BlackRock MD: People are the problem when it comes to machine learning
Forget siloed quant teams, overworked developers or the superiority of the human mind, there’s one big impediment to artificial intelligence really taking off in financial services – the current staff.
If AI is to gain traction in financial services, it will need to vanquish the army of financial services professionals standing in its way. At least, this is the conclusion of Raffaele Savi, a managing director and head of developed markets within BlackRock’s scientific active equities team.
“People are problematic, machines are easy,” he said, speaking at the Newsweek conference on artificial intelligence and big data last week. “People who understand finance and economics don’t think the same way as people who know a lot about computing and machine learning. We need to work together as an industry. AI is where the bulk of alpha will be on a longer term horizon.”
The biggest problem that financial services firms currently face is making the most of the vast swathes of third-party data being created every day. Asset managers and hedge funds need an informational advantage over the competition to gain an edge, and are looking beyond traditional sources of information like sell-side analyst reports and accounting data in order to do this. But most of these data sets are either too fragile, or need extra financial services domain expertise in order to be really exploited.
However, sitting around debating whether humans are better than computers is side-stepping the real issue, says Savi.
“The machines have already won. There is not a human trader alive who can compete against an algorithm,” he said.
Savi believes that hedge funds and asset managers are on the cusp of their “Pixar moment” when it comes to using AI.
“It took 20 years from Pixar being formed to Toy Story finally being released. 20 years of commitment without ever making a movie,” he said. “But two years after the first Toy Story, there hasn’t been another big animation movie made by humans again.”
"When data science and AI finally meet economic theory, the investment game will change for ever," he said.
When this finally hits, expect it to be bloody. For all the talk of using human expertise, job losses in financial services as a result of machine learning – particularly in asset management – could be brutal.
Boutique research firm Opimas predicts that there will be 230,000 jobs lost in capital markets as a result of AI technology by 2025. Around 90,000 of these are likely to be lost within asset management firms, it suggests.
Most financial services organisations are struggling to hire AI and big data specialists because of an extreme talent shortage. But these jobs are unlikely to directly replace the money management roles being eliminated anyway. Opimas predicts that technology and data providers will hire 30,000 people to deal with demand for AI products from financial services organisations.
Photo: Getty Images