‘More dramatic than the internet’: Innovation in AI happens at a breakneck pace and spreads instantaneously, but organizational change happens at the human scale

Generative AI as a broad technology invites comparisons with earlier breakthrough technologies. Like electricity, semiconductors, or the advent of computing, AI is fundamentally changing what individuals and businesses are able to achieve.

However, generative AI is among the first general-purpose technologies to exist during a time of immense connectivity. 

The telephone was invented in 1876, and it took 70 years for it to penetrate 50 percent of US households. Pearl Street Station and electric power went live in 1882, and took more than 40 years to diffuse similarly. Radio spread in 11 years; television in 8; the personal computer came 30 years after the invention of the transistor and 15 years after the invention of the microprocessor and then took 20 years to diffuse; the smartphone took 5; generative AI reached usage from half of the US population in under 2 years from the GPT3.5 moment. 

“When electricity was invented, the world was fragmented and slow. Infrastructure buildout was costly and slow, and innovation could not travel from one area to another quickly,” said Columbia Business School Dean Maglaras, addressing Columbia Global Center Mumbai’s inaugural India Hour event. 

Those constraints have largely disappeared in the age of AI, Maglaras argued. In conversation with Pirojsha Godrej (MBA ’08), Chairperson Designate of Godrej Industries Group, Maglaras described how breakthroughs now move from discovery to application with little delay, and how the pace of infrastructure buildout has no parallels to past technological diffusion epochs.

“You could have an innovative idea or model launched by a frontier lab in San Francisco and, 12 hours later, it is being used in Mumbai in a startup. It doesn't take years or decades for innovation to spread. AI capabilities are improving exponentially fast, and our connectivity allows them to diffuse instantaneously all around the world,” Maglaras said.

The humans in the loop

While global connectivity shortens the distance between AI discovery and experimentation, the pace changes once the technology is deployed in an organization. “The speed of the diffusion of this technology in organizations is going to be slower,” Maglaras said. “And it’s going to be slower because it has to do with humans.”

Employees need time to learn the new tools, understand how their work will change, trust the people directing that change, and see themselves in that wave of change. At the same time, their managers must decide which workflows to automate and make more efficient, how to prioritize and pace change inside the organization, which capabilities to build, and where AI can support entirely new products and lines of business.

Companies have spent the past 25 years investing in data, analytics, and digitization, laying some of the foundations for the present wave of change, a transition Maglaras expects to take years, though not the decades required for the diffusion of electrification or computing. 

Driving that change, he said later, is “a deeply human leadership exercise.” Judgment and an understanding of employees’ fears will remain central to the work of a leader.

From efficiency to reinvention

For many firms, the first phase of adoption will likely focus on tasks they already understand: narrow automation or improvements to existing workflows that drive efficiencies. Maglaras sees a larger challenge ahead. Over the next decade, companies will need to decide whether AI changes what they make, which customers they serve, and where they compete. Organizational design will change along the way.

Maglaras described an “intelligent organization” as one that treats efficiency gains as a starting point, then uses AI to reconsider its products, markets, and services. Established companies may have an advantage because they already possess products, capital, market access, and customer relationships. That advantage, however, rests on a willingness to disrupt familiar ways of working before a competitor does.

The transition also raises the value of people skills. Maglaras pointed to critical thinking, judgment, collaboration, and the ability to motivate others and build trust. Leaders must understand what AI tools can do, how to inspire and support employees, and how to address their fears about how work may change. 

Reskilling helps employees become effective users of AI and strengthens their prospects inside and outside the organization.

“We need to be changing what we’re doing: new products, new markets, new services,” Maglaras said.