Norway wealth fund moves towards some AI-driven decisions but with humans in control

By Gwladys Fouche

OSLO, March 24 (Reuters) – Norway’s $2.1 trillion sovereign wealth fund, the world’s largest, will eventually allow some investment decisions to be made by AI systems under human supervision, but not yet, as ‌the tools still make errors, fund officials said on Tuesday.

At present, around half of the 700 ‌employees at Norges Bank Investment Management code their own AI tools using Anthropic’s Claude large language model, according to Stian Kirkeberg, the fund’s head ​of machine learning and AI.

Staff primarily use these tools to gather information to help them make decisions, Kirkeberg told a fund seminar on AI.

This ranges from monitoring the 7,000 companies the fund invests in for ESG and financial risk, to simulating a contract negotiation or preparing for company meetings.

Kirkeberg said that in time, some AI agents will be allowed to make ‌limited decisions autonomously.

“The principle is that we ⁠make better human decisions by getting AI to analyse it for us,” he told Reuters after the seminar.

“At some stage, we’re going to trust that the agent can make some ⁠of the decisions and we just monitor what it does,” Kirkeberg said.

He added that the fund was moving toward that approach but was not applying it yet, emphasising that human oversight would remain essential.

FIRMS IGNORING AI ARE ‘COMPLETE MORONS’

Chief Executive Nicolai Tangen has ​been ​a vocal supporter of using AI both internally and in ​the companies the fund invests in, once describing ‌firms that fail to adopt the technology as “complete morons”.

He said the wealth fund, which manages Norway’s oil and gas revenues for future generations, is not under the same pressure as short-term investors to automate investment decisions.

“You have investment firms which have automated investment decisions … We’re not doing that. But we are also not a high-frequency trader, … we are a long-term investor, so it’s a bit different”, Tangen told Reuters.

One exception is the fund’s use of AI to analyse ‌when to trade or not, helping to reduce transaction costs.

FUND SEES ​BIG COST SAVINGS

Tangen said the fund had invested “millions of crowns” in ​AI and returned benefits “in the billions”, without giving ​specific figures or a timeframe.

He expected the headcount to remain steady at around 700 across ‌its offices in Oslo, London, New York and Singapore, ​but roles would shift as ​a result of AI toward front-end investment from back-end administration.