Last week I had the opportunity to present an Artificial Intelligence primer to a group of senior leaders at Microsoft. I spent several days putting this content together so I figured I’d share. 🙂
What is AI?
Artificial intelligence (AI) is generally considered the next big technological shift so it’s an incredibly hot topic with a lot of people talking about it. Depending on who you talk to, you may hear varying definitions, but Microsoft generally defines AI as an information system, inspired by biological systems, designed to give computers the human-like abilities of hearing, seeing, reasoning, and learning.
Despite all the buzz, AI is probably one of the least understood technologies. That’s not surprising since AI isn’t something you can see or touch. When AI is implemented well, you may not even realize you are using it. In fact, most of us are probably relying on technology that uses AI without even knowing it. If you’ve recently used Cortana, Bing, or built a PowerPoint deck, you’ve probably used AI.
AI is not one universal technology. It’s an umbrella term that usually refers to multiple technologies like machine learning, deep learning, computer vision, natural language processing (NLP), and many others. These technologies can be used individually or in combination to add intelligence to applications and do things that are like what people can do. For example, we can use AI to:
- Perceive images and sounds, and recognize them (classification)
- Reason over large amounts of data to identify patterns (clustering)
- Help decision-making by making predictions and understanding the relationship between various people, places, or things (regression)
Since AI is an umbrella term, people will often use AI and machine learning interchangeably. Though AI and ML are related, they aren’t the same so it’s important to understand the relationship. Machine learning is considered a subcategory of AI; focused on analyzing data to train models that make predictions. AI leverages ML algorithms to give systems a sense of intelligence.
AI is not a new field; much of its theory and technology has been developed over the last 70 years. However, AI has recently moved into the mainstream due to these factors:
- Massive computing power of the cloud
- Availability of enormous datasets that can be used to teach AI systems
- Breakthroughs in developing AI algorithms and improving AI methods such as deep learning
Microsoft has advantages in every one of these areas and that’s helping us more quickly infuse AI into many of our core products and services. These advantages include the immense computing power of Azure, access to comprehensive data spanning services like Bing, Office, and LinkedIn, and the AI breakthroughs coming out of our worldwide network of research labs. Microsoft also provides a variety of tools for developers and data scientists to build AI-infused applications.
Market Size and Competition
AI is delivering real-life benefits, and many consider it ready to transform software and services. As a result, interest is high. Tractica predicts that revenue generated from the direct and indirect application of AI software will grow from $1.4 billion in 2016 to nearly $60 billion by 2025. A Cowen and Company financial analyst report found that 81% of IT leaders are currently investing in or planning to invest in AI. Many of whom are looking to use AI to improve their existing products and services.
Also contributing to the growth, the number of AI startups and investments has exploded over the past five years, reaching $15 billion. According to Accenture, the total number of AI startups has increased 20-fold since 2011. In 2016, over 550 AI startups raised $5 billion in funding and CB Insights reports that over 200 private AI companies have been acquired since 2012.
Tech giants like Amazon, Apple, Facebook, Google, IBM, and others already use AI technology as part of their stack. They are also aggressively competing for position; poaching talent, setting up research labs, and buying startups. McKinsey estimates the tech giants spent $20-$30 billion globally on AI in 2016, 90% of which went to R&D and 10% to acquisitions. All these factors are creating a sense of urgency, as we all attempt to best each other with AI-infused products and services. Given the similarities in products, services, and other offerings, Microsoft’s primary competitors in the AI market are Amazon, Google, IBM, and Facebook.
Microsoft’s Approach to AI
Microsoft has a unique approach to AI that is based on three pillars:
- Leading innovation that extends capabilities
- Building powerful platforms that make innovation faster and more accessible
- Developing a trusted approach so that AI is developed and deployed in a responsible manner
1. Leading innovation
Microsoft is leading innovation with breakthroughs in fields like image recognition, speech recognition, reading comprehension, and many others. We are using advances in computer vision and image recognition for products such as Seeing AI – an app that helps people who are blind or have low vision do things like recognize currency and get a description of people around them.
Our speech recognition algorithms can recognize the words in a conversation with an error rate of just 5.1%, which is about as well as a person.
We have similar results in machine reading comprehension, which uses AI to read, answer, and even ask questions. This year, we created technology that used AI to read a document and answer questions about as well as a human.
In addition to developing products and services that people can use at work and at home, we’re committed to creating tools that use AI to address societal challenges. We call this effort, AI for Good.
2. Building powerful platforms
To build powerful platforms and make Microsoft AI accessible to everyone, we have created APIs and other tools that developers, customers, and data scientists can use to add intelligence into existing products and services, or to build new ones.
More than 760,000 developers from 60 countries are using Cognitive Services to build apps that do things like recognize gestures, convert speech into text, or identify, caption and moderate images.
In addition, more than 240,000 developers have signed up to use Azure Bot Service to create bots that can interact naturally with customers on websites and in apps.
All these tools run on Microsoft Azure, which now spans 36 regions around the world and provides the backbone for many customers who are developing and running AI-infused products and services.
In addition, we make the Microsoft Cognitive Toolkit — which is used across Microsoft to achieve breakthroughs in artificial intelligence using deep learning – freely available to the public via an open-source license.
3. Developing a trusted approach
Though we are in the early stages of understanding what AI systems will be capable of, developing a trusted approach is important. For now, AI systems are very good at doing certain tasks, like recognizing photos or words, but AI still a long way from being able to do any close to what we see in our favorite science fiction books and movies.
As AI systems get more sophisticated and start to play a larger role in people’s lives, it’s imperative for companies to develop and adopt clear principles that guide the people building, using, and applying AI systems.
Among other things, these principles should ensure that AI systems are fair, reliable and safe, private, and secure, inclusive, transparent, and accountable. To help achieve this, the people designing AI systems should reflect the diversity of the world in which we live.
At Microsoft, we’ve developed an internal advisory committee to help ensure our products adhere to these principles.
Making AI Real
So far, I’ve shared an overview of AI and how Microsoft is leveraging it to enhance the products and services we offer. Now I’d like to talk about how we utilize all our products, tools, services, and readiness materials to help our partners.
We lead by showing how our partners can innovate faster with Microsoft’s flexible AI platform and tools; utilizing Azure’s comprehensive set of flexible AI services, enterprise-grade AI infrastructure that runs AI workloads anywhere at scale, and our modern AI tools designed for developers and data scientists to help create AI solutions easily, with maximum productivity.
For technical audiences, we start by learning as much as possible about our partners’ solutions. For partners that are new to AI, we show them how, with Cognitive Services and Microsoft Bot Framework, they can infuse vision, speech, language, knowledge, search, and/or conversational AI into their solutions; doing so in hours – not days or weeks.
While Cognitive Services and Bot Framework are a great way to get new-to-AI partners started, some partners may require a solution that’s a bit more tailor-made. Cognitive Services also offers customization for several of their services. For partners who need to train and customize a service, without knowing much about various AI and ML algorithms, these custom services are a great option.
In some cases, partners will need solutions that are beyond the scope of our pre-built services. For those cases, Microsoft provides platforms and tools that support all major Data Science languages (Python, R, Node, Java, C++, C#, F#, SQL and more). Equally important, our platforms support many of the popular non-Microsoft deep-learning frameworks like Google’s TensorFlow and Facebook’s Caffe2 as well as Open-Source and commercial software platforms.
It’s also important to note, we have the Azure AI Gallery: Which contains ready to implement AI solution templates created by our growing community of developers and data scientists.
Call to Action
Hopefully, it’s clear that AI represents a huge opportunity for Microsoft as well as our partners. Since Microsoft has been a leader in the AI space for a long time, there’s a lot of content and resources available. Whether you’re new to AI or an AI veteran, there are resources to help you come up to speed or keep your skills sharp.
Educate yourself on what’s possible so you can have AI conversations with your partners. I’d recommend starting with the Microsoft AI Platform whitepaper, the AI Landscape blog post, and the Ethics and Law in Data and Analytics MOOC sites to be very useful.
For business audiences, check out the AI Playbook released by Melissa Mulholland’s team in January. I’d also recommend reading and Gartner’s 2018 MQ for Data Science and Machine Learning.
As you can see, there’s a lot of information available and trust me, there’s a lot more so if you have questions, feel free to reach out.
To close, I’d like to reference this quote from Harry Shum. I believe it perfectly represents the opportunity we have with AI.
AI will transform every business, improve every life and solve some of society’s most fundamental challenges.
Harry Shum, Microsoft EVP AI & Research