A Norwegian board does not need an agreed date for artificial general intelligence to make useful decisions today. It does need to know which assumptions about its business would fail if capable digital work became substantially cheaper, faster or more widely available.
I would organise the discussion around exposure and options. What could change our customers' willingness to pay? What could make our own delivery model uneconomic? Which investments remain useful across several futures? That gives the board a strategy conversation it can revisit when evidence changes.
Keep three different uncertainties apart
Technical capability, reliable deployment and commercial adoption move on different tracks. A system that performs well in a demonstration may still be unsuitable for a Norwegian company's language, data, customers or responsibilities. Conversely, a modest improvement embedded in ordinary software can change purchasing habits without a dramatic scientific breakthrough.
The NIST AI Risk Management Framework treats trustworthiness as a consideration across design, development, use and evaluation. It is voluntary guidance, not Norwegian law. For strategy, I draw a practical lesson: capability announcements should prompt investigation, while operating evidence should determine commitment.
Superintelligence belongs in this discussion as an uncertain scenario, not an agreed forecast. Neither its timing nor its business consequences should be presented as established facts. A speculative timeline is a weak foundation for an irreversible investment.
Three scenarios to test against the same business
Gradual improvement. AI assists more tasks, but people remain the main constraint on reliable delivery. A Norwegian professional-services company benefits from better document preparation and more consistent follow-up. Its priorities remain data quality, work design and employee competence.
Faster commoditisation. Clients can obtain routine analysis or documentation at much lower cost. The same company must explain why customers should pay for its judgement, responsibility, relationships or integration into their operations. Selling saved hours at the old hourly rate becomes a fragile plan if buyers change how they purchase.
Constrained access. More capable services exist, but cost, supplier restrictions, outages or contractual requirements make access uneven. The company's ability to retain its knowledge, operate a fallback and choose acceptable suppliers becomes strategically important.
These are my hypothetical scenarios, not probability estimates. They can coexist in different parts of one business. The board should resist choosing the most exciting one and treating it as the plan.
Watch signals that can change a decision
Ask management to track a small set of business signals: changes in actual tender requirements, customers completing work themselves, price pressure on routine services, independently tested task quality and the cost of supervising a completed workflow. Include supplier concentration and access restrictions where they matter.
Define what would trigger action. For example, a hypothetical firm could investigate a new service model after several customers request fixed-price outcomes instead of hourly delivery. That is a reason to test the proposition, not proof that the entire market has changed. A trigger should lead to a named decision and an accountable owner.
Buy options before making large bets
Some preparation is useful across the scenarios: documented processes, controlled access to company knowledge, portable source material, workforce learning and the ability to measure outcomes. Those investments can make a later AI decision easier even if model progress slows.
Other commitments depend heavily on one forecast: a long exclusive contract, a wholesale removal of review capacity, or replacing a functioning service before an alternative has been tested. Ask which assumptions justify the commitment and how the company would respond if they fail.
A practical annual board exercise is to choose one revenue stream, examine all three scenarios and identify one reversible experiment. Revisit the assumptions sooner if a defined market or operating signal appears. Management should connect the exercise to budgets and customer evidence; a separate “future of AI” presentation can otherwise become theatre.
For the immediate decision, use the board strategy mandate. To preserve flexibility, test the supplier exit plan. Preparation should increase the company's room to act, rather than force certainty where none exists.
Sources and scope
Sources checked on 11 October 2026. The scenarios, signals and exercise are my analysis, not predictions or documented company outcomes. No probability or timetable for more advanced AI is asserted.