Early to Party? 리플 CTO Emeritus Reacts to Ex-SEC Chair Gensler's 2020 AI Paper
리플 CTO emeritus David Schwartzhas reacted to a resurfaced 2020 paper by former SEC Chair Gary Gensler, which pointed out that as deep learning in finance moves to a mature stage of broad adoption, it may lead to financial system fragility and economy-wide risks.
The discussion was revived by X user Andrew Curran, who believes that Gensler might have been "early to the party" in foreseeing the systemic risks posed by broad adoption of deep learning in finance.
According to Curran, Torsten Slok, chief economist at Apollo, warned that mass adoption of AI agents could trigger a bank run as they optimize user investments. He added that Gary Gensler, who served as SEC chair for the last few years, warned that the same issue may arise in the markets: the pursuit of algorithmic perfection at scale destroys system stability.
In November 2020, Gary Genslerco-authored a paper titled "Deep Learning and Financial Stability" with Lily Bailey at the Massachusetts Institute of Technology (MIT) Sloan School of Management before becoming SEC head, which Curran shared alongside his tweet.
The paper noted at the time that the financial sector was entering a new era of rapidly advancing data analytics as deep learning models were adopted into its technology stack. It further stated that existing financial sector regulatory regimes, which were built in an earlier era of data analytics technology, might likely fall short in addressing the systemic risks posed by broad adoption of deep learning in finance.
Schwartz responded to Curran's post, writing: "I think a lot of this makes sense. The part that doesn't is the they'll be so smart that they'll do dumb things part. ".
I think a lot of this makes sense. The part that doesn't is the "they'll be so smart that they'll do dumb things" part.
Schwartz's comment appears to counter the idea that more intelligentAIagents would produce more irrational outcomes, while seemingly acknowledging the general concern that such automated decisions could be made en masse.
Autonomous AI agents are capable of reasoning, planning, and executing multi-step tasks across the internet.
With AI agents increasingly being developed for autonomous decision-making, the question of how thousands or millions of independently operating systems interact could become an important consideration for financial infrastructure and regulators.
출처: U.Today