The world's largest sovereign wealth fund plans to involve AI in investment decisions; human oversight remains indispensable.

Ask AI · What specific role does human supervision play in AI investment decision-making?

IT Home, March 24 news: The Norwegian government officially said on Tuesday that the country’s sovereign wealth fund—worth $2.1 trillion (IT Home note: at the current exchange rate, approximately 1.448 trillion RMB), the largest in the world—will ultimately allow artificial intelligence systems to make some investment decisions under human oversight, but it will not be implemented for now because these kinds of tools still produce errors.

This image is suspected to be AI-generated.

Stian Kjekkebøry, head of machine learning and artificial intelligence at Norges Bank Investment Management, said that of the fund’s existing 700 employees, about half are using Anthropic’s large language model Claude to write their own AI tools.

Kjekkebøry said at an AI symposium for the fund that employees currently mainly use these tools to gather information and assist with decision-making.

Use cases span multiple areas: monitoring environmental, social, and governance (ESG) and financial risks of the 7,000 companies the fund invests in, simulating contract negotiations, or preparing for company meetings.

Kjekkebøry said that in the future, some AI agents will be allowed to make limited decisions autonomously. In an interview with Reuters after the meeting, he said: “The core principle is that, through AI analysis, humans make better decisions. There will come a day when we believe AI agents can make some decisions on their own, and we only need to supervise their actions.”

He added that the fund is moving in this direction, but has not yet rolled it out, emphasizing that human oversight remains indispensable.

Fund CEO Nicolаr Tangen has long publicly supported the use of AI within the firm and at the companies it invests in. He once said that companies that don’t adopt this technology are “extremely foolish.”

He said the fund is responsible for managing Norway’s oil and gas returns to benefit future generations, and does not need to bear the pressure that short-term investors face regarding automated investment decision-making.

Tangen said: “Some investment firms have achieved automated investment decision-making… We haven’t done that. And we are not high-frequency traders; we are long-term investors, so it’s different.”

The only exception is that the fund uses AI to analyze the timing of trades to reduce transaction costs.

Tangen said the fund has invested “millions of kroner” in AI and has received returns of “tens of billions of kroner,” but did not disclose specific data or the time range.

He expects the total number of employees at its offices in Oslo, London, New York, and Singapore to remain around 700, but, affected by AI, the job mix will shift from back-office administration to front-end investment business.

For other managers pushing AI adoption within companies, he advises not to set explicit layoff targets.

Tangen said: “That would only trigger a lot of resistance. I think the goal should be to improve sales, profits, and operating efficiency—expand market share and do your core job better. That’s the more constructive way to roll out AI.”