Salesforce just announced the cutting of 700 jobs, on top of the 8,000 it slashed last year. Google recently sacked another 1,000 employees, after booting 12,000 in 2023.

Another tech titan, Meta, showed the door to 20,000 last year. And yet, by last autumn, Meta was rehiring.

If even some of the most storied tech companies can’t seem to figure out how to right-size their organizations, what does that mean for the rest of the business world? Especially because the U.S. Chamber of Commerce recently noted there are 2.2 million unfilled positions in corporate America, with the professional and business services sector showing the highest number of job openings.

I’m not in a position to critique Salesforce, Google or Meta. Marc Benioff, Sundar Pachai, and Mark Zuckerberg didn’t ask my opinion. But in my experience, leaders looking to right-size their organizations, or at least make them more efficient, too often make the mistake of confusing headcount with the actual jobs and tasks that need doing—and the individuals who do them.

This is why 77% of organizational transformation efforts fail to achieve their objectives. We need to approach things differently. That’s especially the case now that the adoption of AI by organizations is potentially changing the very nature of many types of work and the people who do it.

As a longtime advisor on organizational change, I regularly speak to leaders about their objectives and needs. Lately, there has been considerable discussion of the role that AI might play in making companies more efficient—whether by automating as many jobs as possible or in helping bridge the skills gap by doing work for which the workers can’t be found.

But I wonder if we’re focusing on the right issues when it comes to using AI to reimagine the workforce yet.

AI AS COPILOT

Research has found that a collaborative combination of AI and human skills produces consistently better work outcomes, as well as an average 40% improvement in productivity.

For organizations, AI can play a role in surfacing the data insights that are essential to fully understanding the supply and demand of skills in an organization. But AI still requires careful, thoughtful human oversight. There are too many examples of unsupervised AI producing outcomes detrimental to business objectives and operations. For example, Cruise, GM’s autonomous carmaker, had to recall its entire fleet a few months ago following accidents in San Francisco. Similarly, in the U.K., the delivery company DPD recently had to shut down an AI-enabled chatbot that began swearing at customers and criticizing the company.

And in Australia, a parliamentary inquiry into the Big Four accounting firms ran aground after relying too heavily on a report from a group of academics using Google Bard. The group told lawmakers that KPMG had colluded with 7-Eleven to cheat workers out of wages and that Deloitte had advised a bank on a scheme to defraud customers. Turns out, both supposed cases were elaborate AI hallucinations.

So, it’s clear we must proceed carefully, if we in the business world are to enlist AI’s help in making important decisions. What’s needed are carefully conducted conversations with AI as the copilot.

I’d like to propose three key attributes for such conversations when using AI to help manage organizational change.

USE CLARITY AND PRECISION

Effective prompts require clear and precise language. AI systems interpret input based on the information provided, and any vagueness can lead to irrelevant or off-target responses.

A Human Resources director might compose an initial prompt with a question like, “What are the best initial steps for managing organizational change?”

But a better prompt might begin more precisely: “What are the steps recommended by McKinsey to begin planning a reorganization of a company with 9,500 people?”

UNDERSTAND CONTEXT AND NUANCE

AI often struggles with subtleties. Critical thinkers can stave off potential misunderstandings by first identifying the type of goal to be achieved.

Be clear. What sort of change do we want to manage, and in what industry? Are we meaning to reduce headcount? Shift geographies? Innovate to create a new product line? Something else?

Once the human specifies the purpose of the project, the prompts are the starting point in a conversation with the AI co-pilot, which becomes a deductive process.

CONSIDER ETHICS

With the growing reach of AI, ethical considerations become ever more important. This is where oversight of a human with a moral compass is particularly crucial.

What if the goal is to assess talent, companywide, to determine candidates for promotions to key roles? A poorly prompted AI search might come up with a short list weighted toward people already in supervisory roles or with prestigious academic credentials. But such criteria might be overly reliant on previous promotion criteria that favored certain types of employees while overlooking deserving ones whose ascent might result in a more diverse, representative workforce.

An ethical human must not only initiate the AI conversation but make sure that the data set under review isn’t already flawed by earlier human bias.

As the journey of workforce transformation unspools in 2024, there’s no doubt AI can play a valuable role. But to make this transformation effective and productive, organizations must adopt a new approach to workforce planning.

Carefully conducted, AI-aided organizational change can rely less on bulk firing or hiring and, instead, be more sensitive to the changing skills, attitudes, and needs of today’s and tomorrow’s workforce.