Talent patterns, tacit knowledge and the future of work

Posted by Mike Walsh

8/9/20 10:51 PM

M I K E W A L SH

 

We can be so focused on robots replacing humans, that we miss the essential problem of preserving organizational knowledge. The future of work is more than just automation. One of the most significant opportunities for AI in the enterprise is using it to understand the heuristics, practices, and tacit knowledge exhibited by our best people - before they leave or retire. Especially now. Given the current crisis, finding better ways to define and retain talent patterns has never been more critical.


Every year, a significant number of the most valuable people in your organization will leave. Some will get better offers from rival firms; some will shift careers or find new things to do; others will just retire. In the United States alone, an estimated 10,000 Baby Boomers retire every day, taking all the institutional and human knowledge they have accumulated over a thirty- to forty-year career. This knowledge about how things work and how to get things done is the nuanced, contextual knowledge that doesn't live in spreadsheets, databases, or PDFs but rather inside human brains. I think of it as talent patterns.

The challenging thing about talent patterns is that they tend to be tacit, rather than explicit knowledge. Explicit knowledge is much easier to document and define, and thus preserve. Tacit knowledge falls into the 'we know more than we can tell' problem. Whether it is riding a bike or identifying a cancerous mole - a student or a new employee cannot be simply told how to do it - they have to practice under supervision until they have mastered the skill for themselves. However, as we have discovered with the rise of skin cancer AI diagnosis, machine learning can change the way we think about skills and knowledge.

 



I met Ganesh Padmanabhan some years ago when I was working on my book, 'The Algorithmic Leader'. He is currently the Revenue Officer at Molecula, but at the time, he was a VP at CognitiveScale, an AI startup in the field of augmented intelligence. Ganesh was working on the problem of empowering humans with systems that could learn and improve with time. In this view, this was an important point of differentiation over more basic forms of algorithmic systems, like robotic process automation.

Robotic process automation, which essentially replaces a human operator with an algorithm that replicates their activities on a computer terminal, is a rules-based system. It takes a routine task that a human normally performs, whether it be filing documents or completing forms, and automates it. However, the moment you want to go a bit further and have the system deal with a more complex issue or a cognitive decision-making process, then the automated system hits a wall. Whatever the automated system has been programmed to do, it will continue to do in exactly the same way, whether it has done so ten times, a hundred times, or a million times. It will not and cannot deviate from that.

"The idea behind augmented intelligence is different," Padmanabhan explained to me, as he gestured at the teams of programmers visible through the glass walls of our meeting room. "Augmented intelligence is when you try to mimic human cognitive functions with a feedback loop in it. With a human in the picture, when you're surfacing a particular pattern and say, 'Here's a decision I recommend you make for this particular process,' you're giving them both reasons and the evidence for a recommendation. The human can then use their judgment and either agree or disagree, based on their intuition or experience."

"So, the system learns over time that the output is essentially one that shouldn't be weighted?" I asked.

"Exactly," said Padmanabhan. "The human being in the loop will also ensure the system is trained and gets better as you iterate through it. That is what makes augmented intelligence different from regular AI approaches."

Padmanabhan's vision of the human in the loop becomes even more important when you stop worrying about algorithms taking away human jobs and start imagining how you might preserve the patterns of knowledge that leak from organizations. What makes for great customer experience, a positive interaction in a call center or a caring moment in a medical facility relies on uniquely human patterns that are hard for machines to grasp, or evaluate, without our assistance.

Padmanabhan gave me the example of capturing the patterns of care managers who provide care for cancer patients, so that when a new hire comes in, they are prompted by a recommendation system that says, "These are the right steps to take because Joe Smith did this for thirty years, and he believed it was the best way to actually address this particular function."

You don't have to wait until the end of someone's career to preserve their talent pattern. Rather than simply automating obvious processes, leaders should attempt to identity, record, and replicate the best behavioral patterns across their organization. This becomes particularly important in a hybrid or remote organization - when it can be harder for people to learn in close proximity.

Defining your talent pattern strategy should be a joint project between HR, IT, and business strategy teams - as it incorporates people, AI, and process analysis. Remember, this is not a customer satisfaction survey. Your goal is not to generate a success score, but rather to shift to a different way of working in which doing something, and training an AI to understand what you are doing - becomes inseparable. Think of it as an iterative, data-driven approach that allows you to build a picture of your organization's ideal state.

Only by capturing the best in what people do, can you design a better business.

 

 

This article is excerpted from ‘The Algorithmic Leader: How to be smart when machines are smarter than you’ - available now on Amazon and Audible.

Topics: HR, Talent

New call-to-action

Latest Ideas