(Forbes) AI has actually been bubbling under the surface for several decades now.
Through it all, it has been geeky stuff, beloved by computer scientists but misunderstood and even feared by everyone else.
It's only lately, however, that AI has become a hot topic outside computer science circles -- but it's important to keep it in its proper perspective.
That's the word from Hilary Mason, general manager for machine learning at Cloudera, in her keynote at the recent Open FinTech Forum. For AI to truly make its mark, "we have to make it boring," she said. "We have to say AI is not something that we're excited about; AI is just one tool it's just as exciting as your C compiler."
That's because, ideally, AI will be ubiquitous. operating in the background of numerous systems and applications.
"I think we are in fact building this AI-first enterprise this technology will find its way into many fundamental processes of the businesses that we all run," she said.
"So when I say let's make it boring, I actually think that's what makes it more exciting."
There isn't anything magical or mysterious about AI, Mason emphasizes. AI is "computer programs that are built on top of data that improves with the introduction of more data into those systems and feedback loops. We are not talking about some actual recreation of human intelligence there's some kind of you know science fiction-type thing."
While AI has historically been a dry topic within the domain of computer science and now data science, there's no question it is now a top priority for enterprises. That's where it will be made or broken, Mason points out.
"A lot of people think this innovation specifically in machine learning happens only in academia or perhaps in startups," she relates. "They do good work, but academics are generally not focused on work that will help you build production systems that solve your problems, but rather on ideas that are novel on meeting fairly arbitrary benchmarks that will get their papers published. Don't expect them to solve your problem in production in a scalable and repeatable way."
Startups, likewise, "are highly resource constrained, so they don't even have domain expertise, don't have data, and haven't been doing this for a very long time." That leaves the enterprise, she continues – "we have large companies operating complex businesses huge amounts of human and technical expertise," not to mention "huge amounts of data generally created as a side effect of operating those businesses for some time."
Mason has the following words of advice for building AI into the enterprise:
Drink coffee and solicit ideas from across the enterprise. "Step one is create a very broad sweep -- get as many ideas for potential projects as possible," Mason says. "Then go through and validate capabilities." AI ideas can be transferred between departments or brought in from the outside from other industries.
Consider the economics of an AI implementation. "Is there a meaningful change in economics that would enable you to actually implement something at scale?" Mason asked. Capabilities out of reach just a few years ago -- such as deep learning -- are now affordable. "We knew how to use deep learning effectively for a few years before it became very common, and we couldn't afford it because GPUs, cost of storage, compute costs, were too expensive."
Also, AI components are becoming commoditized with open source solutions, which means "you have robust software and infrastructure you can build on, without having to own and create it yourself."
Review your data options. This includes data from inside and outside the company.
