Why Wall Street's most data-obsessed investors are taking it slowly with generative AI

• A new Bloomberg survey of quant investors found that 54% of respondents do not use generative AI for investing.
• AI usage in quant investing will be linked to data formatting and availability going forward, Bloomberg believes.
• This aligns with the thinking from many quant executives, who have said that AI is not yet able to beat the market.

Generative AI hasn't yet won over the quants.

A majority of people running systematic trading strategies at top-tier asset managers have "not yet begun their generative AI journey," according to a new Bloomberg survey.

The data giant interviewed 151 quants between April and November last year to determine how they've integrated generative AI tools into their investment research process. While this section of the investing world has used machine-learning techniques for years, generative AI has not yet broken through. Most of the respondents, 54%, do not incorporate it into their workflows, the survey found.

This matches the vibe Business Insider reported on in October, where quants at a London-based conference were skeptical of the technology's ability to beat the market and add value to their investing processes. A UBS executive, for instance, said that AI is not going to help win the "alpha war."

Bloomberg's rationale for the slow adoption is focused on data formatting and structure.

Angana Jacob, the firm's global head of research data, said quants require their data to be cleaned and structured a specific way because of the complex systems their strategies run and the amount of capital at stake if there is an error.

"They're working in a very controlled research environment, models need to be explainable, models need to be repeatable," said Jacob, in an interview with Business Insider.

The work required to get the datasets to a point where they can be used is "unglamorous," but "foundational," Jacob said. Her team is creating data products for quants that might increase AI adoption among them, because, she said, the enthusiasm is high for what AI can do once the data catches up. The lack of widespread use in the investing process is more a sign of the diligence of these players.

"It's a good thing, it shows their caution," she said.

Bloomberg is not the only data player that has identified this issue. Former Point72 data executive Kirk McKeown's startup, Carbon Arc, is also focused on structuring datasets for easier ingestion into artificial intelligence models.