Smooth-talking generative artificial intelligence (AI) bots such as the in-vogue ChatGPT model could “revolutionise” the funds management industry, according to Dutch quant investment specialist, Robeco.
In a new think-piece, the Robeco human quartet of Clint Howard, Tim Vogel, Thom Marchesini and Mike Chen, argue the rapidly emerging generative AI systems have the potential to augment portfolio construction, fast-forward research, improve marketing and – perhaps most importantly – write emails.
“A key component in a quantitative researcher’s toolkit is the ability to quickly find and synthesize information to aid in making investment decisions. GPT can act as an accelerator for this phase of the alpha research process,” the paper says.
“… In research, ChatGPT can be employed in scenario analysis for quick surveys on bespoke topics… As an integral component in a quantitative researcher’s toolkit, ChatGPT can accelerate the information-gathering and synthesis process, thereby enabling investment professionals to make more informed decisions.”
But the real productivity gains from ChatGPT et al will likely come from reducing time spent on banal tasks such as drafting monthly investment reports, client engagement letters and emails.
“Composing emails can cost time and energy,” the Robeco report says, with generative AI able to ease the pain – albeit with some associated risks, especially for external communications.
“Errors in such [external] messages can have more severe consequences for the reputation of the company and the relationship with the intended recipient, and at the same time might well be more likely because of GPT’s confident tone and fluency,” the paper says.
Generative AI might also save time and energy by reading the emails, too, but the Robeco paper does not consider this point.
For all its blather-producing powers, ChatGPT does have some downsides including a deficit of “nuance and detail” and an over-familiar tone that sounds more believable (to humans), which combines with a dangerous tendency to occasionally tell lies.
“Finally, as the asset management industry is a highly regulated industry, the use of GPT faces challenges concerning compliance and regulation,” the report says. “Ensuring the confidentiality of sensitive information is paramount, as asset managers often handle confidential client data and proprietary investment strategies.”
Despite the risks, Robeco says fund managers should experiment with generative AI technology given the potential vast productivity gains.
And, in a highly speculative statement, the report says that the growing dominance of machines might even make the remaining humans in the investment industry… nicer.
“For the asset management industry, as machines become more human, one potential implication is that humans can become even more human. That is, people working in the asset management industry must focus more on compassion, creativity, specialized knowledge, and sound judgment,” the paper says.
“For those in creative roles, such as quant researchers, portfolio managers, equity analysts and marketing professionals, creativity is something that generative models cannot replicate entirely, and therefore will continue to stay valued. While GPT can certainly prompt creativity, it does not (yet) have an imagination of its own.”
Financial advice may also be fully roboticised with generative AI, Robeco suggests, after generally disappointing results from the previous robo-advice hype cycle.
“Furthermore, any individual can query a GPT-based application for tailored and specific financial advice. Financial advice prompt templates could be provided to a user, who would then have full ownership and responsibility over how they use the advice that emerges from the GPT-based application,” the report says. “Similar to the explosion in retail trading and the advent of trading platforms like Robinhood, will GPT bring even more options to retail or individual investors without a large professional organization supporting them?”
The Robeco paper coincides with a couple of early-stage reports that show ChatGPT can outperform humans in interpreting Federal Reserve-speak and predict equity prices based on media headlines.
More to come…