世界经济论坛：成功发展人工智能需要彻底反思

研究表明，日益依赖于数据的金融服务行业正面临着新兴风险和监管问题。

根据世界经济论坛（World Economic Forum）的一项研究，随着人工智能和相关技术的日益普及，金融服务的风险、运营模式和竞争机制将发生根本变化。这份长达167页，题为《金融服务新物理学（The New Physics of Financial Services）》的报告被称为“世界上最大的关于人工智能对金融服务影响的研究之一”。该报告与德勤合作，是世界经济论坛“金融服务未来”系列的一部分。

该报告深入探讨了虽然人工智能技术在成本节约、效率提高方面的潜力巨大，但成功地应用此技术需要重新考虑风险管理假设，并颠覆传统业务和运营模式，这非常依赖于从公司内外顺利获取数据。随着金融机构采用全新的竞争机制，他们急需新的战略和合作伙伴关系。

“As AI drives operational efficiency,economies of scale alone will not sutain cost advantages,” Rob Galaski, Deloitte Global Banking & Capital Makrets Consulting leader, said in an August 15 press release.

“In the future,” he went on, “financial institutions will be built onscale of data and the ability to leverage that data.Increasinglybifurcated marketsare already emerging where data sharing is critical to competitive success and first movers are positioned to distinguish themselves by delivering better advice, constant presence, and curated ecosystems. Firms that lag behind are finding that their old strengths may not keep them as competitive as they once were.”

1

厘清定义

The paper is based on research over a10-month period, consisting of more than200 interviewsand seven workshop discussions with subject matter experts, with both incumbent institutions and AI-driven challengers represented.

The authors – led byWEF financial innovation lead R. Jesse McWaters– note that there is no universally agreed upon definition of artificial intelligence and “a marked lack of clarity” around it.

“When business people talk about AI, they typically are not talking about a particular technical approach or a well-defined school of computer science,” the writers say. “Rather they are talking abouta set of capabilities that allows them to run their business in a new way.”These amount to “a suite of technologies, enabled by adaptive predictive power and exhibiting some degree of autonomous learning,” which contribute tomachine-driven advances in pattern detection, foresight, customization, decision-making, and interaction.

图1：该报告称人工智能不存在于真空中，它需要“与所有其他技术创新的发展交织在一起”。

2

转型的意义

The premise of the WEF-Deloitte research is that financial firms must go beyond an understanding of AI's technical capabilities and start to analyzethese technologies' strategic implications and long-term impact, which “may be even more radical and transformative than we first imagined.”

The transformative impacts “will necessitate a level ofpublic-private commitmentto understand and continuously shape the future of financial services,” McWaters said in the August announcement. “And while emerging questions aboutconsumer protections and systemic risksremain the purview of regulators, effectively responding to these challenges will requirecollaboration between public and private stakeholdersin order to resolve regulatory uncertainties and manage the risks and opportunities of AI in financial services.”

The report emphasizes that AI does not stand apart from, and will be intertwined with, such other innovations ascloud computing, blockchain, smart contracts and quantum computing.

“There are a number of technologies that will need to come together in order for AI technologies to work successfully, most notably the use ofcloud along with technology to ensure data security,” says Dilip Krishna, CTO of Deloitte's Risk and Financial Advisory practice.

3

“新的重心”

The physics metaphor – a weakening of “the bonds that have historically held together financial institutions” and creation of “new centers of gravity where new and old capabilities are being combined in unexpected ways” – leads to several key findings about the financial services ecosystem's changing dynamics. These include the growth of collective solutions for shared problems and their consequences; the rise of uneasy data alliances; and the growing power of data regulators and their inevitable impact on the use of AI.

Collective solutions will arise, the report explains, to fulfill therelentless need for data setsthat can be fed into AI platforms, facilitating pattern recognition and problem-solving in, for instance, anti-fraud and anti-money-laundering (AML). Such shared solutions, the document says, canincrease the accuracy, timeliness and performance in non-competitive areas, spreading operational cost savings while improving the overall safety of the financial system.

4

集中化和“令人不安”的联盟

When it comes to shared and centralized solutions, there are questions, not yet fully resolved, aboutaccountability in the event of cyberattacks on shared data,and about responsibility for protecting sensitive customer data.

The rise of uneasy data alliances is said to be inevitable inan ecosystem where institutions are vying for new and diverse data sets to feed and optimize their data-hungry AI engines. In such an environment, the WEF report warns, managing partnerships with competitors and potential competitors will be critical and fraught with strategic and operational risks. It poses the question: “How will institutionsprotect the competitive value of their proprietary datain a world where that data must be shared with competitors to access minimum requirements of efficiency?”

One example of this trend: An employee health insurance partnership of Amazon, Berkshire Hathaway and JPMorgan Chase & Co. that aims to deploy big data and AI to align incentives. “Only time will tell if [such partnerships] drive sustained value,” the report says.

5

数据监管

Another trend cited by WEF and Deloitte is the emerging power of data regulators, with apotential impact on AI usage in financial services,and the uncertainty of how such regulation may play out in different parts of the world.

The report notes that theU.K. has adopted Open Banking, which requires customer-data portability and can erode incumbents' data advantage.Australia, Canada and Singapore are among countries actively considering some form of Open Banking,while in theU.S., Congress has been taking testimony from large technology companies on the topic of privacy and data security. These discussions could lead to new rules.

Of more immediate concern to Krishna is the need for risk managers to start thinking about the internal risks associated with AI as such platforms are rapidly being deployed. “Given the speed with which AI usage could take root internally over the next three or four years, risk regimes need to catch up, in terms of understanding, analyzing and managing the range of potential risks.”

本文作者是Katherine Heires，她是一位商业新闻自由撰稿人和MediaKat公司创始人。本文有删减。

点击原文，阅读《金融服务新物理学（The New Physics of FinancialServices）》报告全文。