AI doesn’t automatically deliver results; it depends on how well the applications align with a company’s operational model. I frequently emphasize that any emerging technology can either support or hinder an organization, depending on how well it fits.

There are also important uncertainties that need to be addressed in business process engineering when integrating AI.

Lian Parsons from HBS highlights this concern: “When we lack transparency in how predictions are made, it becomes hard to establish accountability or build a defensible framework.” This thought reflects the perspective of Mark Esposito from Harvard DCE Professional & Executive Development. “The key issue we face is determining the role of humans and ensuring the technology remains centered around human needs. Even when we reduce human involvement in repetitive tasks, it doesn’t mean we should eliminate human oversight. If that role isn’t clear, we must deliberately define it.”

So, how do we ensure that alignment?

Taking the Treatment

In a recent TED presentation, Karim Lakhani, a professor at Harvard Business School, offers a different analogy—comparing AI adoption to a patient using a new medication.

“We don’t really understand the proper dosage, its effectiveness, possible side effects, or even what complementary practices should accompany this drug, yet it's already being used by hundreds of millions,” he explains. He adds that his team has been running a series of AI “clinical trials,” similar to pharmaceutical testing, to track adoption and observe real-world outcomes.

What appears from these observations, he explains, is a “jagged technological frontier”—a complex network of systems and procedures evolving at different speeds, each bringing its own set of challenges.

“There are specific areas where AI excels,” Lakhani states. “Using it for those functions results in impressive performance improvements. But when applied to areas it’s not suited for, performance suffers significantly.”

Productivity Uplift Seen in AI "Clinical Tests"

Regarding productivity, he shared findings from a study showing that AI users completed 12.2% more tasks on average, completed them 25% faster, and achieved 40% higher quality.

Still, he warns leaders not to become complacent in their oversight of innovation use cases.

Procter & Gamble’s Intelligent Collaborator

Discussing randomized trial results, Lakhani describes AI evolving from a simple tool into a collaborative partner.

“The first noticeable change is a major jump in productivity,” he observes.

Then, the transformation accelerates.

He adds, “While AI model capabilities are growing exponentially—roughly doubling every six to nine months—the ability of most companies to adopt and integrate these changes grows only linearly. That means every six to nine months, the gap between what models can do and what companies can implement widens exponentially.”

According to Lakhani, most companies can't keep up with the rapid pace of progress, which may lead to competitive disadvantages.

He described the challenge of standing out when many businesses are using similar tools, often for as little as $20 per month.

In other words, you must leverage a low-cost tool to generate significant value for your enterprise.

The Four-Phase Journey

Lakhani also introduced a concept he calls the “four-phase journey,” a strategic guide for leadership navigating AI adoption.

Phase one is to learn everything you can about AI—understand the models, systems, and capabilities, and how they apply to your business landscape.

Phase two involves actively using AI tools. He points out that this is where many leaders go wrong—delegating all AI tasks to others instead of engaging personally.

Phase three is envisioning the future of your business with AI, and phase four is turning that vision into reality.

In essence, Lakhani walks us through the transformation of AI from productivity tools to autonomous agents that begin to mimic human behavior.

Will we soon see AI agents in leadership roles, and what will that future look like?

More to come.

The above is the detailed content of The Right Dosage Of AI For Business. For more information, please follow other related articles on the PHP Chinese website!