- 75% of cross‑functional teams fail at core objectives, such as staying on schedule and meeting customer expectations
- Human-AI collaboration can help reduce the burden of administrative work, freeing up managers to stay on task
- Academic research suggests AI tools will eventually become more like teammates that provide emotional bonds for their human co‑workers
In late 2017, Bridgewater Associates founder Ray Dalio announced that his company was developing a software “coach” that will guide managers by crunching data on employees.
The coach is an AI‑powered update to an app that employees at the world’s largest hedge fund already use to rate their colleagues 15 times a week across 75 attributes. Algorithms cull the data to create pointillist feedback maps of skills, strengths and weaknesses. Managers use the maps to build teams in which workers with the right skills take on the right jobs.
“Knowing what people are like also allows us to decide what responsibilities to give them and to weigh our decisions based on people’s merits,” Dalio said in a TED Talk last year.
While many companies are focused on how AI and other machine‑learning tools will work alongside people, there’s a problem that has to be tackled first: People need to work alongside one another in teams to achieve business goals. As it turns out, they’re not always very good at it.
Three‑fourths of cross‑functional teams are dysfunctional, according to a study by Stanford University management professor Behnam Tabrizi, who found that most teams fail at three or more (out of five) core responsibilities: planning a budget, staying on schedule, adhering to specifications, meeting customer expectations, and keeping aligned with company goals.
“Most companies aren’t ready for AI because they still haven’t figured out how to work together as people,” says Dominic Price, the R&D chief at Atlassian, a Sydney‑based company that builds collaboration tools for businesses. “If we don’t fix the human teamwork issues, imagine what happens when you add robots to the equation.”
Tabrizi’s findings point to three crucial areas for improving team dynamics: Better management, clearer goals, and better team communication. New AI‑enabled enterprise tools are bolstering progress in each arenas. They hold promise for putting teams—and teamwork—back on track.
Free the managers
Managers spend 54% of their time on routine tasks like scheduling and managing emails, according to an Accenture report. Just 10% of their time goes toward strategic planning, and even less (7%) toward mentoring talent.
One of the major promises of AI is that it will free managers from rote admin work, freeing them to lean into higher‑value tasks like strategic planning and mentoring talent.
A new class of virtual assistants with machine learning capabilities are nudging managers in that direction. There’s Conversica (for managing email), Amy (for scheduling meetings), HyperScience (for automating data entry), Grammarly (for writing and reviewing drafts) and LawGeex (for reviewing contracts). Natural language generation platforms like Quill are even drafting reports for managers.
Today’s AI tools don’t just handle routine work. They can also take on some of the load of managing a team. Stratejos, for instance, is a tool with AI capabilities that acts as a virtual project manager, tracking team performance through collaboration platforms like Jira, Slack and Hipchat. It also lets managers track employee performance on multiple tasks over time to enable targeted mentoring in skills that may need improving.
“One of the core tasks managers measure success by is their ability to predict,” says Avi Goldfarb, a professor of marketing at Rotman University and co‑author of Prediction Machine: The Simple Economics of Artificial Intelligence. He cites HR managers, who predict who’s going to be a good employee, as an example. “With AI, prediction will become a less important part of the job and coaching is going to become more important.”
Some AI‑driven software even enables self‑management by team members. Yva Pulse is a tool that provides team members with their own performance metrics so they can identify areas of weakness they need to address on their own initiative.
“As machines assume more of the prediction process, things that require judgment and social skills will become more valuable in management,” Goldfarb says.
To be successful, teams need to work towards clearly defined, shared goals. AI can’t call the shots. But predictive AI tools can help companies pinpoint where to focus.
“AI can help people agree on what a goal is by predicting outcomes—but prediction itself isn’t decision‑making,” says Goldfarb. “Goal‑setting is a human task.”
Where AI comes into the equation is in crunching data. Today, there’s more data to crunch, and more insights to mine, than ever before. Armed with predictive, data‑driven insights, teams can set targeted goals that will capitalize on business opportunities in the near future and beyond.
Walmart, for example, uses AI capabilities in SAP HANA, a popular relational database management system, to mine data from transactions at its 11,000‑plus stores in real time. The system churns through over 200 streams of internal and external data, taking minutes to crunch data that once took weeks to process.
Which social media ads will resonate with key demographics? Which new products will fly off shelves in which regions? A company’s internal teams—from marketing and HR to sales and logistics—can use these insights to make complex business decisions to meet (and continually adjust) their goals.
Clear goals aren’t that useful if team members aren’t on the same page about what they’re trying to achieve. Goldfarb predicts that AI will soon help detect subtle variations in individual perceptions of shared goals.
“Let’s say two people are working on the same idea,” he says. “A machine can detect whether they’re working towards the same objective. You could do this through natural language processing or some as‑yet‑invented tool for determining whether two different paragraphs by two different writers are conveying the same idea.”
Nascent AI technology is already moving in this direction with smart speech‑recognition tools like Gridspace. Practically speaking, it’s a virtual note‑taker for business meetings. But beyond taking notes and sharing them with stakeholders, Gridspace detects which parts of a discussion are important based on who says what (it recognizes vocal “fingerprints”), context and even inflection. It then highlights all these points in an email rundown.
Participants can edit Gridspace notes to flag key priorities and share them with colleagues. This feedback loop also helps Gridspace learn how to better recognize company priorities in future meetings.
Keep it clear
In the end, a team lives or dies by its ability to communicate. But even if everyone on your team is yakking on Slack, Hipchat or Jira, they aren’t necessarily communicating effectively.
Several new applications based on machine‑learning algorithms are geared toward ensuring that they are. Virtual assistant Talla, chatbot Howdy and voice assistant MindMeld can all be integrated into Slack and other platforms to let teams tap a company’s internal data and share knowledge in real time.
Even with these artificially intelligent bells and whistles, modern collaboration platforms can actually hinder effective team communication. As Basecamp founder and CEO Jason Fried says, “group chat is like being in an all‑day meeting with random participants and no agenda.”
The mass adoption of new digital communication tools may, in fact, be contributing more to team failure than success. Price recently asked Atlassian’s engineers what makes teams great.
“We got this Tourette’s‑like response where they all just started shouting, ‘Tools! Tools, tools, and more tools,’” he says. “You have to take a step back from being highly reliant on tools to deliver great outcomes and ask, how can we empower these people to alter the way they work.”
Bridgewater is not alone in its expectation that AI will solve that riddle for teams and organizations before they ever figure it out themselves. Some emerging AI applications are geared toward strengthening emotional bonds between team members. Crystal, for one, is an AI algorithm that can identify basic personality traits of colleagues or contacts by sorting through public social media data. It then offers targeted advice on how best to communicate with them.
Bonding with AI
After thousands of years, humans still haven’t mastered the art of collaboration. What new challenges can we expect when AI transitions from work tool to full‑fledged work partner?
It’s not a hypothetical question. By 2022, one in five employees will have AI as a co‑worker in the form of a chatbot, a voice‑ and vision‑enabled AI assistant or even a robot, according to a Gartner study.
Some experts say this shift will impact team dynamics favorably, even predicting that AI will provide emotional benefits to human workers.
“We have an opportunity to create new kinds of interactions,” says Guy Hoffman, a Cornell researcher who designs AI systems and robots. “Any time we have new tech, we get new relationships.”
Hoffman conducted a study in which humans worked with robot teammates who were programmed to behave like humans. As expected, he found that human/machine collaboration increased productivity. He was surprised, however, to learn that his human subjects formed an emotional bond with the machines.
At the end of the study, the human participants reported feeling “tenderness,” “amusement,” “respect,” and “trust” for their new teammates. They also used human pronouns to describe their robotic co‑workers.
On its face, the takeaway is ironic but also optimistic. It may be that artificial intelligence is the missing ingredient that can help human teams succeed.