Nearly four decades ago, when the personal computer boom was in full swing, a phenomenon known as the “productivity paradox” emerged.
It referred to how, despite companies’ huge investments in new technology, there was little evidence of a corresponding gain in workers’ efficiency.
Today, the same paradox is reappearing — this time with generative artificial intelligence (GenAI). According to recent research from McKinsey & Company, nearly eight in ten companies have reported using generative AI, yet most have seen no significant bottom-line impact.
AI technology has been racing ahead — driven by tools like ChatGPT and a high-stakes arms race among tech giants and startups — fueling expectations that everything from back-office accounting to customer service will be revolutionized. However, the payoff for businesses outside the tech sector is lagging, hampered by reliability issues and an irritating tendency for chatbots to “make stuff up.”
Businesses will likely continue to invest billions to stay competitive, but the economywide payoff could take years to materialize as companies gradually learn what truly works.
🧩 The GenAI Paradox
McKinsey refers to this emerging phenomenon as the “GenAI paradox.” Investments in generative AI are expected to soar 94% this year to $61.9 billion, according to IDC. Yet, paradoxically, 42% of companies have abandoned most of their AI pilot projects — up from 17% the previous year, based on a survey of over 1,000 managers by S&P Global.
These failures stem not only from technical hurdles but also from human factors — employee resistance, customer hesitation, and lack of skills. As Alexander Johnston of S&P Global notes, “human factors” are often the deciding challenge.
📉 The Trough of Disillusionment
According to Gartner, AI is now sliding toward what it calls the “trough of disillusionment.” This phase, characterized by disappointment after early hype, is expected to reach its low point next year before stabilizing as AI becomes a proven productivity tool.
This pattern mirrors earlier technology cycles — from personal computers to the internet — where initial exuberance was followed by hard lessons and eventual transformation.
💼 The Current Winners
So far, the primary beneficiaries have been the suppliers of AI technology — Microsoft, Amazon, and Google — alongside Nvidia, which dominates the AI chip market. Executives at these companies claim AI is already reshaping their own operations, cutting out entry-level coding roles and increasing overall efficiency.
Some predict that AI will replace large portions of white-collar jobs. At the Aspen Ideas Festival, Ford CEO Jim Farley stated, “Artificial intelligence is going to replace literally half of all white-collar workers in the U.S.”
However, as Andrew McAfee of MIT’s Initiative on the Digital Economy explains, “The raw technological horsepower is terrific, but it’s not going to determine how quickly AI transforms the economy.” Real transformation depends on how businesses implement and adapt to the technology.
🏢 Real-World Applications
Some companies are already finding ways to incorporate AI — though results vary widely.
USAA, serving military members and their families, launched an AI assistant to support its 16,000 customer service agents. After several pilot projects (some abandoned), the company introduced the assistant to help workers provide faster, more accurate answers. While financial ROI is not yet clear, employee feedback has been “overwhelmingly positive.”
Johnson Controls built an AI app for field technicians, training it on service manuals to suggest repairs and generate summaries. Testing showed it could shave 10–15 minutes off repair calls — a small but meaningful efficiency gain. Currently, fewer than 2,000 of its 25,000 technicians use the tool, but expansion is planned.
Chief Digital Officer Vijay Sankaran envisions a broader impact in five years, as AI becomes embedded across sales, procurement, manufacturing, customer service, and finance — “That’s the game changer,” he said.
🏦 JPMorgan Chase: Scaling AI Internally
Two years ago, JPMorgan Chase restricted access to ChatGPT due to security concerns. Now, around 200,000 employees have access to an internal AI assistant that retrieves data, writes reports, and answers business queries securely using multiple AI tools.
Half of users report saving up to four hours weekly on routine tasks. The bank’s wealth advisers also use a specialized assistant to access research and market insights faster, nearly doubling productivity and sales efficiency.
Lori Beer, JPMorgan’s Global CIO, oversees a 60,000-person technology workforce and admits to shutting down “probably hundreds” of AI projects — but emphasizes this as a learning process. “We’re not afraid to shut things down,” she said. “It’s not a bad thing — it’s a smart thing.”
💡 Closing Thought
As with every technological revolution, generative AI’s journey from hype to productivity will take time. The GenAI paradox reminds us that adoption alone doesn’t guarantee impact — meaningful transformation requires cultural readiness, human adaptability, and time to mature.
