I Know We Fired You, But Could You Come Back Now?

The knee-jerk reaction to AI being able to replace employees and increase productivity and performance has backfired.

Many organisations have come to the hard realisation that their thinking was wrong. Their error was thinking AI would replace employees rather than assist them.

They grasped at the bright, new, shiny thing without careful consideration of the implications.

They are waking up to the fact that they need the employees they fired to come back because those employees possess human qualities AI cannot.

The rehiring is not an AI problem; it is a leadership problem. 

According to TerDawn DeBoe writing for Forbes, “Robert Half reported that a total of 29% of those organizations which cut employees due to an AI related reduction had already rehired into the positions they cut.”

Forrester forecasts that approximately 50% of AI-attributed layoffs will be reversed through rehiring, while 55% of employers already regret making workforce reductions due to AI.

Gartner predicts that by 2027, 50% of companies that attributed headcount reductions to AI will rehire staff to perform similar functions.

The reality check

Several major organisations are reversing AI-driven job cuts and hiring back experienced ex-employees after finding that AI cannot replicate the institutional knowledge, judgement and complex problem-solving.

They are experiencing an AI reality check.

Ford Motor Company

In the early 2020s, Ford aggressively scaled up its automated quality systems by deploying more than 900 AI-powered cameras.

During this period, many senior engineers left the company or retired without a formal process to transfer their deep, unwritten institutional knowledge into the AI data models.

The automated systems missed critical, subtle mechanical flaws. Ford experienced a massive surge in warranty claims and costly vehicle recalls. 

By 2026, Ford had spent three years rehiring veteran engineers, known as “greybeards”, made up of former Ford employees and workers from suppliers. 

Charles Poon, Ford’s vice president of vehicle engineering, said, “We recognised that for us to enhance some of our automation and machine learning and artificial intelligence tools, we needed to ensure that they were trained by the most experienced individuals.”

“Artificial intelligence is a fantastic tool, but it's only as good as the information you use to train it."

Ford is now using the greybeards to train the AI systems and passing on decades of institutional knowledge to younger employees. 

Ford's chief executive officer, Jim Farley, said the initiative contributed to "hundreds and hundreds of millions of dollars" in reduced warranty and recall costs, with the company anticipating $1 billion in cost savings in 2026 alone.

Commonwealth Bank of Australia (CBA)

In July 2025, Australia’s largest bank announced it was cutting 45 customer service specialist roles within its call centres, as it was heavily investing in a newly deployed AI-powered “voice bot.”

The technology could not handle customer call volumes, and instead of improving the customer experience, it left customers without help, and call volumes spiked. 

According to the Financial Review, “Despite claiming the voicebot had reduced call volumes by 2000 a week, CBA was forced to admit calls had actually increased as managers scrambled to offer overtime and even pull team leaders onto the phones. 

The bank had to reverse its decision to cut jobs and apologise to the employees.

CBA spokesperson said, “We have apologised to the employees concerned and acknowledge we should have been more thorough in our assessment of the roles required. We are also reviewing our internal processes to improve our approach going forward.”

The pattern is becoming hard to ignore. AI may process transactions, but it lacks context, judgement, memory, empathy, or accountability.

Klarna

In February 2024, Klarna announced that its AI assistant, built with OpenAI, had handled 2.3 million conversations in its first month and was doing the work equivalent of 700 full-time customer service agents. The company said it had reduced average resolution time from 11 minutes to under two minutes and that this was expected to deliver significant cost savings.

But the limits of the replacement mindset soon became apparent.

AI was effective for simple, repeatable tasks. It was far less effective where customers needed context, empathy, judgement or help with complex issues. Customers do not experience service as a transaction-processing exercise. They experience it through confidence, reassurance and resolution.

By mid-2025, Klarna had begun rehiring human customer service employees. In an exclusive interview with Bloomberg News, CEO Sebastian Siemiatkowski said the buy-now-pay-later company’s heavy commitment to fully autonomous AI support had reached its operational limits. He announced a rare recruitment drive specifically to reintroduce human agents, emphasising that customer service quality had suffered from over-indexing on cost reductions and that giving customers a guaranteed option to speak to a real person was critical for the brand.

The lesson is not that Klarna was wrong to use AI. The lesson is that AI needs to be designed as part of a human-supported service model, not treated as a replacement strategy.

IBM

In March, Twinladder published a case study on how IBM reversed the most famous AI hiring freeze in corporate history. 

“In May 2023, IBM CEO Arvind Krishna announced the company would pause hiring for roles that AI could replace, projecting 7,800 jobs eliminated over five years. In February 2026, IBM's chief human resources officer announced the company was tripling entry-level hiring. The distance between those two statements is a masterclass in what happens when the AI replacement thesis meets operational reality. 

The initial AI rollout focused on HR, procurement and routine administrative functions.

IBM successfully trained its automated systems to handle roughly 94% of routine internal human resources requests. The strategy ran into a critical wall with the remaining 6% of employee interactions.

The AI proved incapable of resolving ethical dilemmas such as complex corporate situations, which require a human with empathy, context, choice, and in-depth corporate knowledge. 

It could not handle complex problem-solving that requires human judgement, and the automated code generation tools introduced subtle bugs that required highly skilled human oversight to diagnose and resolve.

This year, IBM abandoned its replacement strategy and announced it would triple its entry-level hiring across the USA for software developers, HR managers, and other corporate roles, which were the roles targeted for automation.

This story has two lessons. One, you cannot remove the human without due consideration and understanding of the outcomes and consequences. IBM now has employees sitting directly between the AI models and business operations. Their primary mandate is to provide the "human touch," using physical judgement and oversight to make the automated tools genuinely useful. 

The second lesson comes from IBM’s realisation that you cannot simply automate entry-level tasks without losing the institutional knowledge developed in those roles.

Twinladder said, “The problem was not that AI could not perform back-office tasks. It could, and in many cases, it performed them faster and more consistently than humans. The problem was what happened to the organization when the humans who had previously performed those tasks were no longer there. 

Entry-level employees do not merely execute tasks. They learn the business from the ground up. They develop institutional knowledge -- the understanding of why processes exist, not just how they work. They build the relationships, the contextual awareness, and the professional judgement that organizations require at every level above entry. They are, in the language of workforce planning, the seed corn of the talent pipeline.”

Seniority cliff

In a recent newsletter, I wrote about the Seniority Cliff and asked, “Who will be ready in 2030?”

“When we remove the base work that entry-level employees traditionally do, we remove the learning process. In five or ten years’ time, we will not have the senior talent that emerged from employees who worked on the front line, on the shop floor, at the front desk, in the call centre, on reception, on the service desk, etc. Many of the seasoned talent we have today learned their craft in rank-and-file roles.”

This is what researchers are calling the Seniority Cliff. Seniority is not just your age or tenure, but the accumulated knowledge you have acquired over the years.

Human touch 

You cannot remove the human touch. AI should not replace humans; it should augment them. While AI can automate routine tasks, humans remain critical for oversight, questioning, validating, and complex decision-making.

Emily Potosky, Senior Director, Research at Gartner, summed it up.

“AI simply isn’t mature enough to fully replace the expertise, empathy, and judgement that human agents provide. Relying solely on AI right now is premature and could lead to unintended consequences.”

The take-away

The lesson is not that AI has failed. The lesson is that leadership failed to ask the right questions before removing the people who held the context, judgement and institutional knowledge that made the work possible.

AI can accelerate, automate, and analyse. But it cannot replace the human understanding that comes from experience, relationships, ethical judgment and lived organisational knowledge.

The organisations now trying to rehire the people they let go are not correcting a technology mistake. They are correcting a leadership mistake.

The question leaders should be asking is not “Which roles can AI replace?”

The better question is, “Where can AI augment our people, strengthen our capability and protect the knowledge we cannot afford to lose?”

Because once that knowledge walks out the door, getting it back is far harder than keeping it in the first place.