In January, we introduced our 2023 Strategic Agenda based on in-depth conversations with top CSOs in our Outthinker Strategy Network. We will be expanding on one of these trends every week with the intention of supporting your organization’s strategy for the next year and beyond. This is our ninth installment.

Newsletter subscribers will receive early access to each weekly announcement. If you’re not already a subscriber and would like to be, subscribe here. 

______________________________________

By the time Thomas Edison filed his patent for the lightbulb in 1880, most people recognized that electricity was going to change the world. Yet it wasn’t until 1925 that more than 50% of U.S. households and factories had electricity. It took more than 40 years to replace old processes and systems with a new one. What took humans so long to transform around the new technology?

Avi Goldfarb, the Rotman Chair in Artificial Intelligence (AI) and Healthcare and professor of marketing at Toronto’s Rotman School of Management, joined the Outthinkers podcast to share the story and the reason behind it. Before electricity, factories used a steam engine for power. The layout of the factory was determined by the power consumption of the machines—those that required the most power were placed closest to the engine while those that didn’t utilize as much were located farther away.

When factories started transitioning to electrification, factory owners took the obvious route of removing the steam engine and dropping in the electric motor. Workflow remained the same, and factory owners who chose electric power reduced costs by 5-10% per year. But for most factory owners, the trouble and investment of replacing the engine wasn’t worth the incremental savings, so they didn’t bother.

Then, around 1900, people started realizing the real benefit of electrification in factories. By decoupling the power source from the machine, you could reinvent what a factory looked like—for example, by placing workers along a production line. It was then that the adoption of electricity began to accelerate. Goldfarb explains that this became a “system solution”—a total reinvention of the way we do things when a new technology is accepted.

Right now, we are in the “1800s” of AI. Businesses are starting to replace existing processes with automated, more efficient versions. We’ve seen customer service amplified by chatbots, content production expanded by generative AI, appointment scheduling made faster and easier, and predictive analytics better matching customer preferences. In 2020, the IBM Institute for Business Value reported that organizations that adopt AI in at least a pilot phase outperform peers 2X financially and report 5-6 percentage points of direct revenue gain. So far, the outcomes are tangible, but they are not yet transformative.

What will AI’s version of electrification look like?

Goldfarb predicts the most revolutionary transition in AI will be primarily based on the shift from prediction to judgment capabilities. Today’s AI is good at making predictions based on massive quantities of data. These predictions are already helping us process data faster to make better decisions. AI is not yet adept at judging, so humans with superior judgment skills increase in value. This may change the types of people we hire—someone who is skilled at predicting and judging at the same time may not have the same skill at judging when the prediction is already provided.

We’ll also need to get better at quantifying and defining what “success” means for a machine to be able to predict. For example, what is a successful employee for your organization? How successful do you mean? An organization looking for the ultimate superstars might risk hiring a few bad employees on the way. Another organization might care about only hiring very good employees and zero bad ones. We’ll need to clarify what we really want.

The advancement of AI is likely to completely revolutionize the way we work. For a while, we’ll continue to make “point solutions” by ripping and replacing existing tasks with automated ones. Once we’ve gotten used to working with AI, the types of tasks, the places we do them, and the idea of work may change all together. What will AI’s “system solution” look like in the long term?