AI and Bias: Understanding, Implications, and Ethical Design

On the first day of Entrepreneurship Week we welcomed the panel “AI and Bias: Understanding, Implications, and Ethical Design.” The discussion moved beyond theory and into the very real, practical decisions being made by teams building AI products today.

The panel brought together Petra Dalunde, representing the Swedish AI Factory Mimer at RISE, and Laurynas Adomaitis, who leads ethics-by-design work connected to Umeå University. They unpacked how bias emerges in AI systems, why it often goes unnoticed, and what responsible teams can do about it and presentet Mimer – the Swedish AI Factory that provides compute, expertise, data, test and validation and ethical by design to SMEs and startups

Published:

Jan 29, 2026

3 min read

From Entrepreneurship Week 19-23 Jan 2026
Day One Panel Highlights

Bias Is Not an Edge Case — It’s a Design Choice

A key message from the panel was that bias in AI is rarely accidental. It is often a direct result of what we optimise for.

As Laurinus explained:

“When an agent, a conversational agent chats with the users, it has this kind of feedback metric. So it self evaluates whether it’s doing good or not. And usually that feedback metric is engagement. So does the user keep responding to this or not?”

When engagement becomes the primary success measure, systems can unintentionally reinforce addictive behaviour, confirmation bubbles, or overly personalised responses — not because the AI is “evil,” but because it is doing exactly what it was trained to do.

“People want to make addictive, engaging chatbots. Sometimes it’s just fun — sometimes it’s a problem.”

Ethics Depends on Context

One of the strongest insights from the discussion was that ethical AI cannot be one-size-fits-all.

“To me, ethics really needs to be in context. What are you making, and what is it supposed to do?”
— Laurinus

A chatbot designed for entertainment has very different ethical implications than one used for healthcare, education, or decision-making. The panel emphasised that responsible AI starts with clearly defining purpose — and aligning metrics, design, and incentives accordingly.

Ethics by Design: From Checkbox to Practice

Through Mimer, ethics is not treated as a compliance afterthought but as a built-in process. Laurinus described a structured yet accessible approach:

Short, practical ethics trainings for teams building AI
One-on-one ethics evaluations during product development
Progress-based assessment rather than “pass/fail” judgement

Petra highlighted how this process often unlocks new ways of thinking, especially for highly technical teams:

“When very technical people begin to think in these other areas, something interesting happens. They start to mix different competences and ask better questions.”

Mimer as support for AI startups and SMEs

Petra explained that the EU Commission is launching 19 AI Factories and 13 AI Factory Antennas all over Europe to strengthen European sovereignty and competitiveness by supporting AI startups and SMEs.

From an AI Factory or Antenna you can access compute, expertise, data, test and validation and much more for free, if you are a startup or SME. Mimer supports the startups and SMEs through development, testing and deployment.

The best way to come in contact with Mimer is through https://mimer-ai.eu/ or emailing petra.dalunde@ri.se .