Ben Lobaugh
  • Growth & Leadership
  • Giving Back
  • Sailing
  • VW Bus
  • About
AI is not a magic wand: Thoughtful use of AI to enhance your organization

AI is not a magic wand: Thoughtful use of AI to enhance your organization

Originally published at https://benlobaugh.medium.com/ai-is-not-a-magic-wand-thoughtful-use-of-ai-to-enhance-your-organization-135e95d818b3?postPublishedType=initial

Picture a leader who has just finished a promising report on AI. The projections look strong, the board is asking questions, and one idea takes hold: if the tools can do the work, why keep paying for the people who do it today?

The appeal is understandable. It is also a little like planning a budget around a magic wand, and wands are famously hard to procure.

When Harry Potter first went to buy a wand, he learned that the wand chooses the wizard, not the other way around, and that owning one was only the beginning of years of learning to use it. AI works the same way: the tool matters, but the skill and judgment of the person using it decide what it can do.

If you have watched a leader weigh replacing people with AI, or been asked to make that call yourself, you are not alone.

I led an engineering team through a move from human-written code to primarily AI-written code, and what carried the project was not speed. It was testing the tools first, rolling them out in stages, and keeping a person accountable for everything that shipped. The results were large. A roadmap scoped at three years shipped in seven months, and releases that once took most of a day took about an hour, with a human approval gate before anything went out. Those gains followed the groundwork. They did not replace it.

I described those steps in an earlier article. This article looks at the other side of the ledger: what has happened to organizations that skipped the groundwork and reached for the wand.

The promise and the evidence

The promise of the wand is simple: lower cost with quality unchanged. The evidence so far is more mixed, and it leans toward caution.

Forrester’s 2026 Future of Work predictions report that 55% of employers regret layoffs made for AI reasons, and it expects half of those layoffs to be quietly reversed. Gartner forecasts that by 2027, half of the companies that attributed customer service cuts to AI will rehire for similar functions.

Harvard Business Review offers a likely explanation. A survey of 1,006 executives found that AI-related layoffs are driven almost entirely by what leaders expect AI to do, not by what it has already done. The authors warn that this can sour employees on AI and force embarrassing rehiring, and they recommend being incremental and starting a process redesign effort instead.

What the early reversals have in common

Several organizations have changed course in public, and offer us several lessons.

Klarna: decide what you will measure before you cut. In late 2023, Klarna froze hiring outside engineering, and its AI assistant went on to handle about two-thirds of customer service requests, work equivalent to 700 full-time agents. In 2025 its CEO told Bloomberg that it hurt service quality, and the company began hiring again so customers could reach a person. Klarna has said the change is not a retreat from AI itself.

Commonwealth Bank of Australia: verify the result before you act on it. In 2025, the bank declared 45 customer service roles redundant after introducing a voice bot, saying the bot had reduced weekly call volumes. The union representing the staff said volumes were rising, that employees were being offered overtime, and that team leaders were answering calls. The bank reversed the decision, said its assessment had not considered all relevant business considerations.

Ford: capture what experienced people know before they leave. Bloomberg reported that over three years Ford hired, rehired, or promoted about 350 veteran engineers after its automated quality systems fell short of expectations. Ford’s vice president of vehicle hardware engineering explained that AI is only as good as the information used to train it, and that experienced people had left before their knowledge could be captured. Those engineers now lead design reviews, mentor younger colleagues, and help improve the AI tools, which remain central to Ford’s work.

IBM: redesign the work around the tool. IBM’s HR assistant resolves about 94% of routine employee requests, and people handle the rest. Instead of shrinking its early-career pipeline, IBM has said it will triple US entry-level hiring in 2026 and has rewritten those roles, including in HR, where newer staff step in when the assistant falls short. Its chief human resources officer has argued that cutting entry-level hiring trades short-term savings for a leadership gap three to five years later.

A fair counterpoint

The picture is not uniform. The CEO of IgniteTech replaced nearly 80% of the company’s staff over about a year after resistance to an AI mandate, and he reports strong results since. However, he also told Fortune that he does not recommend other employers copy the decision, and that people cannot be compelled to change something they do not believe in.

In these cases, the reversals followed changes made before the results had been verified.

Six checks before you reduce headcount for AI

These checks mirror the six steps of the earlier article, applied to the decision that tempts leaders most. If any answer is no, the decision is not ready.

  1. Settle security and ground rules. Decide what data may go where, which tools are approved, who approves additions, and what review applies before work ships. A cut made before these rules exist leaves fewer people to enforce them.
  2. Test the tool on your own work, and measure quality next to cost. A vendor demonstration is not evidence for your organization. Run the tool against your real workload with your own people, compare the results with what was promised, and track customer outcomes and error rates alongside savings.
  3. Capture what your experienced people know. Ask what they catch that no document records, and use that knowledge to shape and train the tool. Reduce those roles only after the tool demonstrably holds what they knew.
  4. Roll out in stages, and let each stage earn the next. Begin with low-risk work, widen the exposure only when the previous stage is settled, and prefer attrition and redeployment to sudden cuts.
  5. Keep a named person accountable for every outcome. Tools fail, credits run out, and vendors change their terms. The responsibility for the result stays with a person.
  6. Redesign the work, not only the headcount. Ask where work waits and where a human decision is truly needed, then reshape the roles around the answers. Protect the entry-level pipeline, since it is where tomorrow’s experienced people come from.

Use the tool, keep the judgment

AI is a capable tool, and used with care it can strengthen an organization. It does not remove the need for judgment, and it does not remove the need for the people who hold it. Move deliberately, and the results are more likely to last.

Sources

  • Davenport, T. and Srinivasan, L., “Companies Are Laying Off Workers Because of AI’s Potential, Not Its Performance,” Harvard Business Review, January 29, 2026.
  • Forrester, Predictions 2026: The Future of Work, as reported by HR Executive: “The AI layoff trap: Why half will be quietly rehired.”
  • Gartner forecast on customer service rehiring, as reported by HCAMag: “When AI redundancies backfire: Employers now scrambling to rehire humans.”
  • Klarna: eMarketer, “Klarna backtracks AI customer service plans,” May 8, 2025, reporting a Bloomberg interview; and tech.co, “Klarna Reverses AI Customer Service Overhaul.”
  • Commonwealth Bank of Australia: Bloomberg News via Staffing Industry Analysts, “Australia’s biggest bank reverses plan to replace jobs with AI,” August 26, 2025.
  • Ford: Quartz, “Ford is rehiring veteran engineers after over-relying on AI hurt its vehicle quality,” reporting Bloomberg and Ford statements; and Forbes, “Ford Hiring 350 Engineers After AI Failed Shows Human Value In AI Era.”
  • IBM: PYMNTS, “IBM to Triple Entry-Level Hiring as AI Rewrites Junior Roles,” reporting Bloomberg; and AI Magazine, “IBM Triples Entry-Level Hiring Despite AI Layoff Trends.”
  • IgniteTech: Fortune, “This CEO laid off nearly 80% of his staff because they refused to adopt AI fast enough,” August 17, 2025.
  • Lobaugh, B., “Transitioning to an AI-First Engineering Organization: A Practical Overview,” October 17, 2026.
benlobaugh.com
Leadership
Development
Developer Productivity
AI