The real trend is not "AI replacing humans," but rather a renewed consensus on human-AI collaboration.

Author and source: 0x9999in1, ME News



TL;DR

• Ford is rehiring hundreds of experienced engineers to address quality issues that AI automation systems cannot resolve.
• Last year, the Commonwealth Bank of Australia replaced over 40 customer service representatives with AI voice bots, resulting in system failures and a surge in incoming calls, leading to the reversal of the layoff decision.
• IBM's AI human resources system can handle 94% of routine requests, leaving 6% unaddressed, while entry-level hiring in the U.S. is projected to triple by 2026.
• Orgvue report: 39% of corporate leaders have laid off employees due to AI deployment, with 55% admitting to making poor decisions.
• Robert Half data: 32% of U.S. hiring managers have eliminated roles primarily due to AI, only to later rehire for the same or similar positions.
• Analysts' assessment is straightforward: increasing investment in AI while cutting jobs is not necessarily the best path to driving business growth.
• The real trend is not "AI replacing humans," but rather a renewed consensus on human-AI collaboration.

I. A trial conducted without prior approval is being quietly corrected

First, let me ask a question.

Do you remember the frenzy in 2023 to 2025 about "AI replacing XX jobs"?

Executives spoke with confidence on earnings calls. Analysts listed substitution rates in their research reports. Hiring managers quietly removed job descriptions and reviewed them with ChatGPT: Does this role still need a person?

The answer will be revealed in two years.

Required. And more so than imagined.

Ford is the latest to make the shift. This century-old manufacturer is rehiring hundreds of experienced engineers to address quality issues that AI automation systems cannot resolve. Charles Poon, Vice President of Ford’s Hardware Engineering, put it plainly: “AI is a great tool, but its effectiveness depends on the quality of the data used to train it.”

AI isn't perfect; if the training data is poor, it will fail.

And automotive quality issues are precisely the area where cutting corners is least acceptable.

This is not an isolated case. Last year, the Commonwealth Bank of Australia laid off more than 40 customer service staff and replaced them with AI voice bots. What happened? The system collapsed under pressure, customer complaints surged, and call volumes increased rather than decreased—ultimately forcing the bank to reverse its layoff decision and rehire the staff. IBM faced the same scenario: replacing certain HR functions with AI. While the system handled 94% of routine requests efficiently, the remaining 6% proved extremely challenging—encompassing ethical dilemmas, edge cases, and situations requiring human judgment. IBM subsequently announced it would triple its hiring of entry-level positions in the U.S. by 2026.

Triple.

If you had told an IBM executive this number in 2024, they would likely have thought you were joking.

II. Data Doesn't Lie: More Than Half of AI Layoffs Were Wrong

Now that we've covered the phenomenon, let's look at the data.

A striking figure in Orgvue’s report: 39% of business leaders have laid off employees due to AI deployment, and 55% of them admit that these layoff decisions were mistaken.

More than half.

This is not a euphemism like "the results didn't meet expectations"—it's an admission that the judgment was wrong.

Robert Half’s data is more direct: 32% of U.S. hiring managers reported having eliminated a role primarily due to AI, only to later rehire for the same or a similar position.

One-third.

Layoffs are carried out swiftly, but rehiring happens silently. Who bears the cost in between? Employees laid off bear it once; the company bears it again through rehiring. The remaining costs—declining morale, customer loss, and damaged brand reputation—are hidden bills.

Capitol Technology University clearly states: AI is transforming the workplace, but companies are finding that human-AI collaboration is more valuable than fully replacing human workers.

This statement may seem mundane, but it represents a calm conclusion to the two-year wave of AI layoffs.

Why?

Because the things AI excels at and the things humans excel at are not entirely overlapping. AI outperforms humans in efficiency when handling structured tasks, repetitive labor, and pattern recognition. But when it comes to ambiguous judgment, cross-domain reasoning, situational awareness, ethical trade-offs, or conversing with an angry customer—AI falters.

It's not that AI is bad—it's just that its capabilities have their limits.

III. Why did executives collectively misjudge?

Here’s the second question: Why are so many companies falling into the same trap at once?

It's not because executives are stupid. It's because AI-driven layoffs make short-term financial reports look too good.

Layoffs = Immediate reduction in labor costs = Improved financial statements = Positive short-term stock price reaction.

AI hype = embracing cutting-edge technology = analysts assign high valuations = telling a compelling story.

The combination of these two factors formed the standard earnings narrative template from 2023 to 2025: "We enhanced productivity through AI while optimizing our organizational structure."

We replaced humans with AI, and short-term profits increased.

But business isn't a quarterly race.

When Ford discovered that AI couldn't handle quality issues, defects may have already reached the market. When the Commonwealth Bank of Australia realized its voice bots couldn't cope, customers may have already been ranting on social media. When IBM found that 6% of edge cases were poorly handled, employees may have already filed complaints, legal risks may have emerged, and trust may have already collapsed.

Short-term accounting is easy; long-term accounting is harder.

The real issue is that the value of many roles lies not in handling 94% of routine requests, but in managing the remaining 6% of exceptions.

It’s the 6% that determines whether customers stay, employees trust the company, and the brand deserves respect.

AI can now handle 94%, and in the future, it may handle 97% or even 99%. But as long as that final 1% remains, humans must be involved.

And often, when people are not around, that 94% will gradually collapse—because no one is training the AI, correcting errors, or providing feedback to the system.

The most ironic thing about AI-driven layoffs is that you lay off the people who understand the business, leaving no one to teach the AI.

IV. The First Ebb of the AI Bubble Begins with People

Looking deeper, the phenomenon of AI-related layoffs represents the first sign of a retreat from the AI application bubble.

Over the past two years, the core AI narrative has been "replacement"—replacing customer service, replacing programmers, replacing designers, replacing HR, replacing analysts, replacing legal assistants...

One wave after another.

Every time, big companies step forward to say: "We've optimized X% of our roles with AI."

But how many have actually been successfully implemented?

Ford admitted at its own event that AI cannot handle quality issues. This is not a failed experiment by a small company, but a direct retreat by the world’s second-largest automaker. IBM, one of the most aggressive global investors in AI, has an AI HR system that can only handle 94%.

These two cases nearly宣告 the bankruptcy of the "complete replacement" narrative.

It's not that AI isn't capable—it's that the framework of "complete replacement" is fundamentally flawed.

What’s truly working is the "AI co-pilot" model. GitHub Copilot helps programmers write code faster, but doesn’t make them obsolete. Midjourney enables designers to produce drafts more quickly, but doesn’t put them out of work. ChatGPT accelerates analysts’ report creation, yet top analysts’ salaries continue to rise.

AI is a lever, not a replacement.

This understanding took the market two years, countless laid-off employees, and a series of corporate failures to gradually establish.

Five: What are the companies that are truly growing in the AI era doing?

One more question.

If AI-driven layoffs are wrong, then what is the right way to embrace the AI era?

Look at a few negative examples.

NVIDIA has not laid off large numbers of employees. TSMC has not. ASML has not. These three companies are core suppliers of AI infrastructure, and their employee counts are growing, not shrinking.

Now consider the application layer: Anthropic is hiring aggressively, OpenAI is hiring aggressively, and Perplexity is hiring aggressively. Even traditional financial institutions like JPMorgan Chase are massively recruiting for AI-related roles—not to replace existing staff, but to augment them.

The pattern is clear: companies truly benefiting from AI are hiring more; companies rushing to use AI to cut costs and lay off staff are rehiring.

What is the difference between this?

The former treats AI as a tool for expanding productivity, while the latter uses AI as an excuse to cut costs.

Expanding productivity means doing more, entering more markets, and creating more products. Cost compression means squeezing out inefficiencies within the existing operations.

Which is easier to bring long-term growth? The answer doesn't need me to say.

Six: Where did the laid-off employees go?

Speaking of this, we need to address a more serious topic.

Where did the employees laid off from those 39% of companies go?

Some of the laid-off employees have been rehired, but the vast majority have not.

According to Robert Half’s data, 32% of hiring managers admit they made a mistake, but when rehiring, they may not be bringing back the same people. The original employees may have switched industries, left the city, or given up on the field altogether.

This is the cruelest aspect of AI-driven layoffs: the cost of the decision is not fully borne by those who make it.

When executives cut jobs, the cost is borne by employees and their families. When they rehire, they may not find people with the same experience and willingness on the market. Ford is looking to rehire "hundreds of experienced engineers"—pay attention to those four words: "experienced." This is not a gap that new graduates can fill.

The scarcest resource in the AI era is not computing power or data, but seasoned experts who have spent over a decade specializing in a particular field, deeply understanding the business, and are willing to collaborate with AI iterations.

Once these people leave the industry, it’s very hard for them to come back.

Seven: Is this adjustment the end, or just a halftime break?

Final question: Does the AI layoff reversal mean this wave is over?

No.

It can only be said that the phase of "mindless replacement" has ended.

Over the next two to three years, we will see more refined AI application strategies—not asking "Can AI replace this job?" but rather "Which tasks in this role can AI enhance, and which must remain human-led?"

It's not "We laid off XX% of our employees," but rather "Our team's output increased by XX%."

The narrative framework will change.

Truly smart companies will start saying: We didn’t lay off anyone—each of our employees, with the help of AI, now accomplishes the work of three people, and we’ve expanded our business operations.

This story is harder to tell because it requires real business growth, not just talk of growth.

Precisely because it's difficult, it's real.

Eighth, in conclusion

Ford has brought back its engineers. Commonwealth Bank of Australia has brought back its customer service staff. IBM is tripling its entry-level hiring for 2026.

These are not isolated incidents in the news; they represent a shared shift.

An AI frenzy is being corrected by reality.

Analysts’ judgment is so straightforward it almost sounds like common sense: increasing AI usage while cutting employees is not necessarily the best path to driving business growth.

But two years ago, no one wanted to listen to this nonsense.

Now, the market has turned this nonsense into a consensus through real financial costs, a collapsed customer service system, and quality issues requiring rework.

Some lessons can only be understood after you’ve experienced them firsthand.

AI will continue to advance and take over more repetitive tasks across various roles. But those 6% of exceptions—the situations requiring judgment and experience, the ambiguous areas machines still can’t understand—will still need people.

The key is not to drive people away too soon.

It's easy to drive away, but expensive to bring back.

The real question has never been whether AI will replace humans, but: Have you clearly understood what AI can and cannot do?

I’ve thought it through—AI is leverage.

Didn't think it through—AI is a boomerang.

Ford, Commonwealth Bank of Australia, and IBM have already caught their own thrown ball.

Who will be next?

Reference materials

1. Orgvue, "AI and Workforce Transformation Report," 2025 Survey.
2. Robert Half, U.S. hiring manager survey data, 2025.
3. Ford Motor Company, Vice President of Hardware Engineering Charles Poon's related public statements.
4. Commonwealth Bank of Australia: Public information regarding the reversal of adjustments to its 2025 AI customer service system and related layoff decisions.
5. IBM announces data on AI human resources system operations and 2026 entry-level hiring plans in the United States.
6. Capitol Technology University, public research commentary on AI and workplace collaboration.