The dangerous comfort of AI — a growth founder's guide to when AI weakens rather than strengthens

On Nordic Business Forum’s stage, innovation strategist Shawn Kanungo raises one of the AI era’s biggest blind spots: the illusion of progress. From there the 51-minute talk drills into a question growth-company leaders do not yet weigh enough: when does AI weaken a company instead of strengthening it?

The question is more important than it first sounds. We have, over the last two years, seen how every Finnish growth company is building its AI strategy, adding AI to workflows and rolling out copilot-grade tools. This is mostly good, and many benefits already show below the line. But Kanungo raises what the conversation is quiet about: beneath all of this another phenomenon is taking shape that may slow the company’s growth precisely at the moment AI looks like it is helping most.

▶

Source: Nordic Business Forum — Shawn Kanungo: The Dangerous Comfort of AI, 20 May 2026

What the illusion of progress is about

Kanungo’s core observation is simple but uncomfortable. As AI handles more and more small decisions, the human’s decision practice shrinks. As AI writes the emails, the human’s writing muscle thins. As AI plans the customer visit, the human’s ability to notice the wrong moment for the wrong remark does not get sharpened. This does not show in a single workday — on the contrary, AI looks to have made the day more efficient. But over a year and ten years, the question becomes structural: the people who use AI the most learn the least from it.

This is a phenomenon already documented in research, which psychologists call automation complacency. Its classic example comes from aviation: as autopilot became widespread, pilots’ manual-flying skills weakened, and it showed up in several accident investigations. In the AI era the same phenomenon is moving into knowledge work. As copilot writes the code, the developer’s skill in writing complex code themselves weakens. As the AI model drafts the pitch, the founder’s skill in selling their own story weakens. Kanungo’s message is that this is neither good nor bad — it is a structural consequence that must be taken into account.

When AI weakens a growth company

The question is: at which concrete points should the growth-company leader be careful?

**The first is the founder’s own decision-making.** When the founder begins to ask AI for an opinion on every decision before making it, they slowly move into a state where they no longer have their own intuition for new situations. This is especially dangerous for the seed-stage and growth-stage founder, whose whole work rests on the ability to make decisions on imperfect information quickly. I have written earlier in Innohub about Perttu Pölönen’s message that AI changes the ability to trust oneself in a way not yet visible — this is the same observation in the founder context.

The second is the customer encounter. When the sales team starts using AI for customer correspondence, the result improves at first: mistakes drop, the answer is better worded, the customer receives replies faster. But a year later the seller has lost part of their ability to read the customer’s voice between the lines — because they have not written those lines themselves. This shows up sooner or later in deal quality: key customer relationships that previously rested on personal communication start to feel more distant. The customer switches supplier and the seller cannot explain why.

The third is product-development direction. When the product team uses AI to analyse user feedback, statistically strong guidelines emerge quickly: “60% of users want X”. This is useful, but underneath often lies only the weight of the chosen word — AI has classified the feedback one way, but another classification would have given a different result. A good product manager reads individual feedback themselves — and that skill weakens as it becomes slower than reading AI analysis.

What for the Finnish founder

For the Finnish growth company, Kanungo’s message is especially timely, because the transition to AI use has, in the last twelve months, been faster than many expected. I have written earlier in Innohub about Tero Ojanperä’s message that agents are the next wave and are being built into production. This is true — but alongside it, a conscious practice is worth building around the places where the human’s own practice is kept up, so that the comfort caused by automation does not build too deeply.

Concretely this means three practices a growth-company leader can build alongside their AI strategy.

First, define places where AI is not allowed. This sounds contradictory to an AI strategy, but it is not. In practice I mean things like the next quarter’s strategy paper being written by humans, key-customer personal contact being handled in the seller’s own words, and the founder’s important decisions being slept on without AI. These are conscious exceptions that keep up the muscle on which the company succeeds in the longer term.

Second, measure AI’s net result regularly. Most companies’ AI adoption is measured through the benefit side (speed, accuracy, cost savings) but not the cost side (skill atrophy, slower staff development, growing customer churn). This is an imbalanced measurement that gives AI better values than it deserves. Kanungo’s message would be that measurement should be net — both benefits and costs together.

Third, practise slow application of AI regularly. Once a quarter make an important decision without AI’s help. Annually run one big project where AI is excluded. This is not just practice — it is a structural way to verify that the company’s ability to operate without AI still exists. If you notice it does not, AI has already moved from comfort to dependency, and the next stage is already dangerous.

What to take away

Shawn Kanungo’s Nordic Business Forum talk is an unusually needed counterweight to the AI hype that has, over the last two years, filled both media and executive conversations. His message is not that AI is bad — he is himself an innovation strategist building companies around this very topic. His message is that AI is too powerful a tool to be left without measuring both its sides.

A summary for the Finnish founder: build alongside your AI strategy a parallel structure that keeps up the human’s own practice. It can be simple — like “Friday mornings I write all my emails myself, without AI support” — but it must be a conscious decision. The comfort that accumulates over years does not show in a single workday’s statistics. It shows only when the company’s key customer switches supplier and the sales organisation cannot explain why. By then it is too late to fix, unless the structure has already been kept up.