Almost 40 years ago, when the personal computer boom began to get serious, a phenomenon known as the “productivity paradox” emerged.
This was a reference to how despite the large investments in new technologies by companies, there was little evidence of profits corresponding to workers' efficiency.
Today, the same paradox is manifested, but there is generative artificial intelligence (GEN AI). A recent survey by Mackinsey & Co shows that almost Eight Ten companies report using Gen AI, but many have reported that there is no major impact on the final impact.
AI technology is backed by high-stakes arm races between high-tech giants and ultra-rich startups in chatbots like ChatGpt, spurring hopes that everything from back-office to customer service will revolutionize. However, non-tech sector companies are lagging behind in return, plaguing problems, including exciting trends from chatbots.
This means that while businesses must continue to invest billions to avoid falling behind, it can take several years for technology to deliver the rewards of the entire economy as they gradually understand what's best for them.
As McKinsey did in his research report, we call it “gen ai paradox.” According to technology research firm IDC, companies' investment in Gen AI is expected to increase 94% in 2025 to US$61.9 billion (US$79.5 billion).
However, the percentage of companies abandoning most of their AI pilot projects had risen to 42% by the end of 2024, according to a survey of over 1,000 technology and business managers by data and analytics company S&P Global.
Not only technical hurdles, but often projects failed Also S&P Global Senior Analyst Alexander Johnston said that it is because of “human factors” such as employee and client resistance and lack of skills.
Gartner, a research and advisory firm that represents a technical “hype cycle,” predicts AI is sliding towards a stage called “Valley of Disillusionment.”
Gartner Chief John David Lovelock said that technology is expected to have a low point in 2026 before it becomes a proven productivity tool.
It was a pattern of past technology such as personal computers and the Internet. Early vibrancy, difficult slogans of technology mastery followed by industry and work transformation.
The winners up until now have been suppliers of AI technology and advice. It includes Microsoft, Amazon and Google, which provide AI software, but Nvidia is the runaway leader in AI chips.
Executives at these companies boast about how AI is restructuring their own workforce, eliminating the need for entry-level coding work and making other workers more efficient.
AI ultimately replaces an entire herd of human employees, with many predicting perspectives that are widely accepted and echoed by the mainstream of the company. At the Aspen Idea Festival in June, Ford Motor CEO Jim Farley said:
It depends on whether or not that type of innovative change will occur, and how quickly it is, on the real world testing grounds of many companies.
“We're committed to working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with people who are working with them,” said Andrew McAfee, a leading research scientist and co-director at the Massachusetts Institute of Technology (MIT) initiative on the digital economy.
Still, while some companies are finding ways to incorporate AI, the technology remains a long way to replace workers in most cases.
One company where AI promises and flaws are being deployed is USAA. It provides insurance and banking services to the military and their families.
After several pilot projects, some of them were closed, the company introduced AI assistants to help 16,000 customer service workers provide correct answers to specific questions.
USAA tracks AI investments, but there is no calculation of the financial payoffs for call centre software yet.
However, the company said the response from the workers was overwhelmingly positive. There is a software app that answers customer questions online, but the call centers field an average of 200,000 calls per day.
“These are key moments,” said Ramnik Bajaj, the company's leading data analytics and AI officer. “They want a human voice on the other side of the phone.”
This is similar to an AI app developed over a year ago. It is for field workers at Johnson Controls, a large supplier of building equipment, software and services.
The company provided machine operational and service manuals to AI programs trained to generate a summary of the problem, suggested repairs, and brought them all to the technician's tablet computer.
In tests, the app trims repair calls for over an hour for 10-15 minutes. It's a useful improvement in efficiency, but by itself there is little change in the workplace.
Of the company's 25,000 field service workers, 2,000 have access to AI helpers, but the company is planning to expand.
“It's still pretty early on, but the idea is that over time everyone uses it,” said Vijay Sankaran, Chief Digital and Information Officer at Johnson Control.
The long-term vision is that companies will use AI to improve multiple systems, including sales, sourcing, manufacturing, customer service, and finance.
“That's the game changer,” said Sankaran, who predicts the shift will take at least five years.
Two years ago, JPMorgan Chase, the country's largest bank, blocked access to ChatGpt from computers due to potential security risks. Only hundreds of data scientists and engineers could try out AI.
Today, around 200,000 bank employees can access generic AI assistants (essentially business chatbots) from their computers for tasks such as retrieving data, answering business questions, writing reports, and more.
Tuned for your JPMorgan use, assistants tap on CHATGPT and other AI tools to ensure data security for sensitive banks and customer information. About half of workers use it regularly, reporting that they spend up to four hours a week on basic office tasks, the company said.
Bank wealth advisors also employ more specialized AI assistants who use banks, markets and customer data to provide investment research and advice to wealthy clients. The bank says it will help its advisors to obtain information and make investment recommendations as quickly as possible as before, helping to increase sales.
Lori Beer, global chief information officer at JPMorgan, oversees 60,000 global technology staff. Has she shut down the AI project? Perhaps a total of hundreds, she said.
But many of the shelved prototypes developed concepts and code that were folded into other ongoing projects, she said.
“We're absolutely closing things down,” Beale said. “We're not afraid to shut things down. I don't think that's a bad thing. I think it's clever.”
McAfee, a research scientist at MIT, agreed.
“It's not surprising that there is a lack of early AI efforts,” said McAfee, founder of AI consulting firm Workhelix. “Innovation is a process that fails quite regularly.” NYTIMES
