Don't Panic-Invest in AI in 2026

This article was originally published in Dagens industri (Sweden's leading business newspaper) on January 14, 2026.

According to fresh data from Statistics Sweden, AI adoption among Swedish companies increased dramatically during 2025. As usual, we can look across the Atlantic for a preview of what's to come. A North American study from RSM shows that this trend is unlikely to slow down. AI adoption is now nearly universal among mid-sized American companies – 91% use AI in some form. Yet another widely-noted study from MIT reveals that only five percent of all these initiatives deliver clearly measurable business value.

For me, it's difficult to grasp how leadership teams can choose to accept this gap between ambition and impact, and how both government agencies and mid-to-large companies can continue pouring money into something they clearly don't fully understand.

Common mistakes in AI implementation

When a company sets out to implement AI, one of two mistakes is often made: Either they bring in expensive consultants to conduct workshops, interviews, and surveys, or they run phone interviews that typically have low response rates. According to Pew Research, response rates for phone interviews in opinion polls have collapsed from 36 percent in 1997 to just six percent in 2018. Various weightings can be applied to compensate for groups whose voices weren't heard, but it doesn't matter if we miss certain information entirely because those who possess it cannot share it.

The result is a skewed picture and a flawed decision-making foundation. We feel the trunk and think the elephant is a snake, or the leg and think it's a pillar. No one sees the whole, and therefore it doesn't matter how good the solutions we develop are or how much we spend building them.

A better path forward

So how do we create a more grounded understanding and analysis before we start inventing solutions? According to a research article published this spring in the scientific journal "Computers in Human Behavior: Artificial Humans," people are just as willing to share sensitive information with an AI agent as with a human. By using AI to conduct deep, qualitative interviews at scale, we can now do what was previously practically and economically impossible. We can listen to all hundreds or thousands of employees in a week and understand where the bottlenecks in the organization's work actually exist.

For Swedish companies, a great opportunity emerges. We already have a strong culture of listening and consensus-building. An approach where AI strategy is grounded in the entire organization's perspective – rather than management's guesswork – should align closely with Swedish business DNA. While the rest of the world continues guessing, Swedish companies can lead the way by building their AI implementation on understanding and evidence-based insight.

Three steps for decision-makers in 2026

1. Prioritize process understanding over technology purchases

Stop panic-buying tools or hastily building integrations before you've mapped out where the shoe actually pinches.

2. Listen to all employees

The real knowledge of how a company works resides with all the employees who perform the work daily, not just the loudest voices. Use the technology available to scale up qualitative conversations so you can dig into hidden insights and concealed workflows.

3. Use modern technology instead of surveys and feasibility studies

Written forms favor those who are good at expressing themselves in writing and those who have time to spare. Voice lowers the barrier to sharing, but traditional phone interviews are difficult to scale, and follow-up questions require the ability to connect many different data points to isolate the right question to ask at the right time. This is precisely what AI excels at. Through AI interviews, it's possible to hear everyone inclusively and ask the probing questions that reveal hidden insights.

The question for 2026 is not whether we should use AI, but whether we can afford not to understand the challenges AI can help us solve – before we start buying, building, and implementing.

Organizations that continue building solutions disconnected from reality will be overtaken by those who dare to listen deeply and build only afterwards.