John Roese: Not only do I think that there is a huge opportunity here for entrepreneurs to take a business process and apply machine learning, but there’s also a fantastic job market that once you do it, the people you sell it to are the IT people who are looking for relevance as they move into the cloud world. These things have a huge material impact on the business and the productivity.
This is the next wave of productivity that we’re going to experience on a global scale. My message was unambiguous on that second bucket. If there is one place where we are not talking loud enough and are not excited about, it is this idea of real value of AI in enterprises. It will spawn startups. It will spawn entrepreneurs. It will spawn re-transformation and re-skilling of the IT work force.
While many people are thinking doom and gloom about artificial intelligence wiping out jobs, this is going to create immense jobs. It’s just that you must have those two skills. You must understand the business process that you’re starting with and what it potentially could be, and you must understand the contemporary technology that will enable it to get there. If you have those two combinations, you have the makings of being an entrepreneur, a new business, or a next-generation IT person.
Sramana Mitra: It’s high-order skills though.
John Roese: They are very high-order skills.
Sramana Mitra: What is your third bucket?
John Roese: The third bucket is the one in which AI will be the most prevalent. It is what we describe as either invisible or embedded artificial intelligence. The first one was incredibly visible. You’re talking to an AI. The second one is you as a consumer may not see it, but you’ll feel it. Businesses will run smoother. Things will happen better. Packages will be delivered in a more timely manner. Customer support will be simpler.
As a practitioner in the IT world, there are huge business opportunities. It’s a little less visible, but at least you can know when it’s there. You’ll know when a company uses machine intelligence to improve their business process. Inevitably, you interact with it in a way that you understand its presence. The third one is, there is a whole host of use cases that can only be solved by not having any human beings involved.
We will actually cede the entire function to an artificial intelligence that is deeply embedded into the system and no amount of human beings trying to solve this problem will solve this problem. We described it as invisible or embedded AI. Some examples of this are things like autonomous steering in an automobile. It’s very machine intelligence-driven.
I made a joke at Dell Technologies World, “It doesn’t matter how many people you add to a car. Adding people will never make a car autonomous.” You got to shift the burden to a machine that can think. Without that, you’ll never be able to achieve the cost effectiveness that you need and you can’t operate at the speed. When we think about that one, we start to see that there are innumerable examples where this idea of deeply embedded AI will create entirely different fundamental building blocks.
A navigation system with a deeply-embedded AI gives you the ability to do autonomous steering and crash avoidance, which is not possible with any other combination. We’re seeing this market where people are taking web cameras and adding machine intelligence to do things like facial recognition and other things. The way I describe it is, the difference between a security camera in your home without AI and one with AI is when somebody comes into your house and you don’t have that advanced system, you’ll get a message that says, “Some random person is in your house.”
If you do have machine intelligence embedded and the camera knows it’s your child, you get a message that your child is home from school. That is an entirely different experience to the human being, but there were no human beings processing that data. It was a function that was almost autonomously implemented inside of that node or that device. It disappeared so deeply into the infrastructure that we forgot it was there.
Most people don’t realize that modern security cameras are already doing this type of work. Most people are not realizing how much machine intelligence is going to be deeply embedded into the automobile industry and the mobility industry. Even in our world, we just announced some new technology which is the ultra high-end business critical storage systems. About 50% of the mission critical workloads in the world run on those types of systems.
The reality is we announced that we were adding machine intelligence to do performance optimizations and data placements. We’re processing tens of billions of events a day to better manage content to speed the system up. What’s interesting is, most customers will never see that. It just happens. Suddenly, that system with that level of machine intelligence runs at an order of magnitude higher performance. It runs at latency levels that are absolutely critical to power the AI engines that power that second category.
From a user’s perspective, there’s no configuration. You don’t manage it. You don’t even know it’s there. In the first one, there’s fairly few of them but they’re very visible. In the second one, there’s a lot of them but they’re a little less visible to mere mortals but very important to the entrepreneurs and the IT people.
The third one is, they’re innumerable because they’re going to land anywhere that there’s a machine that needs to scale beyond human capacity on a particular function but are likely to become entirely invisible. Once they’re deployed, we’re just going to forget about them.