© 2019 SourceMedia. All rights reserved.

Companies must start creating human-machine collaborations

Last year, my colleague J. P. Gownder and I got to talking about the impact on employees both from companies’ automation efforts and their eagerness to integrate artificial intelligence and robots. We noticed that most of the predictions of how this would go were either utopias viewed through rose-colored glasses or dystopian nightmares darker than most science-fiction novels. Perfect or perfectly terrible future scenarios both sounded unlikely to us.

More importantly, we believed that companies needed a plan for creating future employee experiences that didn’t leave humans either out of work or with jobs that left little for them to do.

And so we recently published a new report to help companies, “Start Designing The Future Human-Machine Workplace Now.” The goal for every company should be to ensure that their humans can thrive when they work alongside robots and AI.

To do that, we lay out three principles for creating future employee experiences:

Free employees to do more important work

There’s plenty of work that we as humans would prefer not to do. The great tragedy of modern technologies is that they have put the onus on high-salaried, busy knowledge workers to do their own administrative work — from scheduling meetings to expense reports.

robots and humans.jpg
A SoftBank Group Corp. Pepper humanoid robot displays GSMA Ltd. branding on a screen during day two of the Mobile World Congress (MWC) in Barcelona, Spain. Photographer: Angel Garcia/Bloomberg

Why not give it all to machines? We see virtual personal assistants in many employees’ futures, taking on predictable, rote tasks such as meeting scheduling, note-taking, and to-do nagging. Companies can gain enormous productive capacity for their most valuable human assets when they do this.

Adapt AI to humans, not humans to AI

For too long, humans have had to adapt to their machines. Whether we’re hunched over our keyboards, straining our necks to hold a phone in place, or trying to remember an obscure command sequence, humans have had to be the flexible partner in the human-machine relationship. That must change.

One way to make AI serve humans on their terms is to layer AI into existing software or interaction models. We call this “everyday AI.” As an example, Microsoft is now using machine learning to streamline several existing tasks. It helps Outlook users sort emails in their inbox automatically or automatically identifies numbers on receipts to populate expense reports. This augmentation of existing processes is a way to streamline or add value without asking much of the humans.

Make humanness a strength

Machines can be naive and unreasonable. So even as companies push for more collaboration between humans and machines, they must honor what makes us human.

One key human strength is our judgment, which helps avoid AI literalism. For example, if the AI suggests something as part of its natural language processing that doesn’t make sense, the human can overrule the AI. The step to check the machine’s work — so to speak — also builds confidence among other employees that what they’re asked to do makes sense for humans. The system suggests answers for humans to give during live chat sessions, but if the employee overrules the suggested answer, that decision feeds back to the database of answers.

There’s plenty more examples and guidance in the report, including a section about how to keep employees comfortable with AI integration.

And for more on this topic, you can listen to J. P. and I discuss the research on Episode 182 of The CX Cast. Or check out the video below in which I preview my talks at two upcoming Forrester events, where I’ll be speaking more about human-AI collaboration, first at Forrester’s Digital Transformation & Innovation 2019 Forumin Chicago on May 14 and second at CX NYC on June 12. I hope to see some of you there.

(This post originally appeared on the Forrester Research blog, which can be viewed here).

For reprint and licensing requests for this article, click here.