You might think that banks like J.P. Morgan have no problem in attracting exemplary talent in artificial intelligence (AI). After all, the U.S. bank successfully hired Geoffrey Zweig, a top natural language processing (NLP) scientist from Microsoft as head of global machine learning in February. One month later it also recruited Jeffrey De Jong, a specialist in experimental particle physics who'd been working for King, the games development company behind the Candy Crush Saga. But for all the Zweigs and De Jongs who come in-house, it seems there are plenty of other top technologists who want to remain unfettered by a big corporation.
"Some of the best technology talent doesn't always want to work in banks," said Daniel Drummer, a VP in fintech at J.P. Morgan's corporate and investment bank. A lot of people want to be entrepreneurs, he added.
J.P. Morgan hasn't given up. It's still working with the world's top AI professionals - on their terms. Rather than assimilating them into its investment bank, it's inviting them into its offices and helping to develop their start-ups as much as it can and - notably - not necessarily investing or demanding anything in return.
"Our objective is not necessarily to acquire or invest in [a company we work with], but to make it industry-ready," said Drummer, speaking at the AI Summit in London today. Instead of investment, Drummer said machine learning start-ups in the wholesale banking space often need access to data and knowledge about how the finance industry works, along with an enterprise to test their ideas out on. This is what J.P. Morgan offers (although it's not averse to investing too if appropriate).
Skittish machine learning technologists seem to like this deal. J.P. Morgan launched its "In Residence" program for fintech start-ups in June 2016. Since then, Drummer says it's met with well over 100 companies using data science, machine learning and AI. Of those, it's chosen to work with a "handful."
Drummer didn't say who those firms are, or exactly what they're up to at J.P. Morgan. The bank has said publicly, however, that it's using artificial intelligence to automate tasks done by its lawyers and to identify opportunities for its investment banking business.
Some of the forthcoming AI initiatives coming out of J.P. Morgan might therefore originate with interesting companies brought under the bank's wing on a relatively informal basis.
What if the AI companies J.P. Morgan nurtures go on to become successful and to tout their wares to competitor banks? Drummer suggested this is just anachronistic thinking: "It's not about being isolated and hoping that no one will find out what you're doing." The future is about companies working collaboratively on AI projects said, Drummer, pointing to Google's approach with TensorFlow.
This may be so, but it also seems that J.P. Morgan's In Residence programme gives it the best of both worlds. It gets to pick top machine learning talent with no commitments either way, and if things work out, JPM can be the first to invest.