

In his Build 2018 keynote, Chief Executive Satya Nadella (pictured) highlighted how Microsoft is investing in AI and other advanced technologies to enrich its core cloud/edge and office-productivity platforms and also its established offerings in collaboration, business analytics and application development.

But before he launched into the product-focused portion of his keynote, Nadella spelled out how Microsoft is addressing three “pillars” of trustworthy information technology:

• Digital privacy: With the deadline for General Data Protection Regulation compliance fast approaching, Nadella assured the audience that, in its internal operations, Microsoft has “been working hard to ensure compliance with [the regulation] by end of this month when it becomes in effect.” He pointed to the fact that they “have hundreds of engineers across the company building the compliance infrastructure” and stated the company is “going to help our customers who use our products and services get compliant.”
• Cybersecurity: To protect society from cybersecurity threats, the scourge of “fake news” and other cloud-borne assaults on society, business and our lives, Nadella stated that “we need to act with collective responsibility across the tech sector to help keep the world safe.” In terms of specifics, he called attention to the fact that Microsoft “recently formed a program to protect our democracy where we’re going to work with the campaigns, the civic society, other constituents, so that we can secure our political process, our democratic process.” He also stated that Microsoft “led a consortium of 34 tech companies with the Tech Accord to ensure that citizens across the world are protected from cyber-attacks. It’s the Digital Geneva Convention of our time.”
• Ethical AI: To prevent “fake news” and other AI-driven applications from going rogue, Nadella signaled the steps that Microsoft is taking internally to bring ethical principles into its R&D, product management and stakeholder engagement activities. “We formed an ethics board inside the company, which is a very diverse group of people who govern the products we build, the projects we engage in. But beyond that, we are also investing in tools, because one of the key things that we have to do is put tools in the hands of developers, designers.” He was short on specifics regarding specifically how it would embed ethical principles into its AI offerings, apart from alluding to the need for better tools for protecting privacy in AI apps and using AI to remove biases and assess trustworthiness of content more accurately.

Foundational technologies

Nadella then presented Microsoft’s investments and announcements in terms of three “foundational technologies” that are driving the shift toward edge-facing intelligent apps:

• Ubiquitous computing: Nadella framed Microsoft’s cloud and edge environments as the foundation for a pervasive computing fabric for a “new app model that is distributed, event-driven and serverless.” Highlighting strong customer growth, he stated that Microsoft is evolving Azure to serve as “the world’s computer,” already having “more regions and more certifications than any other public cloud.” Key to Azure’s future, he said, are its continued development of Azure Stack, Azure IoT Edge and Azure Sphere into a unified computing fabric for the new application model. He stated that Microsoft has launched 130 new Azure capabilities in the past year and launching 70 more this week at Build. He made special note of the fact that AI is rapidly infusing Azure edge environment, stating that “with Windows ML we now have an inference engine for AI that works with the IoT runtime and will be hardware accelerated.” With an eye toward ecosystem enablement, he said Microsoft has more than 230 partners — including those in the computer vision and drone markets — who have certified more than a thousand devices for Azure IoT Edge and he announced Microsoft’s open-sourcing of Azure IoT Edge “so that developers can extend it and contribute to it and take it to all the places that it’s needed in.”
• AI: Nadella emphasized Microsoft “breakthroughs” in AI, such as having achieved human parity of their algorithmic machine-translation technology. He stated that Microsoft’s go-to-market strategy focuses on “commoditizing” AI by putting the technology “in the hands of every developer and every organization.” This, he said, “requires us to be able to scale… AI across both the cloud and the edge [by providing] the most productive tool chain to be able to create … and customize [the technology].” AI’s ubiquity will also require “openness when it comes to frameworks and infrastructure. With that as prelude, Nadella discussed how Microsoft’s AI cognitive services continue to expand to 35-plus offerings speech, vision, language, machine translation, conversational UI, spatial understanding, ambient computing and other customizable capabilities. He discussed Microsoft’s launch of 100-plus new features for its Bot Framework to drive development of AI-powered conversational user interfaces. Furthermore, he emphasized Microsoft’s investment in a growing range of its own AI-powered device technologies, such as Kinect and HoloLens, and support for an expanding partner ecosystem that’s building these capabilities into mobile, embedded, IoT and mixed-reality apps. He highlighted Microsoft’s Cosmos DB as a core data lake for customers and partners to build and train AI apps; Microsoft’s Kubernetes service for tying together an orchestrated toolchain and runtime environment for distributed, containerized and cloud-native neural-network apps; Microsoft’s open development environment that allows developers to build in the AI modeling framework and deploy to the back-end AI hardware platform of their choice.
• Multisense, multidevice experiences: Nadella discussed the ongoing evolution of Microsoft 365 into a “multidevice, multisense experience, service and platform” for knowledge-worker productivity. For developers of these applications, he highlighted advances in Microsoft’s Fluent design tooling, which enables developers to modernize legacy Windows apps by adding experiences that are intelligent, multi-device, sensor-aware, cloud-connected, collaboration-oriented and conversational. He specifically called out advances in Microsoft 365 due to integration of the graph technology launched at last year’s Build and the implementation of this capability into both its own products — such as Power Apps, Power BI, Dynamics 365, HoloLens, and Teams — and those of a growing range of partners. “It’s this [Microsoft Graph] ability to have users’ activity across devices, across their life and work, explicitly represented, that you can use. This data is owned by users and organizations. And they give explicit consent to developers to be able to use that data, extend that data, that means your applications can be part of the Graph. This is not a closed graph, it’s an extensible graph, so that we can collectively add more through our customers with these cross-device experiences, these multi-sense applications.”

These roadmap priorities were consistent with those in Nadella’s keynote at last year’s Build. However, the product-focused substance of Nadella’s keynote this week barely addressed any productization of solutions for helping customers address privacy, cybersecurity and AI ethics challenges.

Developer offerings

Instead, Microsoft rolled out a wide range of new and enhanced offerings to make the lives of developers easier and more productive. Chief among the developer-focused announcements at Build 2018 were the following:

• .NET Core 3.0: This new tool enables developers to use the latest version of .NET and have their application run in a stand-alone .NET environment.
• Visual Studio IntelliCode: This new tool, in preview, provides intelligent suggestions for improving code quality and productivity.
• Visual Studio Live Share: This new tool, in preview, lets developers easily and securely collaborate in real time with team members who can edit and debug directly from existing tools such as Visual Studio 2017 and VS Code. Developers can use Live Share with any language for any scenario, including serverless, cloud-native and IoT development.
• MSIX: The new tool provides a simple way to containerize large catalogs of applications. It inherits all the functionality from Universal Windows Platform, including reliable, robust installation and updating, as well as a managed security model and support for both enterprise management and the Microsoft Store.
• Azure Machine Learning updates: Updates to Azure Machine Learning and JavaScript custom functions allow developers and organizations to create their own additions to the Excel catalog of formulas.
• Windows Machine Learning: This new platform enables developers to develop machine learning models in the intelligent cloud and then deploy them offline and in high performance to the PC platform.
• Azure Kubernetes Service: This new service allows developers to simplify how they build and run container-based solutions without deep Kubernetes experience. AKS, which will become generally available in the next several weeks, integrates with developer tools and workspaces, DevOps capabilities, networking and monitoring tools in the Azure portal. In addition, Microsoft announced that it is now offering Kubernetes support for Azure IoT Edge devices.
• GitHub support in Visual Studio App Center: Microsoft released integrated support for GitHub in Visual Studio App Center. This enables GitHub developers to build apps for iOS and Android devices that automate DevOps processes right from within the GitHub experience.
• Azure Cosmos DB updates: Microsoft announced support for new global-scale multimastering features for cloud and edge scenarios in Cosmos DB as well as VNET support for increased cloud security.
• Microsoft Azure Blockchain Workbench: This new tool makes it easier for developers to build blockchain applications by integrating an Azure-supported blockchain network with such Azure cloud services as Active Directory, Key Vault and SQL Database.

AI projects

The core announcements at Build 2018 were addressed to the growing community of developers working on AI projects, especially those that involve deployment into cloud, edge, IoT, mobility, ambient-computing or mixed-reality devices environments. For them, the following announcements were most notable:

• Edge-app speech-recognition development toolkit: Microsoft announced a Speech Devices SDK for building voice-enabled scenarios into digital assistants embedded in smart cars, smart homes, smart speakers and other devices. The SDK provides superior audio processing from multichannel sources for more accurate speech recognition, including noise cancellation and far-field voice.
• Azure cognitive-service edge-app development announcements: Microsoft announced that its Custom Vision cognitive service will now run on Azure IoT Edge, making it the first Azure Cognitive Service to support edge deployment. This will enable drones, industrial equipment and other devices to take critical action rapidly without cloud connectivity. Microsoft has promised with more cognitive services for Azure IoT Edge to be launched in the next several months.
• Open-sourcing of Azure IoT Edge Runtime: Microsoft has open-sourced this technology, thereby enabling customers and partners to modify, debug and have more transparency and control for applications running AI and other apps in embedded, mobile and IoT endpoints.
• Real-time, distributed, low-latency AI edge-to-cloud technology: Microsoft announced preview of a distributed, real-time, low-latency AI technology for Azure cloud and edge/IoT environments. “Project Brainwave” is fully integrated with Azure Machine Learning and provides developers with access to AI-optimized Intel FPGA hardware and ResNet50-based neural networks within Microsoft data centers. Microsoft claims that Project Brainwave offers five times lower hardware latency than TensorFlow Processing Units for real-time AI. They’re working with system partners and customers to take FPGA and deliver “Project Brainwave” technology into Azure deployments worldwide.
• Mixed-reality and ambient-computing device technology product announcements: Microsoft announced Project Kinect for Azure, which provides a package of sensors for HoloLens developers of mixed-reality apps. Project Kinect, the fourth generation of Microsoft’s motion-sensing technology, includes a time-of-flight depth camera, onboard compute capabilities, fully articulated hand tracking and high-fidelity spatial mapping. Project Kinect packages this hardware technology in a small, power-efficient form factor and integrates with Azure AI for mixed-reality, edge AI and ambient intelligence. Also, Microsoft announced Remote Assist, which enables user to collaborate remotely with heads-up, hands-free video calling, image sharing and mixed-reality annotations. Furthermore, it announced Layout, which enables users to design spaces in context with mixed reality, such as by importing 3-D room-layout models, creating designs as high-quality holograms and sharing and editing these models with stakeholders in real time.
• Computer vision edge-device partnership: Microsoft announced a partnership with Qualcomm Technologies, Inc. to create a computer vision AI developer kit running Azure IoT Edge. This solution will provide the key hardware and software required to develop camera-based IoT solutions. It will allow developers to create solutions that use Azure Machine Learning services and take advantage of the hardware acceleration available via the Qualcomm Vision Intelligence Platform and Qualcomm AI Engine. The camera will also power advanced Azure services, such as machine learning, stream analytics and cognitive services, that can be downloaded from the cloud to run locally on the edge.
• Drone edge-AI partnership: Microsoft announced that it is partnering with DJI, the world’s biggest drone company. The partners have created a new software development kit that enables Windows 10 connected devices to remotely control and stream data in real time from drones. The SDK also allows Windows developers to integrate and control third-party payloads such as multispectral sensors and custom actuators. The SDK is available now as a beta preview to attendees of Microsoft Build and will be broadly available in fall 2018. In addition, DJI and Microsoft are creating new developer tools for Windows and Azure IoT Edge Services to enable real-time AI and machine learning for drones. They are also co-developing solutions that will leverage Azure IoT Edge and Microsoft’s AI services to enable new scenarios across agriculture, construction, public safety and more. In addition, DJI has selected Azure as its preferred cloud provider for commercial drone and SaaS solutions.
• AI accessibility for the disabled: Microsoft announced a new five-year, $25 million program to use AI in augmenting the capabilities of disabled people. The program includes grants, technology and AI expertise to accelerate the development of accessible and intelligent AI solutions in apps, edge devices, back-end Azure cloud services. It will leverage Azure Cognitive Services to help developers create intelligent apps that can empower people with hearing, vision and other disabilities. It will include real-time speech-to-text transcription, visual recognition services and predictive text functionality that suggests words as people type are just a few examples.

The upshot

All of these announcements demonstrate Microsoft’s deep investments in innovative solutions in AI, cloud computing, edge applications and so on. However, few of Microsoft’s product announcements at this year’s Build had any specific relevance to the privacy, cybersecurity or AI ethics themes that the company’s CEO spelled out in his keynote.

Inexplicably, the company missed a prime opportunity to reach out to enterprise information technology professionals who are racing to start complying with GDPR by the end of this month. Seeing as how big-data analytics competitors such as Hortonworks Inc. are going full-tilt into comprehensive GDRP-focused compliance and stewardship portfolios, it’s not clear why Microsoft had no specific announcements to that effect this week.

In addition, Microsoft missed the opportunity to explain how it can help customers use AI to address any algorithmic accountability requirements — such as explaining in plain English how embedded machine learning algorithms made particular automated decisions — that stem from GDPR and other such mandates in many countries. Considering how much cutting-edge research Microsoft is doing into interpretable and transparent machine learning, it’s dismaying that the vendor isn’t putting a big push on commercializing it.

Furthermore, Microsoft failed to discuss the growing need for antiadversarial tools for protecting AI models from hack attacks associated with bogus example data being introduced in training or inferencing scenarios. As I discussed here recently, IBM Corp. is one of the pacesetters in this regard, having launched its pioneering Adversarial Robustness Toolbox recently at the annual RSA Conference.

And Wikibon was disappointed that there was no mention in Microsoft’s roadmap of browser-based AI as an approach that can drive algorithmic decision making at the client side, thereby helping to protect customer personal data from third-party access. Considering the investments that Microsoft is making in Blazor technology for rich progressive Web apps, it would have been interesting to hear if they’ve connected AI-based Web app development approaches, such as the Google Inc.-developed TensorFlow.js, into their roadmap. Again, there was no Microsoft announcement Build to address this emerging new AI development practice.

For Wikibon’s current assessment of Microsoft’s positioning in the big data analytics market, please check out our latest market study’s vendor profiles here.

Photo: Microsoft