From artificial intelligence to augmented intelligence

How is augmented intelligence disrupting traditional notions of AI and how businesses can benefit from it?

When the disruptors become the disrupted, you know the second wave of technology has come - and that's just what is happening when it comes to augmented intelligence. This second wave of AI, which has been described as "a human-centred partnership model of people and AI working together to enhance cognitive performance", according to CMS wire, is changing the way organisations interact with AI and is so powerful that Gartner believes it will create as much as $2.9 trillion of business value and 6.2 billion hours of worker productivity globally by 2021.

While it might sound like mere semantics, augmented intelligence is about the seamless blending of human knowledge and skills with artificial intelligence to solve problems. And while much of the focus on AI has traditionally been rooted in the potential to replace human beings, augmented intelligence is about blending the best of what the human mind and insight has to offer with the benefits of AI.

Some see this as nothing new, really. Chris Nicholson, CEO of Skymind, the company behind Deeplearning4j, and the recently launched Pathmind, a cloud-based AI technology that applies reinforcement learning and simulations to real-world business problems, says it's different than AI only insofar as AI can be autonomous, sometimes, while the purpose of augmented intelligence is to pair AI closely with humans.

In essence it is about how AI is used. Anthony Scriffignano, SVP and Chief Data Scientist at Dun & Bradstreet, explains that, in this way, experts can still apply their thinking, intent and morals, something which is often difficult to represent in machine inference. "The whole idea is that you're still the intelligent one; your intelligence is augmented by something that is able to compute at rates that the human brain simply cannot match and with amounts of information that would otherwise be overwhelming," he says.

Companies are looking to potential applications across a range of areas. Brendan Dykes, Senior Director Product Marketing at Genesys, says: "In the context of customer experience, we see a world where AI is embedded in employee's devices, like mobile and desktop to provide real-time tips and suggestions to employees on the spot, acting like their personal assistant."

Inching towards the singularity?

And for some it is emblematic of a move towards the technological singularity. Andrew Grant, Senior AI Product Director at Imagination Technologies, believes that, while the singularity is still a few steps away, augmented intelligence will take us a step closer to it. "Combining the pervasiveness of artificial intelligence with the benefits of assistive technologies will allow us to achieve more and be more productive," he says.

"He describes augmented intelligence as artificial intelligence personalised, "a series of enabling technologies that will allow us to just get on with what we want to get on with."

Mark Gazit, CEO of ThetaRay, a provider of AI-based Big Data analytics, agrees, also referencing futurist Ray Kurzweil's notion of the singularity as the point at which machines will not only be smarter than humans but will be able to improve themselves. 

"It's a symbiotic relationship that not only amplifies human intelligence, but also helps AI to make better decisions, based on the sensitivities, values and common sense of human beings," he says. "Humans will benefit by being able to use AI to surpass the limits of our own intelligence."

Dan Somers, CEO of Warwick Analytics, believes that augmented intelligence will become the norm. "Eventually we believe that this will involve into ‘business process optimisation' as it becomes more and more normalised rather than new and shiny," he says.

New ways of thinking about AI

But, he adds, we are not really there yet. "The reality though is that even the most sophisticated AI applications ironically require armies of data scientists to develop and maintain them. For many, the Holy Grail in Augmented Intelligence is an application that is trained and guided by a non-data scientist, in particular so that the front-line personnel are not directly doing all the tasks, but they are guiding the bots which help them."

And it will require changing the approach to AI use and development. Peter van der Putten, Assistant Professor of Machine Learning and Creative Research at Leiden University and Global Director of AI & Decisioning Solutions at Pegasystems, explains that if man and AI are supposed to work together, there will be a lot more demands and requirements for the AI to explain itself to the user, in terms of how it comes to its predictions, recommendations and decisions.

Van der Putten adds that there will be more of a need to give humans control of machine learning based AI, to keep the models under control, arbitrate predictions and introduce policies and regulations to implement ethical rules, business strategy or background knowledge.

"The old joke goes that CEOs only invest in software to make money, save money, or stay out of jail," says Josh Poduska, Chief Data Scientist at Domino Data Lab. He explains that, with the potential of unchecked AI to automate a company into very undesirable situations - self-driving cars hitting pedestrians, automated trading platforms crashing the stock market, automated CRM denying loyalty perks to underrepresented segments of society - CEOs are going to be much more likely to approve AI initiatives that are firmly based in augmentation with human decision makers in order to avoid the reach of litigation lawyers.

There are already moves to pass laws and curtail the unrestricted use of AI. "Augmented AI is the most likely path forward," he says.

Poduska believes that augmentation can bring substantial monetary benefits to businesses. "While any decision that has low value per decision and a very high frequency of occurrence begs for automated AI, there are many more business decisions that have slightly higher value, or risk, and occur at a rate that humans can handle. These are perfect for augmented AI," he says.

He notes that humans working with machines usually make smarter decisions than machines alone and adoption rates of AI technology are much higher with augmented approaches. "This is not only true because these solutions are less risky, it is also the case that project planning and product design with the end user in mind greatly increases the chance of success and the impact of the AI asset," Poduska adds.

Augmented applications

Nicholson agrees, explaining that, for example, white-collar office workers spend a lot of time in front of a screen, processing documents and making decisions about information. Augmented intelligence can help them do that. "It can automatically process some information and provide decision support so that workers are more efficient in how they handle documents."

The use of such augmented intelligence is of particular interest in certain areas of business. Genesys, for example, is examining how augmented intelligence can be used to better customer experience. Dykes believes that augmented AI will fundamentally change the contact centre experience as it:

  1. Enables consumers to have the choice to self-serve or talk to a human: AI transforms the IVR and chat experience with smarter bots that can either fulfil requests and answer questions or connect people to the right agent when needed
  2. Limits wait time: by automating mundane tasks with bots (zero wait), optimising agent matching with AI or else infuse AI in forecasting and agent scheduling, consumers are better off with more timely answers to their queries
  3. Results in more knowledgeable customer service reps that already have the context for a call or a chat routed to them: agents and inside sales are better when backed by AI (real time assistance and relevant background info)

It is also being deployed as an approach in the healthcare industry. Jon Payne, Manager of Sales Engineering at InterSystems says that their healthcare product uses augmented intelligence to let users look at different patient records and see whether they refer to the same patient. "Sometimes, the answer is clear, and other times the matching algorithms aren't, so the employee has to manually cross-reference to determine which patient it refers to," he explains. "By applying augmented intelligence that observes decisions and uses them as training data to improve the model that is doing the matching, we've seen an 85% reduction in the amount of work employees do in this area."

But augmented intelligence can be used across organisations, too. Scriffignano cites the example of an organisation dealing with massive amounts of change in the context of crisis intervention. In that context, AI may be helpful in synthesising information, even in recommending strategies, but since the situation is unprecedented, human experts may prefer an approach that allows them to stay in the loop.

"An organisation can then use machines to point out what is different, to interpret massive amounts of information while letting the human experts take actions. The augmented AI methods can learn further from watching the degree to which their advice is taken, thereby converging on advice that is more likely to be taken in the future," he says.

To make best use of augmented intelligence requires a clear and cohesive big data and AI strategy that is supplemented by clear definitions of AI and augmented intelligence. Scriffignano says that organisations should not aim to totally replace their AI strategy with augmented intelligence. "But by having a clear understanding of the function of various AI approaches, they are able to streamline their actions to maximise organisational benefit.

"For example, consider a situation where an organisation must deal with massive amounts of change in the context of crisis intervention. AI may be helpful in synthesising information, even in recommending strategies, but since the situation is unprecedented, human experts may prefer an approach that allows them to stay in the loop. An organisation can then use machines to point out what is different, to interpret massive amounts of information while letting the human experts take actions. The augmented AI methods can learn further from watching the degree to which their advice is taken, thereby converging on advice that is more likely to be taken in the future."

For van der Putten, businesses will need to move away from product, channel or silo thinking and really focus on the customer in order to make best use of augmented intelligence.

Grant cautions that it is still difficult for businesses to embrace even basic first-generation artificial intelligence, today, because there are barriers in terms of getting access to data and having the skillsets available to do something with it. He points out that there are significant challenges in getting small businesses to onboard what seems like complexity.

However, Grant believes the next generation augmented intelligence will be different. "As the efficiency benefits will be quite apparent, it's more likely to be embraced. Though the technology itself may almost be invisible and taken for granted, it will not pose a threat because it's helpful in a myriad of ways to productivity and better decision-support. Just like we take for granted all the benefits of the internet, mobile and Wi-Fi, augmented intelligence will become the acceptable face of the human-machine interface."