AI may not replace engineers, but it could overload the most experienced ones

Much of the debate around AI in engineering focuses on the jobs it could eliminate. But as companies race to adopt the technology, AI may also create new demand for experienced leaders who can manage its risk.

While AI will most certainly accelerate engineering work, companies still need experienced operators to review its output for errors and unnecessary filler. Without that oversight, faster execution can produce poorly functioning products and pose real risks.

The rise of AI in engineering

AI is rapidly entering nearly every industry, and engineering presents both some of its greatest potential and some of its greatest risks.

AI can deliver significant benefits in complex work environments. Automating time-intensive tasks and predicting outcomes are just two ways these technologies are already contributing to engineering. At the same time, many warn of a future in which sophisticated AI tools replace human engineers.

That concern is understandable. Many professionals worry that AI could perform similar work at a lower cost and eventually eliminate their roles. But most people who use AI regularly recognize that this fear is either exaggerated or still a long way from becoming reality. Simply, the technology is not there yet.

Hallucinations (the tendency of AI systems to generate false facts or scenarios) remain a serious problem. Most LLMs in use today also draw from large archives of existing information, which means their output is often an uneven amalgamation of human-created knowledge rather than original reasoning.

Still, that has not stopped business leaders from pushing AI into employees’ daily workflows, which leads to my primary concern about the immediate future of AI: accuracy.

When used alongside a person who can check its output for hallucinations (weak reasoning or “fluff” sometimes called AI slop), AI can speed up workflows and support more accurate work, particularly in engineering. Complex formulas, simulations and experimentation require significant mental effort, so offloading part of that work to an AI tool is understandably appealing.

But the engineer using the tool must take the time to check and then understand the output. Without that human oversight, the product leaving an engineer’s desk could become the physical equivalent of AI slop: uninspired and lacking functionality.

As engineers increasingly use AI tools in their work, organizations need to ensure they have clear standards for reviewing its output. Ultimately, I believe senior leaders will become more responsible for making sure engineered products continue to meet the required quality standards.

Increasing responsibility for senior leaders

Engineering reviews typically follow an established process or rhythm. A senior engineer examines a piece of work, checks key calculations, follows the logic, evaluates the assumptions and determines whether the reasoning is sound. In most cases, they can tell whether the engineer understood the work they produced.

With the advent of AI, senior reviewers must now ask additional questions: What prompt produced this output? What data went into the model? What assumptions were made? What did the engineer accept simply because the result looked complete?

AI is very good at making work appear finished before anyone fully understands it. It can package half-formed thinking in polished language and confident structure. In many fields, that’s a nuisance. In engineering, it’s a gigantic risk.

Whether the work involves product design, mechanical engineering or large-scale civil infrastructure, the quality of an engineered product has direct consequences for the end user. In a consumer product, the result may be poor performance or customer dissatisfaction; a bad product launch. In critical infrastructure, the stakes can include public safety.

That’s why engineering must remain grounded in human judgment. AI can reproduce past patterns with speed and relative accuracy, but only a thoughtful engineer can determine whether the result is fit for purpose.

Larger implications for the engineering world

For COOs, vice presidents and directors, one of the most important questions to ask before adopting AI is whether the organization has a review structure capable of keeping pace with the increased speed of production.

If less experienced engineers can produce more work with AI, companies will need clear standards for how the technology is used and who is ultimately accountable for quality.

That responsibility will likely fall to senior leaders, either by adding oversight duties to their existing workloads or by hiring additional experienced engineers to serve as quality-control leaders for more junior teams. Yet the market for this talent is already tight, particularly as many seasoned engineering leaders approach retirement.

The industry will need to reconcile a workforce made up largely of junior engineers with extensive AI exposure and a relatively small group of senior leaders with the experience and judgment required to oversee their work.