You’re Asking The Wrong AI Questions. Start Here Instead. 

AI is breaking into every corner of business, but most leaders are making the same mistake: treating it like a technology upgrade instead of a massive structural shift. Using AI isn’t just about speed or automation. It’s about rethinking where value comes from, and how work gets done.

AI is a radical way to reimagine growth, all by rebalancing labor, capital and energy. With fewer workers available and capital investments under tighter scrutiny, AI can unlock new sources of productivity, transforming rigid cost centers into fluid, digitally driven operations that enable businesses to do more with less.

Consider Procter & Gamble’s approach. At one of its Berlin facilities, P&G integrated a system of AI-powered sensors into its production lines, allowing quality to be monitored continuously instead of in batches. This not only improves the output and reduces waste, but it also frees employees to carry out less repetitive, higher-value labor. AI converted a cost center into a profit engine by freeing labor and resources to do more productive work.

We’ve seen this movie before.

AI is already a competitive necessity, and it will become more so every day. It’s very much like the rise of the internet. In the mid-1990s, many companies made tentative gestures to adapt, such as putting sales materials onto (diabolically ugly) websites. By the turn of the millennium, however, they were rushing pell-mell to set up new internet-powered businesses and operating models.

Why? Because the competition was forcing their hands.

Of course, much money was wasted during that rush, and many new entrants still succeeded in disrupting the slower-moving old guard.

It’s far healthier to have a multiyear, phased transition to new models, which is why the time to plan your AI transition is right now.

So, what can AI do for you and your team in the near future? Don’t start with the technology; start with the value. Ask three questions about the way you and your people spend your days:

What won’t humans do?

In this category, consider low‑value, ignored or impossible‑to‑scale work. AI can’t always perform miracles, but it’s often better than nothing.

For example, AI-based HR screening systems have flaws and the potential for systematized bias, yet when used responsibly and with human oversight, they can help companies expand the variety and quality of candidates they screen, updating hiring processes that may not have changed in decades.

Unilever, for example, uses AI to analyze videos of job applicants’ interviews and game-based assessments, reducing hiring times by 75 percent while improving candidate quality.

What shouldn’t humans do?

This could include rote, error‑prone or privacy‑sensitive tasks. AI will become acceptable at a broad number of tasks in the next two to three years, just as it has already become excellent at roles like fielding routine customer service queries online.

For example, project and middle managers fritter away hours of their days in alignment meetings. Today, Zoom, Microsoft and Otter provide digital assistants that can synthesize meeting takeaways, outline agreed-upon next steps and highlight unresolved questions.

Previously, a team member would need to handle these tasks, and—let’s be honest—most teams did without them. While digital assistants’ outputs are currently far from perfect, they provide at least some value for a fraction of the time and cost.

What can’t humans do?

Humans cannot perform continuous pattern recognition at a superhuman scale. As a practical matter, AI’s ability to retain and iteratively process data is limited more by economics than by technology. Give it enough computing power, and it will outperform the physical limits of human cognition in unexpected and nonintuitive ways.

For example, digital pathology teams use high-powered slide scanners and AI algorithms to both detect potentially cancerous anomalies in tissue samples and track millions of images so they can be available for remote viewing.

Highly trained pathologists are still needed to review outputs and make judgments on potential cases, but they now do so from the comfort of their home offices. Digital pathology significantly reduces the talent and infrastructure bottlenecks for healthcare providers, allowing smaller teams to provide better care for more patients.

Don’t wait for perfection. Plan for progress.

It’s both exciting and scary to consider how artificial superintelligence will change our society and the way we work. That technology will come at some point, but your organization should prioritize a different benchmark.

Ask yourself, in the next two years:

• Where will AI-enabled systems be obviously faster, better, and cheaper than your existing processes?
• What can be automated or augmented for 20 percent of the current labor-intensive cost?
• What issues can AI handle that you don’t have the capacity for?

AI doesn’t need to be perfect to be useful. It only needs to be better than what you’re doing now. At the same time, it’s developing at such a pace that we need to skate to where the puck is going, so that we have time to get there.