Humans and machines meet in the missing middle

Much has been said about how artificial intelligence (AI) systems can automate some processes to make them more efficient. Far less attention has been paid to AI’s greater power: enabling humans and machines to leverage and amplify each other’s strengths.

It’s about understanding the “and” between human and machine as well as between automation and augmentation.

I truly believe that humans and machines can become symbiotic partners, working together and pushing one another to higher levels of performance. Moreover, companies can reimagine their business processes to take advantage of collaborative teams of humans working alongside machines.

A new book, “Human + Machine,” by Accenture’s chief technology & innovation officer Paul Daugherty and managing director James Wilson, describes this dynamic and diverse space as the “missing middle.” Why? Because almost no one talks about it, and only a small fraction of companies are working to fill it.

The missing middle lies between the business roles commonly associated with humans (leading, empathizing, creating, judging) and machines (transacting, iterating, predicting, adapting).

Here, humans are needed to develop, train and manage various AI applications. In doing so, they enable those systems to function as true collaborative partners. Machines are then enhancing human skills, such as the ability to process and analyze copious amounts of data from a myriad of sources in real time.

In this space, which is still largely unchartered territory, six new roles for humans and machines are emerging: three for humans, three for machines – all of which ultimately benefit people.

The new human roles center around how people can make AI systems work in a way that complements them:

• Trainers teach AI systems how they should perform to become more human-like. Training requires a multitude of roles and jobs. At the simple end of the spectrum, trainers help natural-language processes and language translators make fewer errors. At the complex end, AI algorithms must be trained to mimic human behaviors. Customer service chatbots, for instance, need to be tweaked to detect the complexities and subtleties of human communication.
• Explainers bridge the gap between technologists and business leaders, providing clarity by explaining the inner workings of complex algorithms to nontechnical professionals. These jobs will become even more important as AI systems become increasingly opaque.

Take ZestFinance for example, which helps lenders better predict credit risk and expand financing to borrowers who might not ordinarily qualify. The company enables lenders to analyze thousands of applicant data points and applies cutting-edge AI technology to arrive at a yes-or-no decision. Given the nature of its business, ZestFinance’s customers need to be able to explain the inner workings of the AI system they use to approve loans – and ZestFinance is able to provide that explanation by breaking down the data points considered and what patterns the technology looks for.

• Sustainers then ensure that AI systems are operating as designed — i.e., functioning properly as tools that exist to serve us, making our work and lives easier. By doing so, sustainers will help allay fears of a dystopian future in which robots become sentient and overtake society.

On the machine side, new roles for AI emerge that enable new levels of productivity – augmenting human work. These fall into three categories: amplification, interaction and embodiment.

• In the case of amplification, AI agents give people extraordinary data-driven insights, often using real-time data. It’s like your brain, but better. Drug companies for instance are using amplification to monitor the quality control of pharmaceutical drugs after they’ve been released to the general population.
• Agents of interaction employ advanced interfaces such as voice-driven natural-language processing to facilitate interactions between people or on behalf of people. These AI agents are often designed to have a personality, and they can function at scale – that is, they can assist many people at once. You see them in personal assistant roles and in customer service. IPsoft’s help-desk agent is an example of such an agent that operates in the interaction domain.
• While both amplification and interaction are mostly in the software realm, using interfaces that can, in some scenarios, seem almost invisible, embodiment is in tangible, physical spaces. It’s AI in combination with sensors, motors and actuators that allow robots to share workspace with humans to engage in physically collaborative work.

In these three new types of missing middle interactions companies are not only gaining super-powered employees, but they’re obtaining a whole new way of thinking about the way they run their businesses. Combine this with the three new human roles – trainers, explainers and sustainers which highlight the ways that workers improve the effectiveness of AI – and we start to see the impending shift.

With machines, humans can accomplish more than they can on their own, in the same way that machines are reliant on humans to improve. If done right, it becomes a symbiotic relationship. This is where artificial intelligence becomes “applied intelligence.”