Good morning, picnickers! Today we’re continuing the thread of the conversation from last week about the vicious cycle between LLM tech that promises everything and managers who don’t want to do anything. 

If you are a manager and want to do Something instead of Nothing, Noah Fang has a good collection of principles here.

Reading Material

One observation I’ve made in the past is that the most common AI use cases are not revolutionary but evolutionary, and the thing that they tend to scale is misbehavior. Scams and content mills, deceptive patterns and academic dishonesty – these were among the first use cases of LLM-generated text.

 So it should come as no surprise that the Nothing Manager has leapt at the opportunity to use AI for disengaging from responsibility and leadership more thoroughly than ever before. 

Nothing Management is nothing new. Developing your own point of view has always been a risky proposition; on the other hand, allowing the data to decide for you lifts the risk entirely off your shoulders (until reality catches up with you, and assigning blame is no longer sufficient to keep the stock price up). 

But this old-school flavor of Nothing Management had a key flaw preventing it from scaling: all the humans involved in the actual work could sometimes present critical perspectives, or evidence that the surefire strategy is not actually working. This happens to be one of the superpowers of UX – determining that the methods we are attempting will not help us reach our goals. Unfortunately those in charge often see user research as a buzzkill that needs to be avoided at all costs. 

Enter AI. No sooner did I make a joke that the managers who live in dashboards would soon migrate to living in chatbots that we got a banger of a story about no other than the CEO of Microsoft doing just that. In his own words, Nadella “relies on Copilot to deliver summaries of messages he receives in Outlook and Teams” and views Copilot agents "as his Al chiefs of staff, delegating meeting prep, research and other tasks to the bots. 

This should terrify you.  

Remember that hallucination is a fundamental limitation of all gen AI. Every model hallucinates at least some and more advanced AI models hallucinate more. Meanwhile, AI atrophies your critical thinking skills; this finding was confirmed by a recent study from Microsoft itself, suggesting that a person's critical thinking skills atrophied the more they relied on, and trusted in, AI responses.  

These effects create a vicious cycle of becoming both unwilling and unable to validate whether the AI assistant’s summaries of the state of the world around you have any actual basis in reality. 

The Timeline

This is the apotheosis of the Nothing Manager. A chatbot tells him what has happened, and what he should do. It is hallucinating these outputs some of the time, but validating them would wash out all the “productivity” improvements the bots are providing in the first place. The entire sense-making process – normally driven by multiple cycles of human interaction to identify discrepancies and fill missing pieces – has been replaced with an unreliable black box. 

Nadella is not the only manager to isolate himself in an LLM bubble. AI has been a boon to all managers who want to distance themselves from human beings. Managers are sending AI simulacra to attend meetings and interviews in their place. Those who are responsible for handling customer complaints are reading them solely through AI-colored glasses rather than engaging with them directly, and abandoning analysis of actual data in favor of asking the AI to tell them what went wrong. 

Managers who still want to pretend that they are involved with their teams are generating imaginary conflicts for them to solve (by which I mean, to ask the AI to solve for them) – but not for long. Sam Altman wants to replace all of your colleagues with GPT agents, mirroring Zuckerberg’s efforts to do the same to your friends. 

Perhaps because they anticipate that they will not have to do any of this for long, Nothing Managers have given up not only on strategy, but on org design.

As Tiago Almeida and Jane Ruffino discuss here: productivity gains are now being framed exclusively as getting rid of people. The more Nothing Managers become isolated from workers, the more it’s appealing for them to press the big “layoffs” button – because they have no visibility into the extent that layoffs destroy organizational knowledge and productivity – and pretend that AI will, in some unspecified way, fill the gap. 

But let’s assume for a moment that all those layoffs were actually necessary; that the CEOs were right and precisely 20% of their carefully vetted employees suddenly became underperformers. And let’s ignore for a moment that the organization turning so many “rockstars” into underperformers should first take a long hard look at itself.  

Studies show us that technology can indeed help improve productivity and make up for the loss of contributors. But that result is not a given; when the people organizing digital transformation implement it carelessly, the technology ends up adding more work and productivity actually drops. 

So it is no surprise that when Nothing Managers introduce AI to the rest of their organization, it is in a ham-fisted way that creates more work. 

In 2024, AI in the workplace was optional, and so employees who didn’t find it useful (the majority of them) did not use it. So AI is simply no longer optional. The era of the velvet glove is over; the manager of 2025 pushes AI tooling onto employees with an iron fist.  

Instead of wondering whether the tooling they provided actually helped, Shopify simply made AI usage a factor in performance reviews. Elsewhere, similar AI mandates are driving employees to the brink, with top performers ready to quit because the tools are making their lives harder. Satya Nadella claims that 30% of Microsoft’s code is now written by AI, but public PRs show that this figure comes at the cost of extreme frustration as human engineers wrangle with Copilot’s garbage code. 

This trend is so widespread that this 2-part series is now turning into a 3-parter. Next week, I’ll cover just how much of a disaster Nothing Management actually is, with some cool numbers like “2/3 of CEOs don't see value in AI but are doing it anyway” and “half of CEOs say that their teams can’t align around the strategy.” Gosh, I wonder if those stats are connected somehow. 

— Pavel at the Product Picnic