The investment industry is not yet ready for full-scale artificial intelligence (AI) to replace high-level human decision making, according to a panel.
Speaking on a webinar hosted by Frontier, Melissa Anderson, AllianceBernstein vice president and head of taxable fixed income investment technology, said the industry was not ready for AI to be making decisions on what portfolios were being invested in.
“Where we are today is how do we elevate our talent to be working problems that are not yet solvable through technology? There are many of them,” Anderson said.
“There are many ways that technology supports those decisions but I’m not sure we’re ready pure high-level decision making, maybe smaller-scale rules-based investment decisions.”
David Jenkins, Frontier head of technology and operations, said: “In terms of neural networks and things like that, I don’t feel our industry is ready for those type of things because of the need for playability and predictability. Most of what we use is advanced statistics and machine learning”.
The value of data
Jenkins said he had found data to be a challenge as companies locked it up and turned it into a commercial enterprise.
“Something that you might have been using within your decision-making process suddenly now charges a few hundred thousand dollars to get access for something you previously had for free,” Jenkins said.
“We’re always juggling that balance and finding alternatives, sometimes even if the data is patchy. It’s a commercial reality I guess these days.
“An ever-expanding range of data requires an ever-expanding budget but it’s very hard to trace back the value of that data.”
Lilian Quah, Epoch portfolio manager and head of quantitative research, said it was a business problem.
“It’s something as an industry have some experience dealing with but not necessarily a robust framework for putting a dollar value on data,” Quah said.
“I have seen academic work which is quite sensible in providing a framework for how to do – essentially scoring data.
“Is it worth even the research time? Let along the dollar cost associated with it, but it’s something we’re going to struggle with because more data is going to be digitised and commercially available for a fee.
“It will be up to every organisation to come up with their own plan for tackling data efficiently and judiciously in terms of budget.”
Anderson said this had been never-ending problem that was not going away.
“It’s just how you decide what data to use and what is too much [and] it comes down to a philosophical approach your organisation takes to data,” Anderson said.
“Being at AllianceBernstein and having a robust large technology arm we can rely on for this and also having management in multiple different asset classes, we leverage each other quite a bit as well.”
