Posted on Leave a comment

Quick AI Insights at #MSBuild in an Overstuffed Tech Event Week

We are in the midst of one of the most packed tech event weeks in recent memory. This week alone, Amalgam Insights is tracking *six* different events:

This means a lot of announcements this week that will be directly comparable. For instance, Google, Microsoft, Red Hat, SAP, and ServiceNow should all have a variety of meaty DevOps and platform access announcements. Google, Microsoft, SAP, and possibly IBM and ServiceNow should have interesting new AI announcements. ServiceNow and Red Hat will both undoubtedly be working to one-up each other when it comes to revolutionizing IT. We’ll be providing some insights and give you an idea of what to look forward to.

When it comes to Microsoft Build, we’ve already seen some introductory announcements as Microsoft positions itself as a democratizer of AI. Our quick take is that Microsoft is trying to democratize AI on three fronts.

For experienced developers, Microsoft is providing choice on frameworks and model creation while improving governance and lineage so devs don’t need to spend as much time on raw documentation. Microsoft is highlighting both “MLOps” to help manage machine learning at scale as well as highlighting advances in the ONNX (Open Neural Network Exchange) Runtime support that makes machine learning workloads more portable across chips and platforms.

For current and potential model builders, Microsoft is opening up search and drag-and-drop modeling capabilities. And then for coders who simply want functional AI-based services, Microsoft is providing form, writing, graphic, language, and a new Decision capability to help people make more personalized decisions. Although these are not novel, since IBM has been focusing deeply on language, vision, and reproducability while both Google and Amazon have focused on functionality, Microsoft has over a million developers that access Azure Cognitive Services. That audience of skilled developers who will be building and customizing Azure-based apps provides the most valuable expansion of AI.

The drag and drop model building is interesting, but model building still requires a minimum level of mathematical knowledge to understand what you are building, no matter how simple the interface is. And the ONNX is interesting in terms of the vendors not involved, namely Amazon and Google. This play by Facebook, IBM, and Microsoft to play together in supporting AI frameworks starts to show how there are aspects of AI management and deployment that are becoming commodities and abstracted. This is great for customers as it shifts the focus for customers from a fundamental question of “can we make this work?” to longer term concerns around continuity, portability, backup, and governance.

Tom Petrocelli especially notes that “What Microsoft (and Google and AWS) are doing is trying to create standard machine models that, when fed sufficient data, will work for most applications. This is the black boxing of AI, something we collectively predicted a year ago. Until the average developer without a Ph.D. can use AI tools, it won’t achieve its potential both as a technology and a revenue stream. This is the hard lesson that IBM is learning about Watson.”

(IBM has re-tooled with Watson Studio, Watson Knowledge Catalog, Watson Machine Learning, and Watson OpenScale. It will be interesting to see how much of this is showcased at Advantage 2019.)

Over the next few days, we’ll be taking a look across other shows as well to see where key AI, DevOps, machine learning, ITOps, and IT Cost Management announcements are taking place and how they compare both to contemporaneous announcements as well as Amazon, Oracle, Dell, and other key players in the AI and IT ecosystem.

Leave a Reply