At his Substack last month, AI analyst Gary Marcus noted that companies are spending more on AI but not getting much of a return:

NVidia is down significantly more this year (23.4%) than the S&P (11.5%). Tariffs have driven down the market as a whole, but NVidia’s sharp decline is notable. One reason may be that it is starting to become clear that GenAI is not delivering the ROI (return on investment) people expected. 

“Poor ROI for GenAI,” April 10, 2025

He notes that according to a report at AI Talk-in, on a new survey from Coastal, “67% of companies plan to maintain or increase AI investments, yet only 21% report proven outcomes.”

The disappointing results point to the lack of a clear roadmap, outdated infrastructure, and fragmented data systems. These are typical recipes for failure in any venture:

Coastal CEO Eric Berridge emphasizes that AI tools like Salesforce’s Agentforce demand foundational shifts: “Modern infrastructure, redesigned workflows, and outcome-driven initiatives are non-negotiable.” Companies without these pillars risk wasted investments. 

Baziliks, “AI Spending Up, Results Lag: Coastal Report,” April 4, 2025

It sounds rather as though management personnel think that somehow their firms should be moving to AI — without any clear idea why or how.

But now, in two current instances, large firms are actually retrenching on AI.

Klarna Is Now Hiring Human Help Again

At Gizmodo, tech writer A. J. Dellinger tells us that Swedish-owned company Klarna, which provides online financial services, is backtracking on the AI assistants:

Company CEO Sebastian Siemiatkowski recently announced that the company intends to make sure that customers will always have the option to speak to a human when they need service …

It is a starkly different position than the company took just two years ago. Back in 2023, Siemiatkowski basically threw himself at AI, saying that he wanted his company to be OpenAI’s “favorite guinea pig.” The company instituted a hiring freeze and set out to replace as many humans on its payroll as possible with AI. By 2024, the CEO was bragging about cutting the company’s workforce nearly in half, dropping from a headcount of 3,800 to 2,000 by shifting to AI alternatives. He called the cutbacks “natural attrition” rather than the result of layoffs. 

“Klarna Hiring Back Human Help After Going All-In on AI,” May 11, 2025

Okay, so what happened?

Klarna claimed that AI chatbots were handling two-thirds of customer service conversations within their first month of deployment and went on to claim that AI was doing the work of 700 customer service agents. The problem is that it’s really doing the work of 700 really bad agents, and that quality took a toll. 

”Hiring Back Human Help”

We can infer that sales/satisfaction were down…

We won’t spoil the rest for you but Dellinger goes on to point out that a number of surveys have shown that customers do not want to deal with bots. Not only that but a “Gartner survey found about two-thirds of customers prefer that companies don’t use AI for customer service.”

Specifically, Gartner says they found:

Sixty-four percent of customers would prefer that companies didn’t use artificial intelligence (AI) in their customer service, according to a survey by Gartner, Inc. Furthermore, 53% of customers would consider switching to a competitor if they found out a company was going to use AI for customer service.

A Gartner survey of 5,728 customers conducted in December 2023 revealed that while customer service leaders are eager to adopt AI, customers remain concerned about its use within the customer service function.

“Sixty percent of customer service and support leaders are under pressure to adopt AI in their function,” said Keith McIntosh, Senior Principal, Research, in the Gartner Customer Service & Support practice. “But they can’t ignore concerns about AI use, especially when it could mean losing customers.” 

“Gartner Survey Finds 64% of Customers Would Prefer That Companies Didn’t Use AI For Customer Service,” July 9, 2024

Perhaps it is worth keeping in mind that bots not only don’t buy things themselves but may be chasing away humans who do.

Target and Walmart Are Reversing Course, Cutting Back on Self-Checkout

The Post Millennial reports,

Target is scaling back its use of self-checkout kiosks across its nearly 2,000 US stores, reversing a previous strategy due to rising concerns over theft.

The company had announced plans last March to expand self-checkout across all locations. However, it has since quietly removed kiosks from certain stores, shut them down in others, or imposed strict item limits such as capping self-checkout use to 10 items or less.

Target is not alone in this decision; similar retailers, including Walmart, have pulled back on self-checkouts and restored more traditional cashier-staffed lanes in the last year over shoplifting concerns.

While Target cited a shift in consumer preferences away from self-checkout after the COVID-19 pandemic, the decision also coincides with increasing concerns about retail theft. According to the New York Post, a recent incident took place in California in which a thief used Target’s self-checkout kiosks to steal over $60,000 worth of merchandise in a shoplifting spree that saw 100 visits to the retailer.

“Target cuts back on self-checkout amid growing theft concerns,” May 4, 2025

Perhaps the firms failed to take into account that the humans who have an incentive to shoplift have lots of time and opportunity to learn how to outwit the machines, which have no incentive at all. The shoplifters do not always have so much luck with other humans, who do have an incentive to stop them.

There will be many more of these tales in our future.