"Does AI challenge fundamental trust?" was the title of my article for the renowned Finans/Invest 3/2025. The resumé text was:
Trust is the foundational element of democracies and of our financial systems. Trust involves, among other things, acting in the best interest of others (fiduciary duty). This applies, for instance, to how pension funds manage the savings of their beneficiaries, or to the duties boards of directors owe to their companies and shareholders. But the development and rollout of AI is moving so fast that we risk eroding our shared understanding of reality. Ultimately, this may affect trust itself — and with it, for example, financial stability. Disengaging from the use of AI would be too costly, as it offers overwhelming gains in efficiency and innovation.
The year 2025 will likely mark the point where we begin to allow AI agents to take initiatives on our behalf. If these AI employees are to learn from raw data that includes mis- and disinformation, we risk elevating falsehoods to self-reinforcing truths. The current political decoupling from the United States may also play a role here, as most AI models remain American.
The article was first and foremost a follow-up to my article on AI two years ago. But the article also expresses my deeper concern for how political motives and rammed through development can erode the value of AI.