Automated Decisions

CIOs and HR

I’ve written about how AI can become unpredictable, but I also know that the desire to streamline and reduce costs will lure CIOs to AI as a solution. HR professionals have invested a lot of effort into the R part of their namesake and creating an environment of trust; unfortunately, taking the H out of HR will tend to create discomfort where the average employee is concerned.  If enterprise prefers the road less traveled, where HR becomes a function of IT and AI, ]developing a more human-like way of training AI to make humane decisions could make all the difference.

Given our current training methodologies, a switch to AI is not a line of action that I would celebrate. Many companies boast that their best assets are their employees. Why would they risk alienating or even losing their greatest assets by deploying an AI system to interact with them?  What are you, as a business, willing to lose? Customers? Vendors? Innovators? Your leadership?

For some, the questions are not why and what, but how and when–because much of humans’ jobs will eventually be replaced by AI. That’s just the reality. Admitting that reality, how do we go about making an AI good enough to deal directly with humans inside and outside of each enterprise?

Trees Leave the Forest

Firstly, we need to call Timber! on decision trees. It’s bad enough we have to tolerate the canned responses and pseudo-interactions of telephone software robots. We already use digital twins–virtual copies–for most physical assets such as buildings, equipment, etc., so it should be easy enough to see the huge benefit of starting now on digital twin development of companies–e.g., systems, employees, etc.– so that when in 10 or 20 years a company is ready to replace its HR employees with AI, it will actually be ready–not just willing.  Simulations implementing digital twins would let live employees navigate through various virtual scenarios, like wildcard and potential emergency situations, getting crucial training in a no-risk practice environment while simultaneously allowing the AI to train itself via observational learning.  Gradually the AI’s abilities would grow and eventually the AI could play an advisory role to the humans in the simulations, broadening and deepening the AI’s training until fully capable of the live human’s task.  In this way, perhaps a human HR staff could be effectively phased out, with a reliable, humane and human-like AI agent emerging in its place.

The idea is not so far-fetched.  Erik Lin-Greenberg, a Carnegie pre-doctoral fellow at Stanford University’s Center for International Security and Cooperation, introduced the topic of experimental wargames in a recent article.  This is important because if an entire theater of war as well as the intricate details of a single troop or fighter can be duplicated electronically, then a business should be able to do the same using digital twins of its human resource assets and processes.

Practice, practice, practice…

With a good digital twin environment we can introduce new scenarios involving  AI, robotics, and autonomous equipment. Imagine if you could make your mistakes in virtual scenarios instead of in reality. Employees as well as the AI would learn from mistakes without having to suffer the consequences of having actually committed them in the real world. Businesses could run scenarios repeatedly, looking for the right result, without the old fashioned expenses of try, try again.

Of course, practice does not literally make perfect; the digital twin solution would offer a significantly higher success rate, but reality insists that we stay on our toes.  We will always be looking for the next great solutions. 

Disclosure

The views and opinions in this analysis are my own and do not represent positions or opinions of The Analyst Syndicate. Read more on the Disclosure Policy.

Leave a Reply