From NATO frontline missions to global policy and the boardroom
Artificial intelligence can process information, identify patterns and help people understand complex situations. But when decisions carry significant consequences, a critical question remains: Where does decision support end, and decision authority begin?
This question was at the heart of When the Stakes Are Real, a joint keynote by Aaron Kalvani i and @Rune Køhler at the 4th Data Science & AI Summit in #London on 1 October 2026.
Bringing together perspectives on AI, operational decision-making, global policy and enterprise leadership, the presentation explored a challenge that extends far beyond technology: how can people make sound decisions when information is incomplete, circumstances are changing and accountability matters?
The challenge is not simply having more data
In complex environments, decision-makers rarely have the luxury of waiting for perfect information. Evidence can come from multiple sources. Situations can change quickly. Different stakeholders may have different perspectives, and decisions often involve trade-offs that technology alone cannot resolve.
The presentation highlighted the importance of understanding the relationship between information, judgement and accountability.
More information does not automatically produce better decisions. What matters is how that information is interpreted, placed in context, and translated into action.
For organisations adopting AI, this distinction is becoming increasingly important.
A decision-making framework built around people
One of the frameworks featured in the keynote was the OODA loop:
- Observe: Understand what is happening and gather relevant information.
- Orient: Interpret the information in context.
- Decide: Exercise judgement and determine a course of action.
- Act: Put the decision into practice.
The cycle then continues as new information emerges.
In AI-enabled environments, these stages raise important questions. What information is the system presenting? How reliable is the resulting picture? Who interprets its recommendations? And who has the authority to act?
AI may strengthen parts of the process, but the relationship between information, judgement and action remains critical.
When the picture is incomplete
The keynote used real-world case studies to illustrate the challenges of decision-making in complex situations, including flooding and landslides in Costa Rica, two Air India crashes in the Alps, and infrastructure failures in Ukraine.
These examples were presented through the lens of incomplete or changing information, operational disruption and the consequences of decisions made under pressure.
The broader lesson is relevant well beyond the environments discussed in the presentation.
Whether an organisation is responding to a disruption, managing operational risk or deploying AI into a critical workflow, decision-makers need to understand the limitations of the information available to them.
A system can present a picture of a situation. That does not necessarily mean the picture is complete, nor does it remove the need for human interpretation.
From global policy to the corporate boardroom
The presentation also connected operational decision-making with international engagement, stakeholder dialogue and executive accountability.
For businesses, this connection matters as AI becomes more involved in analysis, recommendations and operational processes.
Effective AI governance cannot stop at defining what a system is permitted to do. It must also address who is responsible for interpreting its output, making consequential decisions and explaining the results.
Three elements are particularly important:
Information: Are decision-makers working with relevant evidence while understanding its limitations?
Judgement: Are context, competing priorities and appropriate decision-making authority being considered?
Accountability: Are decision owners clearly identified, with appropriate records and audit trails?
These questions bring AI governance out of the abstract and into the practical realities of organisational leadership.
The question every AI leader should consider
As AI systems become more capable of shaping the information on which people rely, organisations must think carefully about the boundary between assistance and authority.
An AI system may recommend a course of action. A human may review that recommendation. An organisation may establish rules governing when it can be followed.
But these arrangements need to be explicit, particularly when decisions have significant consequences.
For business leaders, the task is not simply to decide where AI can be introduced. It is to determine how decisions will be made, who remains accountable and how trust can be maintained when the stakes are real.
The future of AI will depend not only on the intelligence of our systems, but also on the quality of human judgement surrounding them.
Continuing the conversation
At Data Science Week, we bring together perspectives from across technology, industry and leadership to examine the questions shaping the future of AI and data science.
Our next European edition, the 5th Annual Data Science & AI Summit Europe 2027, takes place in Frankfurt, Germany, on 24–25 June 2027.
Under the theme From Data to Automation: Enterprise AI, Data Science & NextGen Transformation, we will continue exploring how organisations can move from AI experimentation towards effective adoption, responsible decision-making and meaningful business impact.
A question to leave you with: If AI increasingly constructs the picture from which we decide, where does decision support become decision authority?
We welcome your thoughts in the comments.
Interested in speaking, attending or exploring partnership opportunities? Connect with The People Events.