SMOC.AI's Post — 5 Minutter: AI og utviklere

Kristoffer Kvam · article · 2026-04-10 source ↗ Reader 🖥️

AI as Amplifier AI Labor Disruption Learning Culture & Slack

AI amplifies senior developers' effectiveness but breaks the traditional junior developer training pipeline, because the simple tasks that served as the learning on-ramp are being automated away — and juniors guiding AI-juniors compound weaknesses rather than building competence.

practical · narrative — Practical: argues for a specific AI workflow (spec-first, then code). Narrative form: first-person account of building an AI-driven company, using personal experience as evidence. Part of a serialized LinkedIn content series ('5 Minutter').

Structure

Part	Role
Series framing: documenting the journey to a 100% AI-driven company	Context — establishes credibility and stakes; this is a practitioner report, not theory
AI capability assessment: extremely productive junior developer	Core metaphor — frames the entire argument about who benefits and who doesn't
The effective workflow: spec → architecture review → autonomous execution → self-validation	Prescriptive evidence — shows what working well with AI actually looks like
The junior training crisis: simple tasks disappearing	Core controversial claim — the consequence of AI eating the bottom of the task stack
Generalization beyond software: economics, design, legal	Scope expansion — claims the pattern is universal across knowledge work

Key terms

AI-junior (AI as junior developer) AI Labor Disruption
AI is extremely productive at starting and executing tasks but produces low-quality code without strong guidance. It needs architectural direction, specification, and feedback loops — exactly what a junior human developer needs, but at much higher speed and volume.
evig loop av iterasjoner (eternal iteration loop)
The failure mode when AI works without clear specification: it keeps iterating toward 'ok but far from ideal' output, burning time without converging on quality. The loop is the symptom of missing architectural guidance.

Propositions

explicitAI is most precisely understood as an extremely productive junior developer that requires heavy guidanceconf: high
explicitcontestedSenior developers and technically strong product people get superpowers from AI; juniors do notconf: high
contradictsMartin Fowler & Kent Beck: Frameworks for reinventing software, again and again · p5 — Beck calls AI the golden age of the junior programmer; SMOC argues juniors get no superpowers because a junior guiding an AI-junior compounds weaknesses.
contradictsThe Future Of Software Engineering · p5 — Thoughtworks claims AI makes juniors more valuable by shortening the net-negative phase; SMOC claims only seniors gain superpowers while juniors do not.
contradicts11) AI-spesial med Vilde Opsal og Anders Sveen · p13 — SMOC asserts the plain 'seniors get superpowers, juniors do not' narrative that the Heis panel explicitly disputes, arguing seniors are not uniformly winners and agent-fluent juniors will overtake them.
explicitThe most effective AI workflow is: specification first → architecture feedback → autonomous execution → self-validationconf: high
explicitWhen a junior guides the AI-junior, they compound each other's weaknesses and the result suffersconf: high
explicitdevelopedTraditional junior developer training is breaking down because the simple entry-level tasks are disappearingconf: high
refines ←17) Jonas Slørdahl Skjærpe om svinghjuleffekter, liminalfaser og kunsten å ta heis · p17 — Skjaerpe accepts that traditional junior training is breaking down but insists a staircase into the profession must exist, reframing the question from whether juniors exist to how they learn well.
explicitHaving a senior supervise the junior-AI pair wastes the senior's time on what used to be valuable training work but now just slows progressconf: high
explicitThis dynamic is not limited to software — it will appear in economics, design, and legal servicesconf: medium
implicitThe value of junior training was always partly in the simple tasks themselves, not just in the mentoring relationshipconf: high
implicitSpeed of execution is the primary metric that matters in a startup context — quality of junior training is secondary to shippingconf: high

Arguments

Traditional junior developer training will become unviable because AI eliminates the simple tasks that were the learning on-ramp strength: medium

Explicit premises:

AI can now do the simple tasks that juniors used to do
These simple tasks were the entry gate to building experience
In time-critical environments, using seniors to supervise junior-AI pairs is a net slowdown

Implicit premises:

There is no alternative way to train juniors other than through progressively harder real tasks — classroom learning, pair programming on complex tasks, or AI-assisted learning are not considered
The startup context (time-critical, every decision matters) generalizes to the industry — what's true at SMOC.AI is true everywhere
Junior developers cannot learn by reviewing AI output or by using AI as a teaching tool — the relationship is assumed to be one-directional (junior guides AI) rather than bidirectional (AI teaches junior)

Inferences:

If simple tasks disappear as a training mechanism, the industry faces a pipeline crisis: where do future seniors come from?
The 'AI as junior dev' metaphor may be self-defeating over time — if juniors can't train, there will eventually be no seniors to guide the AI
Companies that figure out alternative junior training models will have a structural advantage in talent development

The optimal AI development workflow requires senior-level architectural judgment upfront strength: medium