I’ve found that rapid changes to how things work together often follow Bruce Tuckman’s four stages of group development: forming, storming, norming, performing1. Ai2 is no different. Generative Ai has been progressing rapidly, and I feel like we’re in the storming stage.

It’s not a rainstorm; rain can be cleansing. Ai is more like a dust storm3: it dirties everything and gets into places you didn’t know existed.

There are two prominent groups responding to this dust storm. At one end are the doomers: vocal critics who warn “Ai will take our jobs” or “It is going to kill us all.”

At the other are the zealots: enthusiasts who can’t get enough, who know every model and proclaim it the greatest thing ever; they think they “have all the skills”4.

This isn’t our first rodeo. When the Mellotron, an early keyboard that played pre-recorded tape loops of real instruments, gained popularity, it terrified traditional orchestral players. In 1967, when the rock band The Moody Blues tried to feature a Mellotron on a televised program, orchestra musicians panicked that the “lifelike string sounds” would put them out of work. Production ground to a halt until the band secured written permission from the musicians’ union to play the keyboard.

The mistake the doomers make is not recognising that Ai serves different purposes; we have more agency over its use than the headlines suggest. The mistake the zealots make is assuming Ai has taste when it does not.

I tend to be pragmatic: I’m somewhere in between. I’m aware of Ai’s power and wary of its limits. I see it as excellent for some tasks and poor for others. Much of the conversation reduces to “it’s not X, it’s Y” — a load‑bearing argument that often misses nuance.5

Our cognition has two broad modes: a logical, analytical side and a creative, aesthetic side. Ai is strong at the left‑brain tasks: writing code, designing algorithms, and trawling large datasets for patterns. For right‑brain tasks (subtle creativity and taste), Ai is much less reliable. Yes, those flyers you see everywhere illustrate that point.

ai flyer

I see many articles and posts with the refrain “Ai took my job”6, especially in software development. Digging deeper reveals a couple of patterns:

1) Some companies attribute mass layoffs to artificial intelligence to deflect from post‑pandemic overhiring and to reassure investors. Admitting to overhiring signals poor management; blaming Ai reframes cuts as forward‑looking efficiency. The available evidence suggests that only a small fraction of layoffs are directly caused by immediate Ai replacement7.

2) Other layoffs affect specialised, routine roles that require less creative judgment; for example, a Ruby on Rails developer who relies on a single stack and lacks broader delivery fundamentals.

Imagine every company replaced every employee with Ai agents using the same off‑the‑shelf models: how would any company keep a competitive edge? Everyone would be the same, much like every generic flyer looking identical.

So how do you survive this Ai dust storm?

T‑Shaped Skills

I’ve written about the importance of T‑shaped skills8 before. Now more than ever it’s valuable to be a generalising specialist: broad across many areas, with the ability to draw on Ai for depth “on demand.”

Exercise your right brain

Draw. Write a short, funny story. Make a joke. These are areas where Ai struggles. Cultivate taste and creativity.

You don’t need to be an Ai expert

There will be people in your networks who follow the latest models and tools: use them as resources rather than trying to become the lone expert.

Don’t stress about using it

Models are improving at left‑brain tasks and at understanding instructional intent. “Prompt engineering” is no longer a thing9; clear, plain instructions usually work. Error rates are dropping.

Summary

  • Thesis: T‑shaped skills help you survive the Ai dust‑storm by combining breadth with specialist depth.
  • Why it matters: Ai amplifies left‑brain, repeatable work; creativity, taste and varied experience remain human differentiators.
  • Actionable: Build broad skills, practice creative right‑brain work, and lean on Ai for depth when needed.

  1. Tuckman’s stages of group development — In this context the “group” is society as a collection of individuals affected by Ai. ↩

  2. Ai with a lowercase “i.” My name is Alister (you can call me Al), which can look like the letters “AI” in some fonts; I use “Ai” to avoid that ambiguity and because I prefer this styling. ↩

  3. The crasser version would be a shit-storm ↩

  4. “Ai skills” here refers to modular markdown files (skill.MD) containing names, descriptions, and procedural instructions that teach Ai agents how to perform specific, multi‑step tasks on demand. ↩

  5. These terms, and the em-dash, are commonly used in Ai generated articles. ↩

  6. Example: a YouTuber documenting their experience — YouTube video ↩

  7. Tech companies are blaming massive layoffs on AI — The University of Sydney (17 March 2026) — The timing and framing of layoffs attributed to Ai warrants closer examination: corporate restructuring, post‑pandemic over‑hiring, and investor pressure are all factors alongside genuine AI advances. ↩

  8. T‑shaped skills (Wikipedia) ↩

  9. Ethan Mollick — X post — Ethan Mollick’s observation that “prompt engineering” may not remain a distinct specialised role. ↩