Humans must be in the loop for workflows that use artificial intelligence, as that will build trust in A.I. at financial services firms, says P. McKay Marschalk Jr., CFA, vice president, director, reporting and distribution product management at FactSet. He took part in this FTF News video, which also features Pat Reilly, CFA, senior vice president, senior director, middle office – Americas, FactSet.
Pat Reilly
“This is a newer technology, right, and people, I think, are going to be naturally distrustful of it. So, what we really are seeing is how clients are incorporating A.I. into their workflows,” Marschalk says. “But also allowing for humans in the loop, editing, checking, and approvals, right. And that’s where I think that our solution really fits that bill. … When commentary gets generated, it gets put into a rich text editor that allows a human to go in and look at that data, look at the commentary that was produced, and maybe that’s their starting point, maybe that’s their rough draft. That gets them 80 percent of the way there.”
Even so, there will be more human-based approvals and a need for checks and balances to make people comfortable with the use of A.I., Marschalk says.
P. McKay Marschalk Jr.
Reilly adds that he has been reminding clients who have concerns about A.I.-based hallucinations is that FactSet’s Portfolio Analysis application has Portfolio Commentary, which creates “AI-generated explanations” of performance attribution analysis. “What I love to point to, and Commentary is that example, is that ‘prove it,’ or that audit ability functionality, where we’re not going to present a number that you can’t validate,” he says. For instance, if an Apple or Nvidia stock jolt causes a portfolio to spike, “we can prove all of that out, and that puts clients at ease.”
The video interview also covers:
- FactSet’s major investment in evolving client reporting solutions;
- How FactSet is optimizing client workflows;
- Why FactSet makes data available via multiple channels;
- How investment firms can have bespoke custom reports and efficient, automated output;
- Why client reporting teams need links to internal risk, attribution, and performance groups;
- How client reporting teams facing budget and headcount cuts can do more with less; and
- How FactSet learns from each client’s challenge.
Leave a Reply