What AI Puts Pressure On
AI, Momentum, and the Work in Front of Us
This is an AI substack.
And this week I wrote about taking my computer away, telemetry, imposter syndrome, rapid change, and a question that keeps showing up in conversation: what’s your model?
None of those are model posts.
None of them are tutorials.
None of them explain how to prompt anything.
And they are all about AI.
There’s a version of “AI writing” that people expect. It usually centers on models, tools, benchmarks, or techniques. That framing makes sense from the outside.
But once you’re inside the work, the center shifts.
What AI changes first isn’t what you can do. It’s what fails under pressure. It changes where judgment is required, where systems become brittle, and where humans lose their footing if nothing is holding them steady.
That’s what this week was really about.
One piece was about boundaries under acceleration. When tools make it easy to keep going, effort isn’t the constraint. Containment is. AI increases throughput. Without boundaries, that doesn’t create leverage. It creates exhaustion. Choosing when to stop turns out to matter as much as choosing what to start.
Another piece was about legibility under speed. Telemetry, not as instrumentation, but as structure. Once systems move faster than human memory, you can’t govern them with stories or after‑the‑fact explanations. You need signals that make what’s happening legible while it’s happening. Telemetry isn’t about measuring people. It’s about keeping systems interpretable when speed removes the option to reason from first principles every time.
I also wrote about orientation amid rapid change. About tools and momentum, and noticing when something starts to feel slightly misaligned. The work I care about right now isn’t linear or scriptable end‑to‑end. It’s investigative. The cost isn’t typing commands. The cost is losing orientation. In that kind of work, visibility isn’t decoration. It’s structure.
Another thread came from a question that keeps surfacing in conversations: what’s your model? It’s an understandable question. It’s also often the wrong one. Models are components, not systems. Treating them as the center of the work obscures where judgment, responsibility, and risk actually live. Focusing on the wrong layer is an easy way to feel productive without being effective.
The imposter syndrome essays weren’t really about confidence. They were about calibration. AI changes feedback loops. Old signals disappear. New ones are noisy. When that happens, self‑doubt gets louder, not because people are less capable, but because the system is harder to read. Learning to interpret yourself accurately becomes part of doing the work well.
And over the weekend, I wrote about choosing a north star. Not as a branding exercise. As a way to decide what to pay attention to when everything is moving. When momentum is high, something has to anchor judgment. Otherwise, the loudest tool, the fastest workflow, or the most visible progress ends up making decisions for you.
Taken together, none of this is about AI as a thing. It’s about what AI puts pressure on.
AI doesn’t replace judgment. It demands it. It amplifies whatever systems, incentives, and habits are already in place. It makes boundaries visible. It makes weak signals louder. It exposes where orientation is fragile and where it’s been quietly holding.
I’m noticing this not in theory, but in my own day‑to‑day work. The questions that keep coming up for me aren’t about which model to use or which tool to adopt. They’re about where to stop, what to trust, how to stay oriented, and how to make decisions that still feel human when everything is moving faster than it used to.
When people ask what model you’re using, or what workflow you’ve adopted, they’re often reaching for something stable. I get that impulse. I feel it too. But the center of the work isn’t there. It’s in the choices we make about what to pay attention to, what to measure, and how we decide when enough is enough.
That’s why the work looks the way it does.
It looks like setting boundaries instead of chasing throughput.
It looks like designing telemetry early so systems stay legible.
It looks like choosing tools based on orientation, not fashion.
It looks like noticing when questions are framed at the wrong layer.
It looks like recalibrating yourself when feedback loops change.
It looks like having a north star when momentum is high.
This is an AI substack because this is the work AI makes unavoidable for me right now.
Not models in isolation.
Not prompts divorced from context.
But the systems, judgments, and humans underneath it all.
That’s what I’m here to name.
If any of this resonates, it’s probably because we’re all learning how to stay oriented while the water around us keeps moving.
Alison + Wiggins