The world's largest sovereign wealth fund has half of its employees coding in Claude, the next step is to let AI autonomously make investment decisions

According to 1M AI News monitoring, around half of the employees of Norway's Government Pension Fund Global, which manages $2.1 trillion in assets (Norges Bank Investment Management), have been using Anthropic's Claude large language model to independently develop AI tools. During an AI seminar, the fund's Head of Machine Learning and AI, Stian Kirkeberg, explained that employees primarily use these tools to assist in decision-making, including monitoring ESG and financial risks of the fund's 7,000 investee companies, simulating contract negotiations, preparing for company meetings, and more.

Kirkeberg stated that in the future, some AI agents will be allowed to autonomously make limited decisions under human supervision. "The principle is to have AI do the analysis for us so we can make better human decisions," he said. "At some point, we will trust the agents to make some decisions, while we only monitor their behavior." He emphasized that this model has not been implemented yet, and human supervision remains essential.

The fund's CEO, Nicolai Tangen, has previously referred to companies that do not adopt AI as "complete morons." He mentioned that the fund has invested "millions of kroner" in AI, with returns reaching the "billions of kroner" level. The fund is currently using AI to analyze trading opportunities to reduce transaction costs. The staff size will remain around 700 people, but positions will shift from backend administration to front-end investment.