A Bain Capital Ventures investor flew out to meet a founder less than a week after a machine-learning tool brought an under-the-radar investment opportunity to her attention.

A mutual-fund manager received a report aggregating the market sentiment of thousands of notes from hundreds of the firm’s analysts before a trip to New York City.

An analyst at the hedge fund Balyasny Asset Management created a presentation of global import and export data that would have normally taken a couple of weeks in a few hours.

Speed and ease — that’s how generative AI is changing the game for finance professionals.

Because AI singled out an emerging trend and identified a startup poised to benefit from it, Bain Capital Venture’s Christina Melas-Kyriazi beat out other investors to lead a fundraising round mere days after discovering it. “We were able to do that because we could sort of identify it fast enough. Speed is really important here,” she told Business Insider.

And for T. Rowe Price’s Sébastien Page to capture the “nuance” of how the firm’s analysts felt about each sector in the past? “It would’ve been pretty much impossible,” the head of global multi-asset and the firm’s chief investment officer told BI. “It would have been a lot more anecdotal.”

Generative AI is threatening to upend many industries with its ability to crunch data and spit out new information in a humanlike way — and Wall Street is no different. As use cases are being built, tested, and scaled among workers, Business Insider talked to 35 people in various roles across the finance industry to hear how AI adoption is happening on the ground. Many of these people were granted anonymity to speak freely about their experiences.

I hope it does replace a lot of the analyst work — a lot of the analyst work is bullshit.

From C-suiters to junior staff, most said they welcome generative AI’s potential to boost efficiencies and cut grunt work. They believe that less administrative friction will allow more time for core work and raise the bar for critical thinking and analysis.

“I hope it does replace a lot of the analyst work,” one former junior banker at a New York-based investment bank told BI, referring to the all-nighters entry-level investment banking analysts spend assembling pitch decks or punching numbers into Excel spreadsheets. “A lot of the analyst work is bullshit.”

Financial advisors and analysts in JPMorgan’s wealth and asset management business are saving a couple of hours a day, BI previously reported. At Man Group, a machine-learning tool can send brokers trades that could offer the best pricing on a specific stock or security based on historical execution data.

“It’s a real source of alpha,” said Eric Burl, the head of discretionary at Man Group, the world’s largest publicly listed hedge fund.

Getty Images; Alyssa Powell/BI

While many express enthusiasm for AI tools that allow them to save time and potentially focus on bigger and better things, others are more cynical. Some raised doubts about the technology’s reliability and usefulness, concerns about their firms’ approach to using AI, and questions about how the technology will affect jobs or work-life balance.

If AI tools can eliminate much of Wall Street’s entry-level work, it could shake up typical career paths. But only time will tell how this will change Wall Street’s work-till-you-drop culture. In a survey of 780 banking and capital-markets employees by Accenture Research, 62% of respondents expect generative AI to increase people’s stress and burnout.

“The hours aren’t going to change, because they’re a product of the culture more than actual workload,” the former junior banker said. “People will always find things for you to do because the expectation is that you work 80 or 100 hours a week.”

Finance has always been and will no doubt remain a competitive business where success depends on the speed at which information is obtained and acted upon.

So it should come as no surprise that in the age of AI, Wall Street firms have burst out of the gate to leverage the technology to get a leg up on the competition. In the race to unlock AI’s potential, they’re testing investing models, hiring top talent, and developing their own cutting-edge research.

Private-equity firms such as Blackstone are building teams to leverage AI’s cost-cutting and productivity-inducing benefits within the companies they own. The middle-market PE firm Thomas H. Lee, which launched a generative-AI coding tool for select portfolio companies, reported that its engineers were up to 30% more productive just four weeks after the rollout.

There’s definitely a first-mover advantage

Quantitative trading firms and hedge funds such as Two Sigma use powerful compute engines and AI chips to uncover new sources of investing alpha. Meanwhile, consumer banks like JPMorgan, with their sprawling technology footprints and troves of data about how consumers spend, save, and invest, have been focused on readying their data strategies to take full advantage of AI.

But there’s no one-size-fits-all approach to tapping AI’s benefits, and finance doesn’t have the best reputation for integrating tech. Because it is such a nascent field, firms are still trying to figure out the most effective, efficient, and safest way to develop and scale the technology before unleashing it on the masses.

“There’s definitely a first-mover advantage,” Keri Smith, Accenture’s global banking data and AI lead, told BI. Finance firms that have already been investing in technology modernization, like migrating to the cloud and making sure enterprise data is well organized and tagged, are poised to step out in front of the pack.

Smith said that after 18 months of experimentation and development, Wall Street’s understanding and use of AI has matured significantly. This has led to thornier questions about how companies can differentiate themselves from the crowd and how to best develop and train talent.

Blackstone is looking to supercharge employee productivity with AI-powered search

These are the puzzles that chief information officers, chief technology officers, and data leaders who oversee their firms’ AI strategies are expected to solve.

Those roles now have “a very important seat at the table,” Ken Griffin, the billionaire Citadel founder and CEO, said in May at the Milken Institute Global Conference. “They’ve got the attention of the CEO. How can we use technology again to really drive productivity?”

The pressure to mine productivity gains has put some of Wall Street’s tech leaders in a tricky place: how to gain a first-mover advantage while not exposing the enterprise to vulnerabilities from moving too fast. Goldman Sachs’ tech chief, Marco Argenti, told BI he had to push back against engineers who wanted the firm to roll out a new generative-AI coding tool more quickly.

“You go slow to go fast,” Argenti said. “It’s such a big revolution that you need to be able to go faster safely. So, at the beginning, you need to take careful steps so that you remove a lot of that toil” involved in building AI applications, he said, like protecting confidential data and retrieving information in a way that will minimize inaccuracies.

“At the beginning of the process, yes, I had to curb people’s enthusiasm a little bit,” he said.

At one midsize Wall Street investment bank, a senior banker uses ChatGPT daily — a sign of how AI is seeping into the analog world of investment banking, where bankers’ interpersonal connections and flair in the boardroom have long reigned supreme.

The banker told BI he’s “retraining” himself to use tools like ChatGPT and Copilot more frequently. It’s transformed how he researches “everything,” from data about industry sectors to ideating new proposals to bring to clients. “Rather than just brainstorming with colleagues, you might brainstorm with a robot now,” he said, “which I’ve actually found to be pretty helpful because it spurs some new thoughts.”

The banker primarily uses AI to compress dense troves of information — whether from research notes or dozens of meetings — into more digestible takeaways. “Now I don’t have to read six reports,” he explained. “I can query those reports through AI and get a pretty snappy summary of what they are, and then I can take that and use my own brain at that point to put it all together.”

His junior team members are hopping on the bandwagon to accelerate tasks like making slide decks, even if they’re sly about it. “I know for a fact that our analysts are using it quite a bit to help them write and create prose and bullet points,” he said, adding: “They don’t readily admit it — until you sit down and have a couple beers with them.”

But with the elimination of some of the drudge work, there’s the risk that some jobs could become obsolete. “Maybe we don’t need as many analysts down the road as we otherwise would,” the banker conceded.

Since most roles in finance include a lot of collecting and processing data, there’s no question that generative AI is set to shake up jobs on Wall Street.

A report from Accenture Research found that capital-markets roles are ripe for AI-related job displacement. The consulting firm estimated that 72% of jobs within investment banks, asset managers, and wealth advisories have “higher potential” to be automated or augmented by AI.

But just as Excel didn’t replace accountants, tech leaders don’t see AI displacing humans.

“Employees with AI skills will replace people without AI skills,” Andrew Chin, the chief AI officer at the $759 billion money manager AllianceBernstein, told BI

In many cases, AI is simply an enabler, Lisa Donahue, the co-lead of the Americas and Asia regions at the global consulting firm AlixPartners, which is best known for its work cleaning up messy balance sheets and turning around troubled companies.

“It enables executives to get information in a comprehensive way faster, which allows you to make your decisions faster and quickly move toward execution,” Donahue said. AlixPartners, which advises private-equity firms, uses a proprietary, AI-powered diagnostics tool that draws on decades of the firm’s consulting work to help buyout shops evaluate potential acquisition targets.

When it comes to implementing and interpreting recommendations, “you’re still going to need experienced people to execute,” she said.

Jobs across the industry vary, obviously, but if you’re in the business of giving advice and influencing outcomes, it’s time for those hard-for-a-machine-to-replicate skills to shine.

Getty Images; Alyssa Powell/BI

One fundamental analyst at a large hedge fund, who’s seen more accurate summaries of earning calls and research reports as the use of generative AI has become more widespread, said that the tech has allowed him “to be more thoughtful” thanks to the additional time he has. He uses the extra time to write notes for his portfolio manager or craft questions for the management teams he covers.

A data scientist at a midsize hedge fund told BI that generative AI models are a “superpower for coders.” One of his biggest use cases is solving coding problems in different coding languages.

He’s been freed up to go deeper or think more abstractly about his projects, knowing the coding chunk won’t take as long as it used to. He compares it to when writers moved from typewriters to word processors, with spell-check, the ability to delete and move items around, and more. He said he’d even take on the expense of ChatGPT himself if his firm stopped covering it.

Not everyone in finance is convinced that generative AI will bring radical changes. Some employees at firms that have long used models powered by AI to seek an edge, such as quantitative hedge funds or trading firms, say a lot of the benefits are overhyped.

“There’s a lot of talk, but I haven’t seen anything yet that changed the world,” a quantitative portfolio manager told BI, adding that “all those GPT models might present somewhat of an improvement, but we haven’t seen anything dramatic like a breakthrough.”

Other cohorts, including fundamental investors who make their living by picking the best stocks, believe their investing style is too nuanced to rely on automation.

It’s a “lack of available use cases rather than a deliberate decision not to,” a fundamental analyst at one of the world’s biggest hedge funds told BI. Two other fundamental analysts BI spoke with agreed there was no explicit use of AI in their processes.

Though some say OpenAI’s debut of ChatGPT represented a step change in AI and machine-learning capabilities, these generative AI models can still spew out misinformation, which means humans need to check the work done by AI. Not doing so is a mistake that some have learned the hard way.

At one large hedge fund, some analysts have had to redo entire reports after realizing the numbers pulled by ChatGPT were incorrect, according to a colleague at the firm. These reports, which typically take half a day or so to complete, were generated by the bot almost instantly, but the analysts realized it used the wrong revenues and profits to draw up the analysis.

“We trust it as much as an intern,” the hedge-fund employee said. “You have to check its work.”

An emerging challenge for Wall Street firms now is closing the gap between the staff and the technology, and some firms are finding a “bit of friction” with adoption, Accenture’s Smith said.

A January report by the firm found that 93% of finance workers and employees are very keen to leverage generative AI, but Smith said their organizations’ rollout of AI tools and approach to upskilling are a source of frustration.

“I wish I could use it more,” one midsize private-equity firm vice president told BI. Their firm blocked employees from using publicly available generative-AI tools and built its own model using OpenAI.

“For some reason, we decided to become software developers and build our own shit versus just buying off-the-shelf stuff from firms who do this for a living,” they said.

Only 7% of banking and capital-markets organizations are actively reskilling their workforces at scale, Smith added.

“I think I’m using it?” said one wealth advisor. His employer, one of the world’s largest brokerages, has developed an internal AI product that analyzes client data and generates reports. One of his interns showed him how to use ChatGPT; other team members have used it to summarize hundred-plus-page private-equity offerings. The advisor plans to spend more time this summer dabbling with it.

An advisor at Northern Trust also cited age as a hurdle to adoption.

“I can see them being very apprehensive,” he said of his colleagues in their late 50s and early 60s. “Why would they want to learn something new when they are so close to retirement?”

At Citibank, employees are encouraged to raise new AI use cases.

One potential use case could be summarizing hundreds of pages of regulatory documents into a list of the relevant obligations, one Citi employee told BI. A human would still need to double-check and validate that the information was correct, “but that just saved you so much time that’s wasted just reading,” they said.

However, the process for getting any type of AI approved is arduous and can take months, the person said.

“A lot of checks and balances, a lot of validating, a lot of evidence-based artifacts need to be provided and committees needing to review and approve,” they said of the AI approval process, adding that “it will just beat you up.”

Some firms are not only automating grunt work but also looking to replicate work that could be considered unique.

JPMorgan’s private bank launched an AI tool that acts as an assistant to its bankers. The firm’s ultimate aim is to use generative AI to replicate the success of its best bankers for all advisors. The firm is working to train the AI copilot on data that breaks down how the bank’s top advisors respond to emails, how they handle client interactions, and what their portfolios look like.

One former JPMorgan executive director told BI he was unnerved by the prospect of an AI tool that helps advisors mimic how top performers respond to emails or communicate with clients. It could cause tension between advisors who think their best practices are being shared with competing advisors in the same market.

“I think it would be weird to have someone’s intelligence feeding a tool,” he said.