Dennis Deery

Since the creation of the modern company, we have become used to identifying workers as individual contributors or as managers of people. Individual contributors needed only follow the orders of a leader, who would ensure that all workers contributed to the same, hopefully effective, goal. While factories tried to deconstruct work activities into simple, easily-repeatable tasks that left no room for individual variations and ideas. It’s fair to say that many managers have tried to accomplish the same breakdown in office work, asking individual contributors to simply do their tasks and not be overly involved in thinking about the overall goal.

This is, ideally, one of the current ways of doing business that will be overtaken and eliminated by the development of artificial intelligence (AI).

Every information worker needs to be a manager now.

And we haven’t prepared anyone for it. What do employees need to know to effectively use today’s AI tools?

In its current form, AI acts much as a new employee does. It can complete fairly complex tasks, given correct instructions. However, the completed work may be right, or it may be wrong. Like an intern asked to “produce a report”, the report could be one page or twenty. It may include citations, or it could include “facts” made up from whole cloth. It may be nicely formatted, or it could be fifty paragraphs of run-on sentences.

As we teach employees to use AI, we need to teach the skill of delegation. What do clear instructions look like? What level of detail is enough to get the results you want? Where should data come from, and how should it be cited and referenced? Any of us who have repeatedly produced reports for a superior know that it becomes muscle memory because those questions have been asked and answered. However, for a new employee like AI, we need to be explicit with our ask. It’s especially critical to be clear with our direction because AI can quickly generate pages and pages of material when a one-sentence answer is more appropriate. It can quickly go from saving the time of an individual contributor to sucking up the time of the audience for such a report.

We also need to teach individual contributors that delegating, whether to a human co-worker or to an AI, does not transfer accountability. As a manager, you’re responsible for the outputs your team produces. It’s great to trust your direct reports, but it’s also important to understand that as their leader you may have better information about what’s needed in creating a report, or a piece of code. Responsibility for its suitability for the task falls to you. Again, AI can produce some wonderful work output, but you still need to review it for correctness. Knowing this will help contributors to build the muscle of creating better instruction prompts for their AI. If I let my AI run wild and produce a fifty page report when one page will do, I’m the one that’s going to have to read that report. Taking that accountability can help individual contributors begin to think more about company goals rather than completing tasks. “Does this information give my boss what I need to do the job?” rather than “Whew, I got that report done, the boss can tell me if it’s right.”

The judgement required to satisfy that accountability is the skill that, for now, separates us from the AI and leaves us as the critical cogs in a functioning organization. AI can produce great outputs, but it often does not have the full context around why that output is needed. When you produce a report for your boss, you pull company data, but you also wrap in your expertise of how to best speak to your boss, and their boss, and you pull in the context from dozens or hundreds of meetings and conversations that the AI does not have access to. That context, and the life experiences you bring to bear, are the secret recipe that you can link to AI outputs to make them uniquely yours, and that make you uniquely valuable to your employer. Use AI to leverage those bits of yourself, not to replace them.

Finally, we have to acknowledge that there are more than a few individual contributors out there that are perfectly happy being individual contributors and not managers. Some of them may have even turned down opportunities to be a manager. There are lots of reasons for this - they may enjoy the work they currently do, they may worry that managing will bring additional stress to their job, they may not want the added administrative work that can come from being a manager.

It’s important to recognize that this shift to “managing” could be unwanted for some people, and will be a major change even for people that welcome it. An individual contributor that excels in their role may feel a loss of identity when asked to manage an AI as part of their work. While AI may help to eliminate rote, repetetive work, there are cases where that’s an enjoyable part of the job for some people. As with any change process at work, we need to ensure that we communicate effectively as we implement AI in our organizations. What are the expectations for use by individual contributors? How does this change expectations around volume of work completed? If AI takes away parts of the job that I love, what are my options? These should be open conversations between leaders and employees. It’s okay for leaders to admit that we’re all finding our way through the changes wrought by AI adoption, and we need to extend grace to each other as we adopt new ways of working.