If digital workplaces are being disrupted by the ongoing development of artificial intelligence (AI) driven apps, by 2021 those disruptors could end up in their turn being disrupted. The emergence of a new form of AI, or a second wave of AI, known as augmented AI is so significant that Gartner is predicting that by 2021 it will be creating up to $2.9 trillion of business value and 6.2 billion hours of worker productivity globally.
Gartner defines augmented intelligence as a human-centered partnership model of people and AI working together to enhance cognitive performance. This includes learning, decision making and new experiences. “As AI technology evolves, the combined human and AI capabilities that augmented intelligence allows will deliver the greatest benefits to enterprises,” Svetlana Sicular, research vice president at Gartner, said in a statement..
AI Or Augmented AI?
So what is the difference between AI as we know it know, and augmented artificial intelligence? Keiland Cooper is a neuroscientist at the University of California Irvine and co-founding director of the non-profit ContinualAI , which is dedicated to researching pathways towards artificial general intelligence (AGI).
He explains that AGI differs from AI in such a way that if you measure your intelligence on a scale where humans are on the top tier, then an augmented AI algorithm would have the same intelligence as the human. Normal AI, like the kind we see today, isn't as smart; not classified as augmented means it can only use the intelligence they do have on specific tasks. “Sure, we don't stand a chance against a chess program, but that is all the chess program can do, chess,” he said.
“Where humans have the edge is our ability to not only play chess, but walk, talk, cook, work, play as well. This is the general aspect of the AI which the name refers, and has been the Holy Grail of AI research as it would signal that we may be onto something about how human intelligence works.” This would also likely be the point where we could say we're the "smartest thing" we know of, but this is where philosophy would need come in.
This, he said, is why companies such as OpenAI receive so much attention, as for some the stakes would be quite high should technology become so advanced. OpenAI has been chasing the goal of developing AGI for the benefit of all. Founded in late 2015, the San Francisco-based organization aims to “freely collaborate” with other institutions and researchers by making its patents and research open to the public. The founders (notably Elon Musk and Sam Altman) are motivated in part by concerns about existential risk from artificial general intelligence.
Related Article: 7 Ways Artificial Intelligence is Reinventing Human Resources
Augmented AI In Use
Augmented AI is not, however, a single concept and can mean many things, from self-standing AI engines through assisted decision support or knowledge management. We believe that's why the impact is forecasted to be $2.9 trillion — because it will be deployed in a number of ways, from AI and machine learning to help companies understand how to manage their fixed asset, Maryanne Steidinger of Los Angeles-based Webaloo, said.
She pointed out that already AI can help to predict and prevent downtime, critical in those industries that need a constant flow of product (such as electricity or water) to ensure customer safety and satisfaction. For companies like food and beverage, AI could help in decision making in production — for example, which line to use for which product, which line has the likelihood of better performance, less downtime, higher quality.
AI can take multiple dimensions and make those predictions based on algorithms that have been developed for those industries with the behaviors of the plant, equipment and people. “AI [and, for that measure, machine learning] by itself has no meaning — it has to have relevant use cases in order to be valuable,” she said. Already there are products that offer some AI-like characteristics: simulation, virtual reality, paper-on-glass but most enterprises are looking for those use cases that will make them ubiquitous and a "must have" for a specific industry.
“It will be up to the vendors to provide those use cases that will apply to the general public, or a specific segment of the market, with measured payback/ROI for scaling [and numerous use cases],” she added.
Related Article: Why Artificial Intelligence Will Create More Jobs Than it Destroys
AI For Scale, Speed
We are emerging through the first wave of AI adoption, and it has left some scars on the industry, Tom Wilde, CEO of Woburn, Mass-based Indico said. In short, everyone jumped on the hype too quickly — that somehow AI was going to be the magic elixir, to not only learn how to do the things we do, but to also invent new and better ways to do them. Today, that's not possible. Instead, we’ve learned that AI is really good at two things: Scale, and speed. “As broad context, the business processes today that have been difficult to automate have been tasks that require some form of judgment or decisioning,” he said. People are really, really good at these types of tasks, and historically, computers have not been capable in these areas.
“We have tried to simulate judgment in computers by building elaborate rule-based or expert-based decision engines. But they have only been as scalable as the programmers' ability to create and maintain the rules based on their understanding of all of the possible permutations of the task,” he said.
In many ways, this was both the promise and ultimately the cause of some of the more notable AI failures such as the IBM Watson debacle at the Anderson Cancer Center. More than an actual failure of the technology, it was failure driven by an incredibly over-hyped marketing-driven expectation.
Augmentation Vs. Automation
The right expectation for AI is much more one of augmentation vs automation. Where AI is best utilized is in delivering scale and speed. AI delivers scale in that it can learn vast amounts of information and apply those learnings across huge amounts of data. In our world, there are many companies that are trying to make sense of tens of thousands of scanned documents from their archive. Something that is beyond any real human scale or reasonable ROI to do with people. In terms of speed, AI is unmatched. It can be accelerated simply by applying more hardware to the problem
The key point here is that AI is only as good as the training data it’s given. It can never be better than the training data. Therefore, if the training data is incomplete or lacking, expect a higher level of human in the loop. Gartner's prediction feels spot on, he said. The analogy we most often like to use is that rather than the popular image of an army of robots showing up to displace people, the better analogy is knowledge workers who are fitted with AI to enable them to accomplish previously impossible tasks because of a powerful set of augmented capabilities delivered by AI.
AI In Control
All this leads to a place that only exits today in science fiction stories, but which Stephen Farkouh, chief information officer at Little Rock, Ark.-based Windstream Enterprise, believes is on the way. The human partnership aspect of augmented AI is only an angle until such time that AI has progressed to a point that we can trust it to operate autonomously and have figured out, as a society and governing body, how to manage through the disruption.
“It makes people feel better to hear that AI helps versus replaces jobs, and that is true right now while AI solutions are in their infancy, but I have no doubt that the models and algorithms will progress to a point where bots can assemble information as good as and likely better than their human counterparts,” he said.
That said, it will take a long time to develop and harden such technology and even longer to gain the confidence to use it, resources to secure it and regulations to unleash it. At least five years until such time and likely five years beyond it, augmented AI solutions will be the lead AI play and it will be as such in every aspect of the enterprise - crunching and processing data at a rate and scale the brain could never approach.