More companies are using AI to help onboard and train employees. Using AI can personalize training to each individual, but companies need to ensure the AI has the right data to work with and realize certain skills need a human touch, experts say.GETTY IMAGES
How do you teach a new customer service representative the difficult skill of dealing with impatient or upset clients? It’s not as if you can hire an actor to role-play with them.
Actually, now in a way you can: Telecom company Telus is one business using an artificial intelligence system to let call-centre workers practice with simulated customer interactions.
“You can actually go and say [to the AI] I want you to simulate a very angry customer,” says Jaime Tatis, Telus’s chief AI officer. “That dramatically reduces contact centre agent onboarding time and, of course, improves the quality of the agents’ performance.”
That reduced onboarding time has had measurable results: According to Telus, the AI agent trainer has saved more than 80,000 training hours and more than $1-million.
Telus is not the only company using AI-powered techniques to help workers learn new skills. Simulating human interaction is just one approach as AI transforms onboarding and training. But observers say that while AI tools offer major benefits, they also come with risks companies must manage.
The power of personalization
A major strength of AI training systems is their ability to draw on large amounts of data to tailor training to individual workers.
“Obviously everyone learns a little differently,” says Steph Daudlin, chief executive and founder of Octopus HR. “AI can assess someone’s learning style and make training recommendations based off of that.”
Some systems can even review an employee’s education and work history to identify existing skills and potential gaps.
Smarter learning
As the Telus example shows, AI models can improve efficiency – especially in human resources departments that oversee employee education, says Anil Verma, a professor emeritus specializing in organizational behaviour and human resource management at University of Toronto’s Rotman School of Management.
“The drudgery can be taken out of those jobs and the humans in the HR department can do more value-added work,” he says.
Making it easy to ask questions
AI may have the biggest impact by making it easier for employees to get answers to simple questions such as “How do I request a sick day?” or “How do I change my direct deposit information?”
New employees can be hesitant to ask colleagues too many questions, says Norman Valdez, emerging technology fellow at the Community Foundations of Canada and CEO and co-founder of BrainTrainr, a Toronto nonprofit that designs AI-literacy programs for nonprofits and small-to-medium enterprises.
Having the option to ask an AI assistant that is available 24/7 “reduces the social friction of ‘one more question,’ ” he says.
A dizzying array of choices
“Everyone and their mother is developing some kind of AI training tool right now,” Ms. Daudlin says. “There’s definitely a lot of overwhelm and stress.”
Before shopping for tools, she advises companies to clearly define their training needs: “Do you want to create a system that people can interface with [themselves]? Are you looking to train them on specific [tasks]? Get really clear on what you want to achieve and then do your research.”
Data is key
An AI system will only be effective if it has data specific to your organization, says Ms. Daudlin. If it’s pulling from a general knowledge base, employees could be misled.
“Every organization should have some kind of documented knowledge base,” she says. If you lack sufficient internal data, she suggests you start documenting now.
Companies must also watch for tools trained on biased data, says Prof. Verma. Some AI facial-recognition systems, for example, have been found to identify white faces more accurately than Black faces. “If your data is not reflective of all different groups and diversity, then the tools will not reflect that,” he says. “It’s very important to vet those tools before you adopt them.”
Know where your data is going
Beyond data quality, companies must understand how providers use organizational and employee information, says Mr. Valdez.
“One of the things that’s important to know is whether the data is being used by the service provider to train their own systems,” he says. If so, see if you can opt out.
Mr. Valdez says it also matters where data is stored. Many data centres are in the U.S., which may operate under different privacy rules than Canada.
Don’t forget the human element
Prof. Verma notes that some essentials – such as organizational culture – simply cannot be taught by AI. “People need to meet and converse and interact for them to know.”
Human oversight also remains critical. “When you leave AI to perform 100 per cent of an output, it will make mistakes,” Mr. Valdez says.
Also, don’t be afraid to seek help and advice. “Reach out to people,” says Ms. Daudlin. “There are literally consultants right now [who] are focused specifically on AI and helping people find the right AI tool and do AI implementations.”