Do Executives Actually Understand AI?
I listened to Cal Newport’s Podcast on Executives and AI
Cal Newport’s podcast episode where he puts executives’ AI confidence under a microscope. Below is the LinkedIn post where I shared my thoughts and why I think this matters . And yes after my post have a look at the long summary or the podcast itself.
Is the impact of AI really understood by executives ?
I listened to a podcast by Cal Newport. I listened how Cal Newport broke down a major software company's AI strategy story from the NYT, and it was very eye opening.
The company's leadership said layoffs from two years ago were "partly due to AI."
Newport points out that in early 2024, there were no real coding agents yet. The tools barely existed. Meanwhile Jensen Huang, Marc Andreessen, and even Sam Altman have all walked back their own doom predictions this year, saying the mass job losses they feared simply haven't materialized.
So why do executives keep saying it anyway?
Because sounding AI-forward has become a proxy for sounding competent.
It's easier to say "we're bravely navigating a coming collapse" than to say "we're not entirely sure what's happening yet, and neither is anyone else."
One statement sounds like leadership. The other sounds honest. Right now the market rewards the first one.
This is the part that matters for anyone leading a team, not just this one company:
• Confidence is not the same as clarity.
• Managing from fear of missing a trend is not the same as managing from an inner compass.
I see this constantly in coaching conversations that have nothing to do with AI.
A leader adopts a narrative because it's circulating, not because they've actually tested it against their own experience.
Then they go on to build strategy, make headcount decisions, and communication to their teams on top of that borrowed certainty.
What to do ?
Stay grounded enough to ask: what do I actually know , versus what am I saying because it sounds sharp?
That question alone would change how a lot of AI strategy gets communicated.
Podcast Summary
There is a strange gap in the AI conversation right now.
On one side, many executives speak with enormous confidence. Jobs are about to change. Roles are about to disappear. Whole sectors are about to be transformed. The tone is often casual, as if the conclusion is already settled and the only remaining question is how quickly everyone can adapt.
On the other side, the actual evidence looks much less dramatic. AI is useful. In some cases, very useful. It is helping with coding, drafting, customer support workflows, and other narrow tasks. But that is very different from proving that mass layoffs are already underway or that widespread job destruction is just around the corner.
That tension matters because it shapes decisions inside companies. It affects hiring. It affects morale. It affects how leaders communicate change. And it affects how people think about their own future at work.
The deeper question is simple: do business leaders actually understand AI well enough to make such sweeping claims?
In many cases, the answer appears to be no.
The claim that sparked the question
A recent New York Times article about SAP framed the issue in a revealing way. The premise was not whether AI might eventually disrupt jobs. The premise was that disruption is already happening, and the real challenge is how to soften the blow.
That is an important distinction.
Once you start from the assumption that AI-driven job elimination is inevitable, every conversation shifts. Leaders are no longer asked to prove the threat is real. They are only asked whether they have a humane strategy for managing it.
In SAP’s case, the suggestion was that the company hoped to reinvent jobs rather than simply eliminate them. That sounds thoughtful and pragmatic. But underneath that framing sits a much larger assumption: that AI is already removing the need for significant amounts of human labor.
That assumption deserves more scrutiny than it usually gets.
Looking to Improve your Leadership ?
Why the “AI caused layoffs” story falls apart
One of the most striking claims in the discussion was that some prior SAP layoffs were tied to AI. On the surface, this sounds plausible. AI has been the dominant business narrative for years, so it is easy to attach almost any restructuring story to it.
But timing matters.
If a company says that job cuts in spring 2024 were partly due to AI, then the obvious question is: what AI capabilities were mature enough at that point to justify that decision?
The answer is: not much.
At that time, the major public tools were still largely chatbot-centered. GPT-4 had expanded multimodal abilities, but serious, reliable programming harnesses were not yet in widespread practical use. AI coding agents were not being systematically deployed across companies in the way they are beginning to be discussed now. Most organizations were still experimenting. The conversation was dominated by future potential, not proven workforce replacement.
So when leaders casually link those earlier layoffs to AI, the explanation sounds less like a grounded operational reality and more like narrative packaging.
This point becomes even harder to ignore when prominent voices close to the technology itself push back on the same story.
Nvidia CEO Jensen Huang called claims of large AI-driven job elimination “ridiculous” and criticized executives for scaring people to sound smart.
Marc Andreessen argued that software job losses were better explained by post-COVID overhiring and higher interest rates than by AI replacing developers.
Recent labor data showed a rebound in tech hiring, with software development openings rising to their highest level in years.
Even Sam Altman acknowledged that early fears about entry-level white-collar job elimination had not materialized the way he expected.
That does not prove AI will never reduce jobs. It does prove something narrower and important: the confident story that AI has already been the main driver of major recent layoffs is weak.
And weak stories become dangerous when repeated often enough to sound obvious.
The leap from “useful tool” to “economic upheaval”
A second problem appears in how some leaders reason from current AI usage to future economic collapse.
The logic often goes something like this:
AI is getting better at producing code and text.
Therefore many current job tasks can be automated.
Therefore many current jobs will disappear.
Therefore leaders must reinvent the workforce before it is too late.
The first step is true. The rest do not automatically follow.
Take software development. AI tools are now deeply involved in coding workflows. Developers increasingly describe what they want, review outputs, correct edge cases, and iterate with AI agents rather than writing every line by hand. That is a meaningful change in how coding gets done.
But a change in workflow is not the same thing as a collapse in demand for developers.
In fact, both can happen at once: the mechanics of coding can change while demand for people who understand systems, architecture, debugging, product intent, and quality assurance remains strong or even increases.
This is where many discussions become slippery. The phrase “AI writes code now” gets treated as if it means “programmers are becoming unnecessary.” That leap skips over the hard part: software work is not just typing syntax. It is specifying goals, managing ambiguity, integrating systems, testing assumptions, handling tradeoffs, and making judgment calls. AI can assist inside that process without replacing the whole process.
That distinction is central not only for tech leaders but also for anyone thinking seriously about productivity and sustainable performance at work. Tools can reshape execution without eliminating the need for human discernment.
The workforce may change without shrinking
One executive line of thinking suggests that companies will not necessarily become smaller, but they will become “very, very different.”
That idea is more reasonable than the layoff narrative, but even here the rhetoric often outruns the evidence.
Yes, roles evolve. They always have. Some work becomes less manual. Some work becomes more supervisory. Some new specialties appear around managing tools, shaping prompts, validating outputs, or integrating AI into existing systems.
But saying the workforce will look different is still not the same as proving that large categories of jobs are about to vanish.
In fact, history suggests a more modest and more common pattern:
new tools first augment existing workers,
teams then reorganize around the new capabilities,
job descriptions slowly shift,
and only later, if ever, does deep structural replacement happen.
This slower reality is less dramatic. It is also usually more accurate.
What often gets missed in executive commentary is that introducing AI frequently creates new coordination burdens. Someone has to decide what the tool should do. Someone has to verify whether the output is correct. Someone has to notice when the system fails in subtle ways. Someone has to own the consequences.
That does not look like instant labor elimination. It looks more like labor reconfiguration.
What companies are actually doing with AI right now
One useful way to cut through hype is to ask a very plain question: what are organizations doing with AI today inside real operating environments?
In the SAP example, the uses described were familiar:
improving or drafting patent-related applications,
handling some customer support requests,
generating software prototypes,
helping developers add features more quickly.
None of that is trivial. These are real applications. They may save time. They may improve throughput. They may reduce friction in specific workflows.
But they are also narrow use cases that have been discussed for quite a while.
This is the key mismatch: the practical use cases sound incremental, while the surrounding language sounds apocalyptic.
When the implementation evidence is “AI helps draft, clean up, support, and prototype,” it is hard to justify speaking as though the broader economy is on the edge of labor-market devastation.
That mismatch should make any careful reader pause.
So why do executives sound so certain?
An emerging explanation is that many business leaders are not reasoning from deep understanding. They are reasoning from atmosphere.
That atmosphere is built from repeated slogans:
“Use AI or be replaced by someone who does.”
“Every company is becoming an AI company.”
“White-collar work is next.”
“This is bigger than the internet.”
These statements may contain fragments of truth, but fragments are not strategy.
What often happens in organizations is that leaders feel intense pressure not to appear behind. In a climate like that, being wrong in an AI-forward direction can feel safer than being cautious and correct.
If an executive says, “We are boldly reinventing ourselves because AI will change everything,” that sounds visionary.
If the same executive says, “We are still trying to understand where AI genuinely adds value, and so far the effects are narrower than the headlines suggest,” that sounds less glamorous, even if it is more honest.
This creates an incentive problem. The social reward goes to certainty, not nuance.
And once that certainty spreads through conference talks, leadership memos, consulting decks, and LinkedIn commentary, it begins to harden into common sense.
That can become deeply frustrating for employees. They hear grand directives to “use AI” without clarity on what problem is being solved, which part of the workflow should change, or how success will be measured. The message becomes less like strategy and more like ritual.
The cost of AI vagueness inside organizations
When leaders speak in vague, inflated terms about AI, the damage is not only intellectual. It is operational.
Confused narratives lead to confused behavior.
Teams adopt tools without a clear use case.
Managers pressure employees to integrate AI into tasks that do not benefit from it.
People become anxious about replacement even when no concrete replacement model exists.
Communication becomes performative rather than useful.
Over time, that creates a workplace culture of low-grade stress and strategic noise.
Leaders who want to avoid that outcome need a much more grounded approach. That means identifying actual tasks, defining real constraints, evaluating measurable gains, and being candid about tradeoffs. It also means helping teams stay centered rather than reactive, which is closely tied to broader questions of stress management in demanding professional environments.
Not every technological shift should be processed as an emergency.
Looking to Improve your Leadership ?
The two groups the public struggles to trust
Part of the confusion around AI comes from an unusual credibility problem.
The people closest to the technology often alternate between confidence and doom. They promote capability with one hand and existential anxiety with the other. This makes their claims hard to interpret cleanly.
Meanwhile, many of the people most directly affected by the technology inside companies may not understand it deeply enough to assess it well. They rely on broad impressions, secondhand talking points, and trend pressure.
So the public ends up caught between two unreliable narrators:
those who may know the most but communicate in sensational ways,
and those who must make important business decisions but may be operating on thin understanding.
No wonder the conversation feels distorted.
A better standard for AI reporting
One of the strongest ideas to come out of this debate is that journalism should treat AI reporting with the same skepticism used in political reporting.
That does not mean being hostile to AI. It means being appropriately critical in a high-stakes domain where incentives, agendas, and blind spots are everywhere.
A stronger reporting model would look like this:
When an executive claims AI caused layoffs, ask for specifics.
When a company says jobs are being transformed, ask which tasks changed and how.
When someone predicts large-scale displacement, compare the prediction with current labor data.
When a source makes a sweeping claim, bring in independent technical experts who can evaluate what the tools can actually do.
When business language sounds dramatic, separate proven present capability from speculative future possibility.
This matters because AI is not just another gadget story. It has become a lens through which people interpret work, economics, education, and status. The reporting standard needs to match the impact.
Good skepticism does not slow progress. It improves judgment.
What a measured view of AI actually looks like
A measured view does not deny the importance of AI. It simply refuses to confuse possibility with proof.
A more balanced summary might sound like this:
AI is already useful in narrow but meaningful ways.
Programming workflows are changing, perhaps significantly.
Some writing, support, and coordination tasks can be accelerated.
Many organizations are still experimenting and learning what works.
Claims of broad, near-term job destruction remain much less substantiated than executive rhetoric suggests.
That version is less exciting. It is also more actionable.
It encourages companies to test carefully, train intelligently, and communicate honestly. It allows workers to stay adaptive without succumbing to panic. And it creates room for leadership that is thoughtful rather than theatrical.
For people in leadership roles, this is especially important. Clear thinking under uncertainty is one of the defining qualities of effective leadership. If you want to strengthen that capacity in a practical way, resources on leadership development can help translate abstract pressure into better decisions and steadier communication.
The real challenge for leaders
The real challenge is not proving that you are “AI forward.”
The real challenge is knowing what is true, what is premature, and what remains unproven.
That requires intellectual discipline.
It requires resisting prestige narratives. It requires not repeating dramatic claims just because they are fashionable. It requires asking boring questions about systems, labor, incentives, and actual workflows.
Most of all, it requires humility.
AI may indeed reshape many kinds of work over time. Some roles will evolve. Some tasks will disappear. Some sectors will reorganize in serious ways. But if leaders want trust, they need to stop speaking as if every possibility has already become reality.
There is a big difference between saying, “We are studying how AI can improve our work,” and saying, “The economy is about to be upended unless we reinvent everything now.”
Only one of those statements sounds mature.
Final thought
The most useful posture toward AI right now is neither blind enthusiasm nor cynical dismissal.
It is careful attention.
Care about the technology. Track what is changing. Learn the real use cases. Notice where productivity gains are genuine. Notice where rhetoric outruns evidence.
And when business leaders speak with total certainty about AI-driven job destruction, it is worth asking one more question before accepting the premise:
Do they actually know what they are talking about, or are they simply repeating a story that makes them sound current?
That question may be one of the most valuable AI skills anyone can develop right now.