Trump’s coming push on AI and banks

Three-quarters of companies are adopting AI, according to a new Randstad survey of 12,000 workers, but only 35% of employees report getting any training in AI.

On this score, banks are ahead of the curve. Year on year, according to our latest data, the number of employers offering AI-specific training has jumped from half of the 50 banks tracked in the Evident AI Index to nearly two-thirds.

Why it matters: In a highly competitive market for talent, banks that are committed to developing it are better able to attract, retain, and not least, promote it. Those that offer any AI training to employees demonstrate a 1.4x higher growth rate of their AI staff than banks that feature more generic training in data and analytics. Banks that offer any AI training saw a 17.1% increase in their AI workforce, compared to 12.2% for those that had none. Among Data Engineers specifically, that advantage grows to 1.6x.

We regularly spotlight AI use cases in finance, but it’s rare to see what they look like. This week, Evident’s Alex Inch sat down with Arta Finance CEO Caesar Sengupta for a peek behind the curtain. With $90 million in funding and support from investors like former Google CEO Eric Schmidt and ex-ING and UBS chief Ralph Hamers, Arta is hoping to democratize access to wealth management. From a portfolio explainer in Singlish to collaborative agents that build investing portfolios, Sengupta showed Alex how Arta is using AI to make it happen.

This version has been edited for clarity and length. Check out the full interview, including more demos of AI features, here.

INCH: One of the ways you use GenAI is deploying multiple AI agents to draft an investment portfolio. Could you tell us more about how that works?

SENGUPTA: What it does is first take the overall theme and use reasoning to break it into sub themes, and each of these kicks off a separate co-pilot. Each co-pilot has access to different data sources based on whether they're public, proprietary, or we’ve constructed the data source ourselves. Next, a separate master LLM checks if each of these investment copilots has done their work correctly. It's exactly the way a hedge fund would do it with analysts, right? They would send an analyst off to research a particular data stream, which is how we've comprised the whole thing. Once all of this comes back, it gets put into one consolidated investment memo. For example, if you wanted to invest in Nvidia’s supply chain, it would understand things like specialty gas and material providers that will give you exposure. For the user we'll describe what the steps the agents went through, as well as give people links to the resources the agents used so they can consult these resources to validate it.

When you're implementing features using AI, what is the balance of time spent on the AI side of that feature versus the UX side of it?

It’s a hard question to answer, but we deeply care about usability. AI can be quite intimidating at times. Especially in the investment world, being able to understand what the AI is doing is super important, that's why we've spent a lot of time thinking through the UX. I don't know if I can say 60% on UX versus 40% on AI, but all I can say is we've spent a huge amount of time on it. It’s also very important to corroborate and substantiate the AI’s findings. It adds validity to the AI and makes it much more usable.

Some employees at banks are concerned about accelerating AI adoption, maybe because of job replacement or task automation. What can business leaders do to keep teams onside?

Today most banks want to serve a lot more clients. Most RMs (relationship managers) would gladly take on 10x more if they could, but they spend a lot of time dealing with very basic stuff. My sense is that really progressive leaders are helping their teams understand that AI can actually help them serve more people and expand the business a lot more. The vast majority of people today don't have access to wealth management, and most banks are raising the minimum levels because they don't have a way to solve for it. There are also markets across the world where the number of wealth managers is reducing. So I think if you have a good wealth manager or a financial advisor, you can supercharge them with AI. That's a very compelling story. That's my approach of how to think about AI, at least in our field. There might be other fields where AI is much more threatening to jobs in the short run, but I think in the wealth space, there's just so much opportunity for expansion and democratization that AI is just going to make it more possible for people to get high quality services.

Ahead of the tentpole event on the AI research calendar – the NeurIPS 2024 conference in Vancouver next month – we looked at which research produced by the banks has gotten the most traction. Here are this year’s standouts, ranked by citations:

#1  Explainable Artificial Intelligence (XAI) 2.0

Contributor (Current Bank Affiliation): Freddy Lecue (JPMC)
Description: Challenges and opportunities of improving AI explainability

#2 Crosscodeeval

Contributor (Current Bank Affiliation): Bing Xiang (Goldman Sachs)
Description: Evaluation benchmark for code completion models

#3 Exposing flaws of generative model evaluation

Contributor (Current Bank Affiliation): George Stein & Layer 6 Colleagues (TD Bank)
Description: On the issues with performance metrics used to assess generative AI image models

#4 Proximal reinforcement learning

Contributor (Current Bank Affiliation): Andrew Bennet (Morgan Stanley)
Description: Improvements to machine learning techniques

#5 For sale: State-action representation learning for deep reinforcement learning

Contributor (Current Bank Affiliation): Edward Smith (RBC)
Description: Novel approach for improving image based machine learning

The top paper this year addresses a pressing concern for all users of AI: How to explain the way an AI model came up with its answers. Models are growing bigger and more complex, and no one’s figured out how to get explainability right. The risks are obvious, as Air Canada’s faulty AI chatbot showed.

One week until the Evident AI Symposium – an unmissable gathering of the top minds shaping AI in banking.

Following the latest Evident AI Index update, this exclusive event will dive deep into where banks are succeeding and struggling in their AI transformations. Including which banks are leading the way? Where are they delivering value? And what are the headwinds and opportunities on the horizon? This is your chance to get a first hand look at the future of AI in banking for 2025.

Plus, unpack the secrets to JPMorganChase’s Index success with Teresa Heitsenrether’s Opening Keynote Speech and listen to our expert panelists on:

• How is AI driving insights and knowledge creation?
• How can banks benefit from the most promising AI research?
• How are geopolitical dynamics affecting AI and business?

View the latest agenda and recently confirmed speakers here.

We look forward to seeing you – either in person, or virtually.

In this week’s Use Case Corner of the best we’ve seen, two newly announced generative AI examples (including a first timer) and one on fraud detection.

#1 Customer Insights Extraction Using Transformers and NLP

Use Case: Customer Retention
Vendor: n/a
Bank: BBVA

Why it’s interesting: Multilingual tool that pulls responses from customer satisfaction questionnaires and categorizes them by topic. That lets the bank analyze large amounts of qualitative data better to respond to customer needs.

Potential ROI: Increased customer satisfaction and engagement, better operational efficiency
Reported ROI: n/a

#2 Generative AI for Performance Reviews

Use Case: Process Automation
Vendor: n/a
Bank: Standard Chartered

Why it’s interesting: Standard Chartered’s first-ever generative AI use case is meant to improve employee and company performance feedback and goal-setting.

Potential ROI: Improved review quality and employee development
Reported ROI: n/a

#3 AI-Enhanced Scam Detection and Security Warnings

Use Case: Fraud Detection
Vendor: n/a
Bank: ANZ Bank

Why it’s interesting: Announced in their annual report last week, the bank's new tool uses more traditional AI approaches to allocate a score depending on how risky a transaction is for customers using its online banking system.

Potential ROI: Reduced fraud-related losses, improved customer security and trust
Reported ROI: n/a

Have feedback on or ideas about use cases? Let us know at [email protected].

"I’m particularly proud of the fact that one of my students fired Sam Altman. And I think I better leave it there"

- Geoffrey Hinton, “Godfather” of AI, after receiving this year’s Nobel Prize in Physics

“AI will fundamentally change the way we operate,” said ANZ CEO Shayne Elliot on last week’s earnings call. The bank chief cited “profound” productivity opportunities, particularly with tools like coding co-pilots.

Introducing generative AI to a use case is “only 5% of the job,” according to ING’s Chief Analytics Officer Bahadir Yilmaz. The other 95%? Good risk management systems: Bahadir’s teams go through a 20-step process to evaluate an AI system for 140 risks.

BBVA expanded a partnership with the University of Navarra to offer an AI training program for their managers to learn how to improve productivity, specifically through generative AI.

Morgan Stanley is big on India, recently opening a tech hub in Mumbai. Sal Cucchiara, chief information officer for wealth & investment management technology, called India the “AI hub for startups.”

Even if we made no more progress on AI, there’s still so much benefit to be had, said Cohere’s CEO Aidan Gomez. As models get bigger, their effectiveness does taper off over time, he said – aligning with growing speculation that models are hitting the “scaling ceiling”.

Banks are ramping up their hiring of AI Product Managers – the people who turn AI into something useful for employees or clients.

• TD Bank advertised for a Technical Product Owner focused on the delivery of machine learning solutions.
• CommBank is looking for a Senior Manager of their Gen AI Playbook, responsible for ensuring its GenAI platforms and services best serve customer needs and boost the company’s overall product approval scores (NPS).
• JPMC is hiring an AI Product Manager specializing in conversational AI to manage the strategy and adoption of firm-wide analytics products.

There’s a correlation between having more AI product managers and how AI mature a bank is:

Any noteworthy people moves you know about? Share them with us at [email protected].

AI didn’t feature much on the campaign trail (see “What the election means for AI,” The Brief, Oct. 31), but there’s no shortage of fresh speculation about Trump’s impact on tech policy (see Forbes, Fortune, TIME, TechCrunch, Vox). Without Rohit Chopra at the Consumer Financial Protection Bureau and Lina Khan at FTC, does AI investment thrive in an “America First” landscape... or does Big Tech face new and unexpected hurdles to overcome?

One of many big personnel moves this week was Elon Musk’s nomination to co-run the new Department of Government Efficiency (DOGE, get it?). Americans for Responsible Innovation (ARI) says Musk could defund parts of Biden’s Executive Order on AI, including the U.S. AI Safety Institute (which enjoys broad support in the Senate). Beyond cost-cutting, Musk’s mercurial views on AI, which oscillate between alarm (10-20% chance AI “goes bad,” he says) and self-interest (support for California’s SB 1047 legislation that would disadvantage xAI’s competitors), can more easily find a presidential ear now.

As for us, we see a couple things clearly about Washington’s new approach to AI.

First, very little can be derived from sideline speculation regarding the ongoing Cabinet Hunger Games. Michael Kratsios, former White House CTO and current managing director at Scale AI, is handling tech policy during the Trump transition. Lee Zeldin, the pick to head the Environmental Protection Agency, mentioned energy and AI policy in the same breath. Big whoop… The jobs that matter are antitrust chief at the DOJ and the head of the FTC.

Second, whatever comes next will surely disrupt the status quo inside the Beltway. This is not necessarily a bad thing for Wall Street. It buys banks precious time to get their houses in order with regards to deploying priority use cases, documenting ROI, implementing AI-specific governance, and coming up with an agentic AI strategy. Whichever banks figure all that out will be well placed once the dust settles in Washington.

Third, 2025 just got a lot more interesting. Buckle up…

The Research Corner section in this Brief was updated to remove the attribution of RBC to the paper "Exposing Flaws of Generative Model Evaluation". In addition, due to an error in our data source, we mistakenly included a paper that wasn’t co-authored by a researcher from a bank. The list was updated to include the 5th most cited paper from a bank researcher.