To gain from AI, organizations must reward people using it (and promise not to downsize them)

Interested in more work-life content? Check out our weekly Careers newsletter. Sent out every Sunday afternoon.

As an academic studying innovation, Ethan Mollick was well placed when ChatGPT came out to put it and competing large language models through their paces, evaluating the potential of artificial intelligence for organizations. He labels it a General-Purpose Technology and notes it has grown faster than any other product in history – including the internet – and will have a greater impact than its predecessors.

“We have invented technologies, from axes to helicopters, that boost our physical capabilities, and others, like spreadsheets, that automate complex tasks; but we have never built a generally applicable technology that can boost our intelligence. Now humans have access to a tool that can emulate how we think and write, acting as a co-intelligence to improve (or replace) our work,” he writes in his new book, Co-Intelligence.

Steam power, he notes, which ushered in the Industrial Revolution, improved productivity by 18 to 22 per cent when put into a factory. Researchers have had trouble finding any actual long-term productivity impact from computers and the internet over the past 20 years, despite the deep transformation that occurred. Yet early studies of AI across a wide variety of job types have found productivity gains of 20 to 80 per cent.

Managers must grapple with how to use AI effectively, getting their own work done, and how to take advantage of it organizationally. His advice starts with these principles:

• Always invite AI to the table: Try having AI help you in everything you do, barring ethical or legal barriers. You need to explore the invisible wall that marks the frontier open to us with AI, so that you know its advantages and weaknesses for your particular work. AI cannot only assist us in our job tasks but also in our thinking, giving us another perspective that counters biases. It is a thinking companion, although thoughtlessly handing decision-making over to AI could erode our judgment. Remember the frontier will continue to shift, as the technology advances.
• Be the human in the loop: For now, AI works best with human help and you want to be that helpful human. As AI gets more capable and requires less human help, you still want to be that human, providing judgment and expertise. AI, contrary to the false impression it gives, doesn’t really know anything. It’s simply predicting the next word in a sequence and can’t tell what is true and what is not. It is also prone to providing an answer you will like, even if that is not accurate. You must provide perspective and oversight.
• Treat AI like a person: There is an advantage to changing your notion of AI as a human-built machine to a person, albeit an alien. In many ways AI is like a team of interns, eager to please. “They can be creative, witty and persuasive, but they also can be evasive and make up plausible, but wrong information when pressed to give an answer,” he writes. To make the most of this relationship, you must establish a clear and specific persona, defining who the AI is as well as the problem to tackle. Perhaps for the task at hand it should be Shakespeare or Ernest Hemmingway or Bill Gates or a friendly critic of the writing you submit or a nasty critic. That changes the response you receive and can be illuminating.

Some prominent organizations initially banned AI because of legal concerns, leading to subterfuge as employees secretly used it on their phones, fearing they would be discovered. Ironically, as Prof. Mollick points out, some of the value of AI comes from people not knowing you are using it. Research shows that if individuals learn they are receiving AI-created content they judge it differently.

Augmenting this quiet, shadow use of AI is the justified worry that workers might be training their own replacements. “If someone has figured out how to automate 90 per cent of a particular job, and they tell their boss, will the company fire 90 per cent of their co-workers? Better not to speak up,” says Prof. Mollick.

Organizations that want to take advantage of AI will have to reassure their work force about employment. “There are many reasons for companies not to turn efficiency gains into head count reduction or cost reduction. Companies that figure out how to use their newly productive work force should be able to dominate any company that tries to keep their post-AI output the same as their pre-AI output, just with fewer people,” he argues.

Typical organizational responses to technology are too slow and too centralized, he warns. The IT department can’t easily build an in-house AI model, and consultants and system integrators have no special knowledge about how to make AI work for a specific company.

Given the job fears and the bureaucratic obstacles, organizations will have to provide strong incentives for AI users to come forward and expand the number of people it. These should be substantial rewards for people finding significant opportunities for AI to help. “Think cash prizes that cover a year’s salary,” he advises.

Systems and organizational charts will need to change. He notes that the common hierarchical system we employ was originally conceived to run railroads in the 1850s. Technology and algorithms can be used to control work, keeping tabs on people, and large language models could supercharge the possibilities. Instead, he hopes we can have an honest conversation: Much of work is boring and it is possible now to improve the human experience of work.

We’re writing that story now. The ending remains unknown.

Cannonballs

• Andrew McAfee, co-director of the MIT Initiative on the Digital Economy, tells Charterworks that as with electrification, we’re currently just refining tasks as we bring in AI – the equivalent of replacing the steam engine in the basement with an electric one. The true advantages came when electric engines were put in place throughout individual organizations and the systems that connected them. “The era of generative AI is a year-and-a-half old now. It’s really early to expect that deeper reimagining and restructuring to be in full swing yet, but it’s going to happen relatively quickly,” he says.
• Let your favourite employee move to another team when that’s desired by the individual. Human resources professor JR Keller and consultant Kathryn Dlugos note their research shows people know which managers hoard talent and if you want to be known as a talent magnet let your people go.
• Whining is not a display of vulnerability, warns executive coach Dan Rockwell.

Harvey Schachter is a Kingston-based writer specializing in management issues. He, along with Sheelagh Whittaker, former CEO of both EDS Canada and Cancom, are the authors of When Harvey Didn’t Meet Sheelagh: Emails on Leadership.