If anyone thinks the annual Microsoft Build! Conference is only for techies, than you are wrong!
First of all, there is an enormous Data & AI zone where anyone can talk to the Microsoft product engineering teams. A huge opportunity if you want to know how Microsoft looks at building its platform and how their tools can help you achieve more.
Secondly, Satya’s keynote was very Business oriented. Almost half of his talk was about AI and more specifically how we should all contribute to responsible and ethical AI. He answered questions like “How are we using AI at Microsoft?” or “How do we at Microsoft think you should deal with biased data?”
During our first day at Build! It became clear that Microsoft made it its mission to help any company become an ‘AI First Firm’. They want to provide an environment, a platform that enables companies become data driven, using the data of one tool to make predictions for another.
One of the most interesting talks we saw was from Lila Tretikov. She said that “AI will eat software”, meaning that new software will always have a form of AI incorporated. AI will augment software and ultimately help us achieve more.
Like every year, a bunch of new stuff was announced:
Explainable Boosting Machine (EBM)
This is a new ML algorithm that kills many other algorithms because of accuracy and speed but most of all because it tells you why it predicts certain value. Business stakeholders more and more want to know why an AI makes a certain decision. The idea that AI is a black box that just makes certain decisions and isn’t able to explain why and how it made that decision is no longer acceptable, and this is also why Microsoft is now contributing to what we can call XAI (“explainable AI”).
EBM is an open source algorithm that is available from now.
Have a look at: https://github.com/Microsoft/interpret
Form Recognizer is a new solution that will help automate form scanning and extraction of the data in that form. We’ve seen demo’s on both typed as handwritten scans.
Homomorphic Encryption is a new type of encryption technology that makes it possible to compute on encrypted data. It means that the owners of the data can see the data unencrypted (client side) but is stored and send to the server encrypted. But encrypted in a way that you could still make summaries or build machine learning models on top of it.
With Microsoft SEAL, the cloud operators will never have unencrypted access to the data they are storing and computing on. This is possible through homomorphic encryption technology, that allows computations to be performed directly on encrypted data. Data privacy relies on state-of-the-art cryptography (mathematics) and all information released will be controlled by the customer.
Stay tuned for more news from Build! 2019.
This article was originally posted on our blog; https://blog.arinti.be/build-2019-day-1-ef548dc0bd90