AI-First is Just Corporate Theater

Every C-suite exec has the same script these days. "We're going AI-first," they announce at the all-hands, usually followed by some hand-wavy nonsense about "transforming how we work" and "staying competitive in the AI age." The subtext is always the same: get more productive or get out.

Then they head over to social media and start their performative posting, you’ve seen it on LinkedIn, about how they’re now an “AI-first organization”, to the applause of the hype crew. I see it, you see it, I click their profiles, and yup, there is usually nothing of substance that would suggest that they understand the technology or change management.

These executives are setting their teams up to fail spectacularly, alienating their best workers, and destroying their culture… if they had any culture worth protecting.

The Theater of AI Transformation

As an educator and technologist, I have a lot of conversations about AI, whether it’s in my fractional advisory capacity or building modern, effective reskilling and upskilling experiences. What the layperson doesn’t often get to see is the gap between boardroom and investor rhetoric and the ground-level reality.

While leadership celebrates on social media about their AI transformation, the actual workers are trying to figure out what the heck this technology is supposed to do. And this isn’t guesswork on my part; the media narrative is definitely shifting in the post-GPT5 world.

MIT just released a report that 95% of generative AI pilots are failing. Most of the conclusions I agree with, especially the identified education gap.

No Training

Many companies are expecting employees to become AI-savvy overnight magically. You see this in firms where they mandate that employees must use AI as part of their daily workflow.

I previously wrote about Goodhart’s law and how it’s likely to impact these initiatives. But the mandates make the assumption that executives, not the people actually doing the work, know what’s best for… You know, I’m not even sure what they’re looking for except a reduction in human capital.

That sounds harsh, maybe a bit extreme, but the numbers don’t lie. High failure rates, pretty much zero measured impact on GDP, and productivity isn’t increasing. If AI were really “10x’ing” productivity, we’d be seeing it in the macroeconomic numbers by now.

If a company wants to explore AI effectively, it first must educate its workers. But for the most part, all I see are vague directives about “using AI more”, with all the strategic clarity of a motivating cat poster at the dentist.

Where are the workshops on LLM fundamentals? What about security and data protection? And I certainly don’t see much in the way of ethics.

The top problem that potential clients are expressing to me is the effectiveness of the hype. They’re making large investments based on big promises with minimal results. “Eric, we need to educate our leadership teams so they don’t get taken by these shady vendors.”

Yes, you do. And I can help with that. But training isn’t enough! You’re still going to fail because you have

No Measurement Framework

I was running a workshop for a Fortune 500 company. The topic was a common one that many organizations lack: Data Literacy.

Gartner defines data literacy as:

“the ability to read, write and communicate data in context, with an understanding of data sources and constructs, analytical methods and AI techniques. Data literacy helps people identify, understand, interpret and act on data within a business context to influence business value or outcomes.”

Part of the workshop was discussing ways to turn data into actionable information. I asked the leaders in the room, “Do you trust your data?”. A bunch of nervous chuckles, and not a single confident “yes”.

How do you expect to be AI-first if you lack good data for it to access? If your team lacks good data literacy and you can’t trust your data, how in the world are you going to know if your AI initiatives are working? Hopes and dreams? A small study by METR showed what every tech leader should know: tech workers are awful at estimation.

“When developers are allowed to use AI tools, they take 19% longer to complete issues—a significant slowdown that goes against developer beliefs and expert forecasts. This gap between perception and reality is striking: developers expected AI to speed them up by 24%, and even after experiencing the slowdown, they still believed AI had sped them up by 20%.”

Does your team have the time and resources to clean up your data and then design and run controlled experiments? I rarely see it. What I see most often is vibes-based AI adoption.

But that’s ok, you have

No Infrastructure

Want to share learnings across teams? Good luck. Most companies haven't built systems to capture and distribute data from AI experiments. Every department is reinventing the wheel, making the same mistakes, and learning nothing from each other.

Do you like data silos? This is how you get data silos. Also, if you’re so hot to be AI-first, what answers do you expect to get from the tool without a single source of truth? What are you going to do when every team is using different vendor tools that do the same thing? Do you know where your data is being sent? Is it being mined? Maybe it’s posted online for everyone to see?

The Predictable Failure Rate

Let’s get back to that MIT report. 95% of AI projects are failing. Ninety-five percent. Let that sink in.

Sure, some of that failure is baked into the technology itself. LLMs hallucinate and get things wrong. It's not a bug to be fixed because it's fundamental to how their architecture works. Scaling has hit a wall. There is a lot of evidence that they can’t reason.

But a massive chunk of those failures? That's on leadership. That's on executives who mistake announcement theater for actual strategy. The need to understand

What AI-First Actually Looks Like

When I’m working with clients, the goal is to set realistic expectations and provide practical steps for exploring the benefits of AI tools. If done right, companies don’t have to mandate performative productivity increases by threatening job security. Unless the only reason they’re interested in AI is for wage reductions… then I guess they can have at it.

1. Education first: People need to understand what AI can and can't do, how to evaluate outputs critically, and how to design meaningful experiments.
2. Infrastructure investment: The team needs systems for experimentation, measurement, and knowledge sharing. It’s not just about buying enterprise ChatGPT licenses and calling it a day.
3. Cultural change management: Fundamentally, AI adoption is a human problem, not a technical one. People need psychological safety to experiment, fail, and learn. Create a clear plan about what you will do if the tool increases productivity.
   
   Author’s note: Smart companies will invest that productivity into improving their products and services. Dumb ones will gleefully reduce staff.
4. Realistic expectations. Stop promising magical 10x productivity gains. Start with specific use cases where AI can provide measurable value.

If you ignore or shortcut this process, it will become apparent that

The Real Threat to Your Business is Leadership

It is a failure of leadership to prioritize performative signaling over substantial change management. While these leaders are busy posting about their “AI journey” on social media, competitors are quietly building the organizational capabilities that drive meaningful impact by investing in people and processes.

Workers see through the theater. They know the difference between real support and empty mandates. Want to experience productivity loss? Wait until you annoy your best workers enough that they decide to move on.

The nature of work is always changing, and we need responsible leaders to steer organizations responsibly.

The best AI transformations happen when executives stop performing transformation and start doing the unglamorous work of enabling it.