As money managers look for ways to leverage artificial intelligence in building portfolios, the generative technology is also making its way into climate modeling for physical risks.
“On the physical climate side, there’s a lot of movement towards emulation of climate models, bringing in some of the catastrophe models,” said Matt Goldklang, climate scientist at the Boston-based quantitative equity manager Man Numeric. “I say emulation is basically take all the physics and say ‘OK, physics doesn’t really matter anymore, let’s just throw at it and see what we can re-create.’”
At an April 17 panel at the BloombergNEF Summit in New York, Goldklang noted that the climate science community tends to be “a little wary” because it tends to prefer physical-based processes. “But there’s a slow marriage developing in which you’re enhancing the models with AI.”
At the firm — which is owned by the $167.5 billion alternative investment manager Man Group — Goldklang said artificial intelligence is “allowing us to use climate models more readily” and “increase what we call ‘downscaling,’” the procedure of using large-scale climate models to make predictions with higher quality in a more spatial scale.
Man Numeric also uses AI when looking at and considering what actions it can take in regard to adaptive capacity — or the ability to adjust to climate change to mitigate potential damages and make the most out of opportunities.
“I know we’re developing models based off of GPT technology to read company reports and understand what that adaptive capacity looks like,” Goldklang said. “Rather than hiring a team of 15 analysts to help us do that, we have a computer, which has been really helpful.”
“It’s really just becoming a very helpful assistant,” he added.
