Generative generation — Is AI coming for your finance job?
- William Hoffman
- +Shubham Saharan
The days of junior analysts cutting their teeth in the world of finance through data entry or late nights creating slide decks may be coming to an end — if firms embrace generative AI technology like ChatGPT.
The technology is still very much in its infancy, and at this point is perhaps better at lying convincingly than it is at giving sound financial advice. But market participants who have viewed private demos for uses in finance are all highly impressed and think the technology will play a big role in the industry if it continues to improve.
“Very few things completely blow my mind, and AI is one of those things,” said Pete Stavros, co-head of Americas private equity at KKR, during a panel discussion at the Milken Conference in LA last month. “I would not underestimate what could happen in the next five years.”
We spoke to nearly a dozen bankers and investors about the implications of this new tech, which many felt could reduce the number of analysts in the industry and dramatically alter the landscape for those that remain. As AI speeds up some of the menial tasks that junior finance professionals have to deal with, what could the industry gain — and what could it lose?
Grunt work go away
Few market participants expect AI to render analysts entirely obsolete, but it seems more than possible that some of the more menial tasks they perform could be done by technology.
“We’re literally just making pitch decks and we’re told what to put in,” one analyst told 9fin. “We're moving logos, we're running financial models, and if there's technology that can do that for you, that's definitely threatening to the traditional analyst role.”
AI tools could also help analysts greatly increase the number of companies and sectors they analyze, and to review credit agreements, OMs and earnings reports more quickly.
That could free up analysts to focus on other, more analytical tasks, such as adding nuance and detail (not to mention human imagination) to financial models.
“By eliminating the due diligence and formatting audits and automating the most tedious manual tasks, you free up a lot of time to go focus on actual analysis,” said a buysider.
Theoretically, this kind of grunt work — often seen as a gauntlet that junior bankers have to run in order to advance their careers — could allow analysts to get more hands-on experience in the dealmaking process.
“Spreadsheet jockeying and editing slideshows, it's all just backbreaking work for the sake of giving someone backbreaking work,” said one senior credit analyst who made it through that gauntlet and out to the other side.
Technology is already having a dramatic impact on these roles even without generative AI technology, noted an investment-grade credit banker. The electronic bond book ordering project Direct Books is threatening some sales positions; other software can spit out investment models in seconds.
On top of that, it’s hard for people to trust any technology — but especially AI — fully.
“I don’t think this technology is more dramatic or more drastic than other technological developments that we've seen,” the investment-grade banker said. “You're always going to need the human side of this. Someone to say ‘This doesn't look right together,’ or somebody to say ‘The AI has completely screwed up and it's doing something wrong.’”
Back to basics
Many senior bankers and portfolio managers still see the value in human analysts doing these junior-level tasks, and say it’s important training for good credit analysis.
Learnings the basics of how a trade is booked, what the clearing process is, how to put together a pitch book or slides, and understanding market dynamics like the relationship between interest rates and credit spreads is all valuable experience.
“To remove people from that, I would think it would have a negative impact,” the investment grade banker said. “It's an important component of the business and of the analysis process that you're no longer learning and experiencing.”
There’s another problem: models and technology are only as good as the data that’s put into them, and the finance industry is often confronted by situations with few modern precedents.
How would AI models have reacted to Covid-19, or Russia’s invasion of Ukraine, or the strongest jobs market in a generation? As much as answering those questions can be assisted by technology, the process ultimately requires a human touch, sources said.
“There is an art overlaid on top of the data, and you have to have a feeling for what is happening in the market,” one UK-based banker said. “If you’re not working mechanically and preparing documents, you might not have that embedded knowledge needed for decision-making.”
Using AI tools doesn’t make it impossible to get that experience, but sources said it will likely require institutions to rethink how they train younger staffers.
“We're definitely going to have to think about how to train the next generation,” a senior credit banker said. “I think we'll use AI, but we have to figure out how to train new people and prepare them for those more senior roles.”
Fewer opportunities
Most people agree that no matter what analyst roles look like going forward, there are going to be fewer positions available thanks to AI.
As much as AI may save firms money by reducing analyst headcount, that reduction also shrinks their pool of future rainmakers.
“If you eliminate the number of entry-level jobs, that cuts your pool in an already very intensely competitive industry,” said the senior analyst.
Indeed, one effect of AI could be to make getting a job in finance even more competitive.
Some junior analysts said they are already struggling to find higher-level jobs within their institutions, and in recent months, many are choosing to stay put rather than looking for an early pivot to a different market (such as going from investment banking to private equity).
This could even impact people who haven’t yet joined the industry. Recent grads who are looking for analyst positions may have to get creative in order to stand out, demonstrating the value they can add to the deal-making process beyond the rote tasks that AI tools can execute, said a third-year investment banking analyst.
It’s also possible that the early experiences associated with junior finance roles could be found elsewhere in the corporate world, the senior analyst added. Such candidates could gain valuable knowledge and skills in the back office of a manufacturing company or the treasury department of a retailer, for example, and then take that experience to banking or investing.
“Go see how an actual business is run and get your hands dirty in the industry itself,” the senior analyst suggested. “Then you can go and use that experience to become a better investor, focused on that sector.”
Talking points
Some corners of the financial industry are feeling the pressure faster than others. For example, sales and marketing teams are firmly in the crosshairs.
One prime services banker said they asked an AI service for “the bull case for investing in private credit,” and received a detailed list of arguments in favor of the asset class.
“My sales team was like, ‘That's the email that I send to investors, those are my talking points,’” the banker said. “So in the world of marketing and fundraising, it's changing things already.”
Bankers also expect AI to speed up underwriting in auto and consumer lending — which, again, could lead to a reduction in headcount.
AI is already helping Wells Fargo speed up income verification on auto loans, said Tanya Sanders, head of Wells Fargo Auto, speaking at the Fintech Nexus USA 2023 conference in New York on 10 May. Chatbots are another way that generative AI could improve the lending experience for customers, she said.
“I think with any automation, whether it's machine learning or AI, it does imply that there's some impact to the staffing that we have within our underwriting departments,” she said.
“But what I tend to look at is [does] it provide scalability for our business. Once we have those tools and technologies deployed, we can scale up and down with fewer impacts over time to staffing.”
Special sauce
Generative AI may also be coming for financial researchers. Richard Hunter, chief credit officer at Fitch, said research teams could use AI to pull data more quickly — but that humans will do the analysis and provide opinions.
“Opinion is a higher moat for ChatGPT to cross,” said Hunter. “It's obviously not an insurmountable moat, but it's a higher moat.”
“Unfortunately, it does mean that people drift from talking about facts, which is actually quite a useful thing to talk about, into maybe doing more opinion where they feel they can compete against ChatGPT on a better playing field.”
For the graduating class of 2023, the good news is that capital markets are notoriously slow to adapt to new technology.
Sources noted they are not overly worried about generative AI disrupting an asset class that has struggled for years to even implement electronic trading.
“Fixed income markets will become more electronic over time,” said Andrew Relph, a managing director at Concise Capital. “But right now, you still have to pick up the phone and call the broker to make a trade. So there's going to need to be a human element to it, for the time being.”