Welcome back to Untangled. It’s written by me, Charley Johnson, and supported by members like you. Today, I’ve done something a li’l different: I used my STEWARD framework to analyze this incredible piece of journalism about how nurses are navigating the adoption of AI. It’s ultimately a story about a workflow that was never designed to protect the judgment it depended on.

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

On to the show!

🏡 Untangled HQ

​This Week:

• Stewarding AI: I launched my new course, “Stewarding AI: How to Build Responsible Policies, Workflows, and Practices.” Be the first to hear when it’s open for enrollment. ​

• Untangled Collective: I hosted Untangled’s March community event on practical approaches to world building and mapping backward. Next month’s event is all about mapping power.

Learn How to Map Power

Coming Up:

• Stewarding Complexity: Bad news: the stable, predictable world your governance frameworks were built for doesn’t exist anymore. Good news: we’re figuring out what to do about it—together. Join Aarn and I this Tuesday for our next event on Stewarding Complexity.

• The Facilitators’ Workshop: Every team has a nay-sayer, a bomb-thrower, and someone who’d rather let tension quietly fester than have a five-minute uncomfortable conversation. This free one-hour session is for everyone who has to work with them.​

• Systems Change: Tomorrow I’m opening enrollment to the waitlist for Cohort 7 of Systems Change for Tech and Society Leaders. Those already on the waitlist will get 40% off the new price ($1200) if they sign up by March 13 -- so ... get on the waitlist today!

Get 40% Off

🧶 Deep Dive

​

I want to tell you a story about a nurse named Adam Hart.

Hilke Schellmann — journalist and author of The Algorithm — has been investigating how AI is reshaping work across industries. In her reporting on what’s actually happening on hospital floors, she follows Hart to the bedside of a patient flagged by a sepsis alert. An algorithm generated an order. His charge nurse told him to comply. Hart refused. He had noticed something the algorithm couldn’t: a dialysis catheter. Administering fluids would have harmed the patient. He also had “that feeling” — a somatic, non-algorithmic form of knowing built over years of watching bodies decline and recover.

This isn’t a story about one difficult charge nurse. It’s a story about a workflow that was never designed to protect the judgment it depended on. I’ve spent the last year developing a seven-step framework called STEWARD for exactly this situation. Not just in hospitals, but in any organization deploying AI into complex human work. Let me walk you through what Schellmann’s reporting reveals, step by step.

S — See the System

The hospitals in Schellmann’s reporting mostly skipped this step entirely. They asked, “What can we automate?” rather than, “What future are we trying to create, and does this tool help or hinder it?” The sepsis alert and the BioButton arrived with hype-laden launch narratives — UC Davis called BioButton “transformational,” and the industry promised continuous, around-the-clock monitoring that no human clinician could match — which substituted for genuine systems analysis. Nobody mapped how the alert would reshape authority on the floor, or the choreography between nurses and physicians. They bolted the tool on and called it innovation.

T — Trace What Must Stay Human

This is the heart of Hart’s story. His refusal wasn’t defiance — it was the exercise of exactly the kind of judgment STEWARD says must remain human. The “feeling” Hart and his colleague Beebe describe — documented carefully by Schellmann — isn’t soft. It’s non-delegable. It exists nowhere in any training dataset. It develops through years of getting things wrong, reading skin, and noticing how a patient’s breathing changes between 3am and 4am.

Hospitals that deployed these tools without first asking “what must stay human?” created conditions where those capacities were treated as obstacles rather than assets.

Subscribe now

E — Evaluate New Risks, Accountability Shifts & Loss

When the sepsis alert generated an order, it quietly became a verdict. The algorithm redirected authority, compressed deliberation, and placed the burden of proof on the clinician who disagreed. When something went wrong, responsibility was impossible to locate but nurses absorbed the consequences either way. This is the “problem of many hands” — and it’s not an accident of bad design. It’s what happens when you deploy AI without explicitly asking who is accountable when the machine is wrong.

W — Workflow Redesign

The sepsis protocol is the clearest failure in Schellmann’s reporting. The workflow wasn’t redesigned when the AI arrived — it was just augmented. Alert fires → charge nurse issues order → bedside nurse complies. No mandatory interpretation moment. No named override authority. No procedure for when the alert is wrong. STEWARD asks: who interprets this output before anything moves? What triggers a legitimate override? None of those questions appear to have been answered before deployment. The result was a policy built on a model that substantially underperformed its marketing, with nurses left holding the consequences.

A — Adjust Interfaces & Tempo

The BioButton story Schellmann documents is interface design gone wrong in almost every way. The alerts were vague, frequent, and hard to interpret. They pulled nurses away from patients already flagged as high-risk to investigate false positives. As one nurse explained,

“I have my own internal alerts—‘something’s wrong with this patient, I want to keep an eye on them’—and then the BioButton would have its own thing going on. It was overdoing it but not really giving great information.”

The interface encoded urgency without specificity, training nurses to respond to pings rather than to patients. STEWARD asks: does this interface support critical thinking or passive compliance? Is uncertainty visible? Are alternatives presented? In almost every case Schellmann describes, the answer is no. The alert arrives with apparent authority and no explanation. As one researcher she spoke with put it, AI models should be able to explain why they’re recommending something.

R — Review What the System Teaches

The nurses witness, in real time, what these systems teach. They’re teaching clinicians to wait for an alert before acting. They’re teaching deference — “the AI said so” as a dead end of clinical reasoning rather than a starting point. One system’s alerts are estimated to be false positives half the time, yet policy requires a response to every one. Over time, this erodes exactly the clinical instinct the story celebrates. If the next generation of nurses learns to wait for an alert, the “feeling” Beebe and Hart describe stops developing. That’s the most dangerous long-term cost, and the one least visible in any ROI calculation.

D — Detect Drift & Design Corrective Moves

The drift Schellmann documents is already institutional. What began as a tool to support clinical judgment has, in some places, become a substitute for it. The nurses see it. They’re staging demonstrations, testifying before city councils, and going on strike — not because they’re anti-technology, but because they’re watching the root system get plowed under in real time.

UC Davis eventually stopped using BioButton because nurses were catching deterioration faster than the device was. That’s a monitoring system working. The problem is that it took a year, a pilot, and significant friction to get there.

Schellmann closes her piece with a line worth sitting with: “the ultimate value of the nurse in the age of AI may be not their ability to follow the prompt but their willingness to override it.”

STEWARD would add one thing: organizations need to be designed so that overriding is easy, expected, documented, and studied — not treated as defiance.

The fact that Adam Hart’s refusal landed as defiance isn’t a personnel problem. It’s a systems problem. And it’s entirely preventable.

I sent Schellmann a draft of this in advance to get her feedback. She had these kind words to say: “What a powerful way to use reporting to better real-world practices - I am so thrilled this work can be utilized to not only show what we need to work on, but also exactly how to implement AI responsibly in organizations.” This is the first time I’ve used my frameworks to map real world case studies and share it as a newsletter -- what did you think? Please respond to this poll or hit ‘Reply’ and let me know. Thanks! ​
​

⚒️ Systems Change

​Think your organization has a healthy relationship with technology? There’s a diagnostic for that -- and it’s free!

Get free tool

​

💫 Work With Me

​Here are 4 ways I can help:

• ​Facilitation: I can help facilitate your team through complex and fraught dynamics, so that they can achieve their purpose.

• ​Advising: I can help you navigate uncertainty, make sense of AI, and facilitate change in your system.

• ​Organizational Training: Everything you and your team need to cut through the tech-hype and implement strategies that catalyze true systems change. (For either Stewarding AI or Systems Change for Tech & Society Leaders)

• ​1:1 Leadership Coaching: I can help you facilitate change — in yourself, your organization, and the system you work within.

​

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​supported​ by members like you. ​As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

On to the show!

🏡 Untangled HQ

• New: ​I launched a referral program!​ Know someone who’d love Untangled? Send them our way. Every ​referral​ gets you closer to free courses, a hard copy of AI Untangled, a 1:1 coaching session, and more. The more you share, the more you get back — and honestly, it would make my whole week. ​Share your unique referral link and start earning rewards.​

• Next Week: Wednesday’s ​Untangled Collective​ event offers a practical guide to world building that stays grounded — and actually changes how you act today. ​Join me and 25 others.​

• Coming Up: On March 9th, Cohort 7 of Systems Change for Tech & Society Leaders will open to those on the waitlist. ​Be the first to hear about it. ​

• Tomorrow: I’ll pop into your inbox with a very exciting announcement. Stay tuned!

🧶 Deep Dive

Sorry Dario, Work Is Weirder Than That

The biggest risk to work isn’t AI — it’s acting on the prophecies of AI leaders who misunderstand what work actually entails.

Anthropic CEO Dario Amodei and Microsoft AI CEO Mustafa Suleyman ​are both warning​ that AI could eliminate half of all entry-level white-collar jobs. I have my doubts, but let’s assume that they believe these predictions; that they aren’t simply propagating a hype-y, self-serving narrative. In this scenario, they’re still making a critical error — actually, two of them.

The first is a category error. As Dave Edwards of the Artificiality Institute writes in ​a great new essay​, they’re confusing tasks with work, and completing tasks with value creation. Think about your job — how much of what makes you effective can be quantified, and turned into data? How much of your work is consistent and bounded by explicit, objective success criteria?

Yes, large language models and agents can perform many tasks — especially those that require low discretion, explicit inputs, and clear success metrics. But the bigger risk is that we mistake tasks for judgment, ​meaning making​, imagination, discernment, taste, instinct, and creativity. The senior partner who knows which clause will blow up a deal because she’s seen that client panic before. The marketing lead whose best campaign idea emerged from a story overheard at lunch. These aren’t tasks, nor would they show up in training data.

Which brings me to Studs Terkel and his incredible book, Working. Terkel spent years in the 1970s interviewing hundreds of Americans about their jobs — steelworkers, receptionists, farmhands, executives, waitresses, spot welders. Working is one of the most empirically thorough accounts of human labor ever assembled. And what it documents, over and over, is that the official description of a job almost never captures what makes that job valuable. The hospital aide who knows which patients need what kind of attention before they ask. The stonemason who points to a wall he built fifty years ago that will still be standing after he’s gone.

Subscribe now

What Terkel documented — and what AI leaders systematically cannot see — is what ​Edwards calls the root system​: the complex human ecology underneath the visible work that makes the visible work possible. The trust that took years to build. The judgment that developed through getting things wrong. The institutional memory that tells you which rules are real and which ones everyone routes around. The relationships that turn a difficult conversation into a resolved one. Remove all of this and the organization becomes something qualitatively different and less capable.

If Amodei and Suleyman aren’t hyping AI but making a genuine category error, the question becomes: why? In the 1970s, political theorist Langdon Winner developed a concept he called ​reverse adaptation​. The idea is deceptively simple: we assume technologies are designed to serve human purposes, but over time the relationship quietly inverts. We start reorganizing our purposes to serve our tools. In this way, the technology’s logic — what it can measure, what it can process, what it rewards — gradually becomes the organization’s logic. What the tool can’t see stops counting as real. I wrote about this awhile back in ​”Trapped by the Pursuit of Efficiency.”​

This is why the category error Edwards calls out is so problematic. If Dario or Mustafa were trained as anthropologists, they’d see a fundamentally different system. They wouldn’t see work as a collection of discrete tasks. What reverse adaptation explains is why this framing feels so natural to them, and why it’s so dangerous. It’s that the tools they’ve built have become the lens through which they see everything — including the human systems those tools are supposed to serve. The tool has adapted them.

This brings us to the second error — the strategic one. This isn’t just a risk to workers; it’s a risk to companies and the broader economy. As Edwards ​writes​:

When a CEO announces that AI can replace their knowledge workers, they’re not making a bold bet on the future. They’re making a confession about the present…They’re telling you that they’ve replaced their unique differentiation with something available to every competitor on earth for $20/month.

Right, if you replace your knowledge workers with AI, you’ve replaced something unique and hard to replicate — accumulated judgment, institutional memory, relational intelligence, productive friction — with a commodity. Remove all the complex human stuff and you don’t get the same work done cheaper. You get a different, diminished kind of work — and eventually, a different, diminished kind of organization.

Share

So what’s the alternative? I offered a starting point in my essay ​What Remains Human​ and the question: how do we redesign work that takes advantage of what machines do well while protecting what makes us irreplaceably human?

This means taking seriously what Terkel documented: that the most valuable things happening in organizations are often invisible to management, unmeasured by metrics, and completely absent from job descriptions. It means ​designing workflows​ that protect those capacities rather than hollowing them out. But it also means watching — actively, continuously — for the signs that those capacities are eroding anyway. It’s building the habit of asking, regularly and honestly: what is this system actually teaching people to do? Is the tool adapting us?

Terkel’s interviewees knew something was being taken from them — the craft, the discretion, the sense that their judgment mattered — even when they couldn’t name it precisely. The organizations lost what truly mattered without ever noticing it was there.

We’re at a similar inflection point. The difference is that we can see it coming. The question is whether the people making decisions about AI have read enough Terkel — or enough Winner, or enough Edwards — to watch for the drift before the root system is gone.

⚒️ Systems Change

Finally: a free way to map power without having to become a shadowy elite! Your system is made of different actors. Anything that makes a difference in a system counts as an actor -- both people and technologies. Learn how to map them and shape them with this free tool.

➕ More From Untangled

​Untangled Podcast: YouTube | Apple | Spotify ​

• For Funders: Funders Must Rethink How to Lead Through Tech-Inflected Uncertainty​

• For Regulators: What If We Regulated Chatbots Like Any Other Product?​

• For Tech Leaders: What Must Remain Human.

• For Everyone: The Intelligence of a Hunch: What AI Will Never Have.

💫 Work With Me

Here are 3 ways I can help:

1. ​Courses & Trainings: Everything you and your team need to cut through the tech-hype and implement strategies that catalyze true systems change.

2. ​Advising & Facilitation: I can help you navigate uncertainty, make sense of AI, and facilitate change in your system.

3. ​1:1 Leadership Coaching: I can help you facilitate change — in yourself, your organization, and the system you work within.

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​valued by members like you.​ This week, I’m sharing ​an essay that Michelle Shevin and I published earlier this week in Tech Policy Press​ about how funders must rethink how to lead through tech-inflected uncertainty.

AI promises to optimize away uncertainty. But generative …

Read more

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​valued by members like you.​ This week, I’m sharing a process any organization can use to reconfigure human-machine workflow. Because the key question for companies and organizations adopting AI isn’t “what can we automate,” it’s How do we design work that takes advantage of…

Read more

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​valued by members like you.​ Today, I’m sharing my conversation with Ben Winters, Director of AI and Privacy at Consumer Federation of America, about ​The People First Chatbot Bill​—model legislation for regulating chatbots that’s been endorsed by over 70 organizations.

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

🔦Untangled HQ

The ​Untangled Collective​ held its third community event earlier this week. Here’s what one participant had to say:

Join The Collective

On Tuesday, I’m launching another community with Aarn Wenneckers: ​Stewarding Complexity.​ This one is for boards, CEOs, and organizational leaders who need to step outside formal governance structures and practice making sense of complexity in real time—together. ​Join us?​

🧶 Chatbots Don’t “Just Happen.” Companies Make Choices.

Tech companies have successfully made chatbots seem like mystical, uncontrollable entities while simultaneously claiming they can be trusted without regulation. Yet, as Ben points out, every aspect of a chatbot—from training data to interface design to what responses get blocked—represents a series of choices by companies. When those choices foreseeably lead to harm, companies should be held accountable.

In our conversation, Ben and I dug into the key provisions in the Bill, including:

• Product liability: The bill leverages centuries of product liability law to hold companies accountable for design choices, rather than treating chatbots as neutral tools.

• Data minimization over consent: Instead of relying on checkbox fatigue, the bill prohibits using personal data from outside chatbot interactions.

• Private right of action: Harmed individuals can sue directly, not just rely on overwhelmed state attorneys general.

We also discussed how lessons from failed social media regulation informed this Bill —why content-neutral design matters, how consent-based models cement the status quo, and what it takes to overcome platform lobbying that claims regulation will “kill innovation.”

But more than any specific recommendation, the Bill serves as a reminder of the kind of world we could live in. It articulates an alternative future that we could inhabit. And here’s the good news: we know how to get there and state legislators are increasingly receptive.

Share

As civil society organizations look for what policies to push, and as states face push back from companies saying regulation will stifle innovation or that chatbots are too complex or that China will win etc., I hope they pick up a copy of ​The People First Chatbot Bill.​

It’s a lot simpler than the mystique that surrounds these bots — we just need to treat them like the products they actually are.

👉Before you go: 3 ways I can help

1. ​​Advising:​ I help clients develop AI strategies that serve their future vision, craft policies that honor their values amid hard tradeoffs, and translate those ideas into lived organizational practice.

2. ​​Courses & Trainings:​ Everything you and your team need to cut through the tech-hype and implement strategies that catalyze true systems change.

3. ​​1:1 Leadership Coaching:​ I can help you facilitate change — in yourself, your organization, and the system you work within.

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​valued by members like you.​ Today, I’m writing about why so-called “world models” can’t save AI, and the intelligence of a hunch.

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

Read more

Today I’m announcing a fun new addition to Untangled: subscriber chat.

This is a conversation space exclusively for subscribers—kind of like a group chat or live hangout. I’ll post questions and updates that come my way, and you can jump into the discussion.

I don’t yet know what exactly I’ll use this space for or how often I’ll start a chat. B…

Read more

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​supported​ by members like you. This week I’m sharing my conversation with Miranda Bogen (Director, AI Governance Lab, Center for Democracy & Technology) about what happens when your AI assistant becomes an advertiser.

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

Don’t forget to sign up for ​The Untangled Collective​ — it’s my free community for tech & society leaders navigating technological change and changing systems, and ​the next event is coming up!

Join The Collective

🏡Untangled HQ

🔦NEW: I’m teaming up with (complexity expert and author of ) to launch ​Stewarding Complexity​, a private, confidential gathering space for boards, executive teams, and organizational leaders to step outside formal governance structures, speak candidly with peers, and practice making sense of complexity — together. ​If that’s you, join us!​

🚨Not New, But Important: Every organization I speak with is facing the same two questions: How do we build strategy for uncertainty—and what should we actually do about AI?

My course, ​Systems Change for Tech & Society Leaders​ provides a structured approach to navigating both, helping leaders move beyond linear problem-solving and into systems thinking that engages emergence, power, and the relational foundations of change. ​Sign up for Cohort 6 today!​

Because why not: here’s a free ​diagnostic framework​ I use in the course to help you assess how your organization understands and uses technology across its strategy, programs, and operations.

Get Free Guide

🖇️ Some Links

How Certain Is It?

I’ve written ​a lot about why embracing uncertainty matters​. Chatbots do the opposite—they collapse uncertainty into confident-sounding responses, packaging blind confidence as a feature. But what if we designed these tools differently? What would it take to preserve uncertainty rather than erase it? A ​new paper​ tackles this challenge, arguing we need to protect the messier, harder-to-quantify forms of uncertainty that professionals navigate through conversation and intuition. Their proposed fix? Create systems where professionals collectively shape how different forms of uncertainty get expressed and worked through.

Blackbox Gets Subpoenaed

Job applicants are suing Eightfold AI, claiming its hiring screening software should follow Fair Credit Reporting Act requirements—giving candidates the right to see what data is collected and dispute inaccuracies.

Eightfold scores job applicants 1-5 using a database of over a billion professional profiles. Sound familiar? It’s essentially what credit agencies do: create dossiers, assign numeric scores, and determine eligibility.

The lawsuit argues: if it works like a credit agency, it should be regulated like one. As David Seligman of Towards Justice put it: “There is no A.I. exemption to our laws. Far too often, the business model of these companies is to roll out these new technologies, to wrap them in fancy new language, and ultimately to just violate peoples’ rights.”

Threatening Probabilities

Every time a chatbot threatens or blackmails someone, my inbox fills with “proof” of sentience.

But a ​new paper​ shows these behaviors aren’t anomalies—they’re just extreme versions of normal human interaction: price negotiation, power dynamics, ultimatums. Our surprise comes from assuming chatbots should only reproduce socially sanctioned behavior, not the full spectrum of how humans actually act.

Threats and blackmail don’t signal consciousness. They signal the model is drawing from the complete statistical distribution of human behavior—including the parts we don’t like to acknowledge. It’s probabilities all the way down, even when they’re uncomfortable ones.

🧶When Your AI Assistant Becomes an Advertiser

OpenAI just announced it will start testing ads in ChatGPT’s free tier. The press release was carefully worded—reassuring users that “ads will not change ChatGPT answers” and that “your chats are not shared with advertisers.” But as Miranda Bogen, director of the AI Governance Lab at the Center for Democracy and Technology, pointed out in a recent conversation, these statements are misleading and miss the entire point. What’s coming is a fundamental shift in who these systems serve—and what that means for people, privacy, and inequality.

To understand why this matters, we need to look at three things: how AI changes advertising signals, what “privacy” really means in this context, and why this could be harder to detect than anything we’ve seen before.

The Signal Problem

The question is: what happens when your AI assistant becomes an advertiser?

Answering that question, according to Miranda, starts by recognizing that advertising is all about high fidelity signals of intent—data that accurately predicts what you want to buy or do. When an ad interrupts your experience on Facebook, it’s hoping that you’ll care; that perhaps something you clicked awhile back will still be relevant. That’s not a great signal. Searching offers a better signal. You’re typically using Google because you want something.

But ChatGPT is different. You’re not searching for information. You’re often thinking out loud, revealing what matters to you, what you’re struggling with, what you’re planning or hoping for. Each conversational turn reveals deeper context about your intent—creating rich data for advertisers.

Now, OpenAI wants those signals but, if you read the press materials, they’re clearly concerned about losing users. For example, they bend over backwards to say that your chats won’t be “shared with advertisers.” But according to Miranda, this is technically accurate but completely misleading. The platform doesn’t need to send advertisers a list of your conversations. That’s the whole point of advertising infrastructure—OpenAI will target ads on behalf of advertisers, shielding your specific data while making the connection happen anyway.

The press release also promises you can “turn off personalization” and “clear the data used for ads.” But there are multiple layers of personalization happening simultaneously (e.g. raw chat logs, explicit memory stored about you, etc.) and it’s unclear what exactly OpenAI is referring to. Plus, even if you did turn off all personalization and erased all memory in the system, the amount of information a chatbot has about you in a specific context window offers plenty of signal for advertisers.

The Relationship Problem

On Facebook or Google, it’s clear you’re dealing with an advertiser. Your intent is your own. The experience is transactional. But as Miranda argues, when your AI assistant or AI co-worker starts subtly suggesting new products or services, something fundamentally different is happening.

It’s closer to influencer marketing where paid recommendations come wrapped in the veneer of authentic social connection. But an influencer’s audience typically knows that they’re being paid to sponsor a product. With an AI assistant, the lines start to blur. It has been helping you draft emails, think through career decisions, process relationship struggles. You’ve built relational trust with it over months, so when it suggests a therapist, lawyer, or contractor, you might perceive it as trusted advice without knowing, of course, which providers paid to be in the pool the AI draws from. The persuasion is invisible, wrapped in the same helpful tone the AI uses for everything else.

The Visibility Problem

Personalized ads and privacy harms are a big albeit old problem. These tools will of course propagate discrimination, exploit people at vulnerable moments, reinforce stereotypes and biases, and shape what opportunities people see (and don’t!). But this evolution of the advertising model brings something new: these harms will be even harder to identify.

Why? Because these systems are being built to connect with each other. AI agents will call other tools, connect with your bank and service providers, exchange information across an ecosystem of interconnected systems. There will be money and incentives flowing through this network in ways that are nearly impossible to track.

As Miranda put it:

“Even just tracking where any of this is happening, where exchanges of money and incentives are happening behind the scenes and where that might be shaping people’s experiences will just be even more challenging to keep up with over time.”

If your inner monologue so far is “this all sounds very bad,” well, I get it. But we didn’t end the conversation without imagining alternative business models and policy solutions. Listen to the end for these, and hear what Miranda would do to shift power back to users if she were advising our next (fingers crossed!) President four years from today.

👉 Before you go: 3 ways I can help

1. ​Advising:​ I help clients develop AI strategies that serve their future vision, craft policies that honor their values amid hard tradeoffs, and translate those ideas into lived organizational practice.

2. ​Courses & Trainings:​ Everything you and your team need to cut through the tech-hype and implement strategies that catalyze true systems change.

3. ​1:1 Leadership Coaching:​ I can help you facilitate change — in yourself, your organization, and the system you work within.

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​supported​ by members like you. This week, I’m sharing my conversation with , journalist and host of the thought-provoking podcast, .

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

On to the show!

Subscribe now

🔦 Untangled HQ

I launched ​The Facilitators’ Workshop,​ a community of practice for leaders who want to perfect the craft of facilitating groups through conflict and ambiguity—so they can actually achieve their purpose. Our event on January 23, ​”From Conflict to Clarity & Connection,”​ will give you a structured process for diagramming conflict—a way to slow down, make invisible dynamics visible, and understand what’s actually happening before deciding what to do next.

I’m spinning up a lot of new things that I’m excited to tell you about. The best way to stay up to date on upcoming events and workshops is by joining ​The Untangled Collective​.

Join Today!

In season 1 of , Evan cloned his voice, hitched it to an AI agent, and then put it in conversation with scammers and spammers, a therapist, work colleagues, and even his friends and family. ​You can listen to that conversation here.​

In season 2 of Shell Game, Evan explores what its like to run a company with AI agents as employees. A real company building a real product with users and interest from venture capitalists. This is the future that Silicon Valley is actively trying to bring into existence. Sam Altman recently shared that some of his fellow tech CEOs are literally betting on when the first one-person, billion-dollar company will appear.

Now, all the hype would make you believe that we should welcome this future with open arms. Productivity will skyrocket. Time will feel abundant. Work will become frictionless and maximally efficient. That’s the story, anyway. You won’t be surprised to find that the gap between the hype and reality is, uh, massive. Evan and I talk about that gap, but Shell Game helps us see around the corner to what it might actually feel like to work with AI agents. It’s a story about:

• What’s lost when an organizational culture becomes sycophantic.

• What its like when your colleague regularly make stuff up, commits it to memory, and then repeats that thing in the future as if its real.

• Why words like ‘agent’ and ‘agentic’ belie the reality that these large language models don’t really do anything on their own.

• The costs and complexities of anthropomorphizing agents, and how we’re voluntarily tricking ourselves.

• What humans are uniquely good at, and what it means for automation and the evolution of work.

• What Silicon Valley misunderstands about the world they’re creating and what’s at stake in confusing fluency and judgement.

Share

Shell Game is smart, thought-provoking, and really funny. I can’t recommend it enough. I hope you enjoy my human to human conversation with Evan Ratliff.

🧶Want to go deeper?

If you finished our conversation thinking, “Okay… I need to think about this more,” let me help.

• ​Flattery as a Feature: Rethinking ‘AI Sycophancy’​

• ​There’s no such thing as ‘fully autonomous’ agents​

• ​It’s okay to not know the answer​

• ​AI isn’t ‘hallucinating.’ We are.​

That’s it for now,

Charley

Welcome back to Untangled. It’s written by me, ​Charley Johnson​, and ​supported​ by members like you. This week, I’m writing about hope, and why it is a complex systems skill.

As always, please send me feedback on today’s post by replying to this email. I read and respond to every note.

On to the show!

Read more

Three announcements before getting into it:

1. I launched a new workshop, Mapping Your System: How to Navigate Power, People & Change in Uncertain Times. Join me?

2. The next Untangled mee…

Read more

• Last week’s post, Responsible AI is Not a Principles Problem. It’s a Network Configuration Problem, is now available for free subscribers.

• I launched a new 2-hr workshop, Mapping Your System: How to Navigate Power, People & Change in Uncertain Times. You’ll learn to map the invisible relationships, dynamics, and power arrangements that reliably shape your system’s behavior — and pinpoint where meaningful intervention is actually possible. (The number of participants will be capped at 30 - hope to see you there!)

• Disney is licensing its characters for use on OpenAI’s social video platform, Sora. So… prepare yourself for deeply cursed videos of your favorite animated icons doing things you can’t unsee. But! In far better news, the first real licensing standard that gives creators control over how their content can be used by AI dropped this week. That’s genuinely exciting.

• A great new paper by on why the pursuit of “algorithmic fairness” is a category error. It pairs nicely with a piece I wrote earlier this year, Can AI Be ‘Fair’?—and you will be thoroughly unsurprised by my answer.

• A rare form of cancer connected to toxins in the local water supply is taking off in an eastern Oregon community that is home to a 10,000 square foot Amazon data center. Basically, big farms in the area contribute tons of wastewater laden with nitrates from fertilizer. Normally, the aquifer deal with runoff. But the water demands of the data center pushed the system beyond its limits. The aquifer became oversaturated—and then contaminated. The result? The data center has been drawing in toxic water to cool its operations, and that water has cycled back into the community’s supply. Tests now show nitrate levels eight times higher than Oregon’s safe limit.

• 🔦 Community Spotlight: A sharp and important distinction between ‘alignment’ and ‘coherence’ from . This might sound like semantics, but as Aarn argues “Organizations do not fail for lack of alignment. They fail because the system cannot generate coherence.” Don’t miss the conversation Aarn and I had on the mindsets and muscles leaders need to navigate uncertainty and complexity.

Subscribe now

Follow the Flows

Most organizations talk endlessly about “data” and “communication.” But few talk about what sits beneath both: information flows—the way signals, stories, numbers, and interpretations actually move through an organization. Flows determine what becomes visible, what gets ignored, what feels urgent, and what quietly shapes people’s day-to-day decisions.

If you want to change a complex system, you have to learn to see—and eventually reshape—those flows. So let’s do that, shall we?

The first thing to know about flows? They’re not neutral. Every flow is shaped by three intertwined forces:

1. Technical Infrastructures: The protocols, algorithms, or technical artifacts that physically route the data.

2. Context & Norms: The organizational routines or incentives that shape what gets shared (and what gets hidden).

3. Power: The dynamic of who controls what counts as “legitimate” information.

4. Positionality: How different stakeholders experience the same flow differently (e.g., a dashboard that feels empowering to a manager but exposing to a worker).

Share

Why This Matters Flows don’t just transmit facts; they shape how an organization perceives itself. They configure behavior (people act on what they see) and create blind spots (missing flows hide critical dynamics).

To see the flow clearly, you need to categorize your stakeholders differently. Pick one critical information flow (e.g., a weekly status report or a customer feedback loop) and run this three-step mapping exercise:

Step 1: Categorize the Actors

Don’t just list names. Label them by their role in this specific flow :

• Producers: Who creates the information?

• Mediators: Who filters, curates, or amplifies it? (These are your gatekeepers).

• Receivers: Who relies on it to act or decide?

Step 2: Label the Flow

Draw the arrow connecting them. Is this flow Formal (reports, dashboards, KPIs) or Informal (conversations, rumors, Slack channels)?. Hint: The informal flows often carry the real truth of the system.

Step 3: Make it Sociotechnical

Ask, “What technical artifact mediates this flow?” Is it an algorithm, a spreadsheet, or a rigid meeting agenda? How does that artifact shape the data before it reaches the Receiver?

But information flows don’t always … flow. They’re often blocked, distorted, or hidden. Here’s a li’l diagnostic checklist you can work through:

Read more

• Read my rant on why your organization doesn’t have a ‘tech problem’ — it has a mindset problem. Then check out the free tool I developed to help you and your organization move toward a sociotechnical mindset that accounts for people, power, and context.

• CNN partnered with prediction market, Kalshi this week. Proponents claim these markets si…

Read more

Be the first to sign up

Prediction markets went mainstream in 2025. Companies and governments continue to use predictive algorithms despite research continuing to show that they … don’t work. Social tokens, which I wrote about in 2022, are now on the rise — creators are tokenizing themselves and you can speculate on their “value.” Pump.fun—one of the platforms fueling this tokenization—hit $1B in revenue this year. AI gambling agents are starting to take off. None of this should surprise us. In a year defined by political volatility, climate crisis, and rapid technological churn, prediction and speculation have become our default societal response to uncertainty.

This entanglement isn’t new. As historian Jamie Pietruska shows, prediction emerged as a ubiquitous practice in the 19th century to soothe our collective anxieties of modernity. Media scholar Devon Powers argues that today’s forecasting industries go a step further: they actively manufacture the feeling of overwhelming change so they can sell prediction as the antidote. But speculation isn’t just psychological and social—it’s structural. As Aris Komporozos-Athanasiou argues in Speculative Communities, “our imagination becomes more and more speculative in nature” as we become “increasingly impatient with capitalism’s future promises.” It turns out, speculation becomes a rational survival strategy amidst chronic instability.

The point is, to navigate uncertainty requires that we’re clear eyed about the role prediction is playing in society. But … what actually is prediction? We casually toss the term around as if it’s a simple guess about the future. But it’s far more than that — here are four ways to think about it:

1. Prediction Is a Proxy for Power

Prediction consolidates authority in the present. By selling the idea of constant, overwhelming change, forecasters position themselves as indispensable guides, effectively monetizing uncertainty. The ‘AI’ industry has essentially become a forecasting industry — they sell uncertainty and change so that they become an authoritative voice in the face of transformation, creating the conditions for their own relevance.

This dynamic reveals that prediction is fundamentally a proxy for power: it determines whose expertise is legitimized and whose vision of the future is prioritized. As historian Amanda Rees notes, the homogeneity of those doing the predicting—often privileged, well-connected men—narrows the horizons of possibility. Predictions don’t merely anticipate the future; they police it. They encourage compliance with trajectories that keep existing power structures intact.

50% Off Untangled

2. Prediction Is Diagnosis

Read more

• I co-authored an academic article, “Troubling translation: Sociotechnical research in AI policy and governance” in the Internet Policy Review. Cool, huh?

• A new paper argues that you should stop thinking of automated decision systems as predictive and start thinking of them as policy interventions.

• Do you think a chatbot would make better dec…

Read more

Join!

Read more

• Ever wonder what would happen if you put a large language model in control of a robot vacuum? Apparently, the vacuum experiences a full-blown ‘existential crisis’ about its role in the world.

• What if we centered the voices of young people when trying to understand the impact of social media on their lives? If you were persuaded by The Anxious Generation, you need to read this retort from Maximilian Milovidov.

• A great new piece in The New Yorker explores “the data centers that train AI and drain the electrical grid.” The essay ends on the question, what happens when an economy built on AI and data centers runs out of high quality data? I explored that question a few year ago in “The Doom Loop of Synthetic Data.”

• Will using generative AI make you dumber? Quite possibly — but also, maybe not. The difference between the two is how yo use it, and how you make sense of it is, and what it’s not. That’s the summary of the latest literature on AI use and learning. But the broader trend is clear: we’re reading less, we’re writing like we text, and our individual and collective attention is full. So where do we go from here?

• A new diagnostic tool any funder can use to examine how their mental models, tech & data practices, and strategic routines either reinforce or challenge dominant power structures in society. Work in philanthropy and want to join an invite-only webinar covering the diagnostic and my course?

Free Funder Only Webinar

If you’ve sensed a shift in Untangled of late, you’re not wrong. I’m writing a lot more about ‘complex systems.’ To name a few:

• What even is a ‘complex system’ and how do you know if you’re in one?

• How to act interdependently and do the next right thing in a complex system?

• Why if/then theories of change that assume causality are bonkers — and how to map backward from the future.

• How do you act amidst uncertainty?

• How should we think about goals in an uncertain world?

• Here’s a fun diagnostic tool I developed to help you assess how your organization thinks, acts, and learns under complexity.

I am obsessed with complex systems because the world is uncertain and unpredictable — and yet all of our strategy frameworks pretend otherwise. We crave certainty, so we build plans that presume causality, control, and predictability. We know in our gut that the systems we’re trying to change won’t sit still for our long-term plans, yet our instinct to cling to control amid uncertainty is too strong to resist.

And honestly, in 2025, this shouldn’t be a hard sell. Politics, climate change, and AI are laughing at your five-year strategy decks!

Subscribe now

Complexity thinking helps us see this clearly — that systems are dynamic, nonlinear, and adaptive — but it, too, has blind spots. First, it lacks a theory of technology. The closest we get is Brian Arthur’s brilliant book, The Nature of Technology: What It Is and How It Evolves, which explains how technologies co-evolve with economic systems. (Give it a read, or check out write-up in Technically Social). But Arthur was focused on markets, not on social systems — not on how technology is entangled with people and power.

That’s where my course comes in. I’m trying to offer frameworks and practices for creating change across difference, amid uncertainty, in tech-mediated environments — approaches that honor both complexity and the mutual shaping of people, power, and technology. (And yes, Cohort 5 of Systems Change for Tech & Society Leaders starts November 19.)

Second, complexity is hard to talk about simply and make practical (that’s why my Playbook turned into a 200 page monstrosity!) Every time I use the words “complex” or “system,” I can feel the distance between me and whoever I’m talking to widen. I’ve been searching for thinkers who bridge that gap — who write about systems with both clarity and depth — and recently came across the brilliant work of , who writes the great newsletter (Subscribe if you haven’t yet!)

After reading his essay, Systems Thinking Isn’t Enough Anymore, I reached out and invited him onto the podcast. I’m thrilled to share that conversation — one that digs into the mindsets and muscles leaders need to navigate uncertainty and constant change, the need to collapse old distinctions between strategy and operations, and what it really means to act when the ground beneath us keeps shifting.

Before you go: 3 ways I can help

1. Systems Change for Tech & Society Leaders - Everything you need to cut through the tech-hype and implement strategies that catalyze true systems change.

2. Need 1:1 help aligning technology with your vision of the future. Apply for advising & executive coaching here.

3. Organizational Support: Your organizational playbook for navigating uncertainty and making sense of AI — what’s real, what’s noise, and how it should (or shouldn’t) shape your system.

P.S. If you have a question about this post (or anything related to tech & systems change), reply to this email and let me know!

• A short reflection on why I feel out of place at AI events.

• Your conversations with Claude, ChatGPT, and Gemini are being used for training. Chat carefully!

• Why paper is one of today’s most important innovations.

• If medical multi-agent systems repeatedly lost evidence, suppressed correct minority opinions, and failed to prioritize high-risk clinical outcomes, would you trust it? Me neither.

• Incredible report on how to resist, refuse, reclaim, and reimagine seemingly (BUT NOT AT ALL!) inevitable AI narratives.

Share

This week, I spoke with Harry Law, Editorial Lead at the Cosmos Institute and a researcher at the University of Cambridge, about AI and autonomy. Harry wrote a terrific essay on how generative AI might serve human autonomy rather than the empires Big Tech is intent on building.

In our conversation, we explore:

• What the Cosmos Institute is — and how it’s challenging the binary, deterministic thinking that dominates tech.

• The difference between “democratic” and “authoritarian” technologies — and why it depends less on the tools themselves than on the political, cultural, and economic systems they’re embedded in.

• The gap between agency (Silicon Valley’s favorite word) and autonomy, and why that difference matters.

• How generative AI can collapse curiosity — closing the reflective space between question and answer — and what it might mean to design it instead for wonder, inquiry, and self-understanding.

• Why removing friction and optimizing for efficiency often strips away learning, growth, and self-actualization.

• The need for more “philosophy builders” — technologists designing systems that expand our capacity to think, choose, and act for ourselves.

• Harry’s provocative idea of personalized AIs grounded in our own values and second-order preferences — a radically different vision from today’s “personalization” built for engagement.

Subscribe now

The conversation around generative AI has gone stale. Everyone is interpreting it through their own frames of meaning — their own logics, values, incentives, and worldviews — yet we still talk about “AI” as if it’s a single, coherent, inevitable thing. It’s not.

My conversation with Harry is an attempt to move beyond the binary — to imagine alternative pathways for technology that place human autonomy, curiosity, and moral imagination at the center.

If you’re fed up with imagining alternative futures and want to do the hard, strategic work of changing the system you’re in, and set it — and you! — on a fundamentally new path, sign up for Cohort 5 of my course, Systems Change for Tech & Society Leaders. It kicks off in three weeks and there are still a few spots available.

Enroll today!

Before you go: 3 ways I can help

1. Systems Change for Tech & Society Leaders - Everything you need to cut through the tech-hype and implement strategies that catalyze true systems change.

2. Need 1:1 help aligning technology with your vision of the future. Apply for advising & executive coaching here.

3. Organizational Support: Your organizational playbook for navigating uncertainty and making sense of AI — what’s real, what’s noise, and how it should (or shouldn’t) shape your system.

P.S. If you have a question about this post (or anything related to tech & systems change), reply to this email and let me know!

Everyone’s Talking Past Each Other About AI

I attended a bunch of “AI” conferences last year. The single theme that cut across all of them? Everyone was talking past one another!

With 100 people in a room, there were nearly as many perspectives on what “AI” even is:

• A p…

Read more

📖 Weekly Reads

• The top-10 strategic mistakes philanthropies and civil society organizations make (1-5 & 6-10) — and what to do instead.

• Picture a teeter totter with ‘tech power’ on one side and ‘people power’ on the other. Right, it doesn’t look good. I’d like to to live in a world where technology serves us — our relationships, our goals, and our system…

Read more