American businesses have been paying Goldman Sachs investment bankers to help them with the process of putting their shares on stock markets and conducting an IPO for 156 years. However, David Solomon, the CEO of Goldman Sachs, stated earlier this year that artificial intelligence can now perform a large portion of that labor.
A six-person team at Goldman Sachs Group Inc. (GS) would typically need two weeks to produce an IPO prospectus, but AI can accomplish 95% of the work in minutes, Solomon claims. Solomon stated to the Financial Times earlier this year that “the last 5% now matters because the rest is now a commodity.”
Solomon is hardly the only CEO on Wall Street evaluating how AI may affect his company and its operations. The bottom rung of Moelis & Co.’s (MC) pyramid of investment bankers will probably get smaller, according to Navid Mahmoodzadegan, the company’s incoming CEO, who stated this on an investor call in June. This is because AI will take over some of the tasks that were formerly performed by youthful, aspirational financiers.
“We’re never not going to have a pyramid,” Mahmoodzadegan declared. “If technology advances as I believe it will, will we be able to be a bit more efficient on the pyramid later on? I believe that is a possibility.
Bankers and investment managers on Wall Street are evaluating artificial intelligence (AI) and predicting what the coming years will bring to financial firms and their employees.
Corporate America is seeing a boom in artificial intelligence. Andy Jassy, the CEO of Amazon Inc. (AMZN), stated in a memo to staff members last week that the company’s use of AI tools will result in the need for fewer workers in some positions. As economists, governments, and workers attempt to manage the effects of the new technology, New York State mandates that firms reveal if a layoff is due to AI-related technological progress.
According to Prasanna Tambe, an associate professor at Wharton and co-director of human-AI research at the business school, finance organizations have been among the biggest and fastest users of AI tools since they became available a few years ago. However, a large portion of adoption has been experimental thus far, as businesses start to comprehend the potential applications of AI technologies and their dependability. Furthermore, he noted that some of these tools are being updated on a weekly basis, which is expanding their capabilities quickly and making it challenging to predict their whole influence on the job market.
“The general rule is, anything that can be automated, will be automated,” stated Ted Mortonson, managing director and technology strategist at Milwaukee-based investment bank Baird.
In terms of how AI technologies will be incorporated into procedures across various departments, Mortonson told MarketWatch that his company was still in the discovery stage. Mortonson has participated in a process where the company’s business leaders have been collaborating with the IT department to test OpenAI’s enterprise services and improve the tools to better suit their requirements.
The information-gathering process has historically been carried out manually, with vast volumes of data being gathered and arranged, according to Mortonson, who is in charge of technology sector research. He can accomplish the same amount of work with a smaller crew, though, because the AI tools he has been testing are enabling him to automate a large portion of that labor.
“The number of people used, whether it be a junior banker, analyst or a salesperson, you just don’t need to throw bodies at it anymore,” Mortonson stated. “Technology can actually help you optimize it. Hiring is therefore decreased. Companies aren’t calling out for bodies. People are shouting for technology.
The role of AI on Wall Street
According to Mortonson, AI systems may be trained to efficiently and automatically scrape news, press releases, and earning outcomes. They can then categorize the data and make it available for a portfolio manager to access when needed. A smaller staff will be needed for even jobs like researching and creating financial models or creating PowerPoint presentations, as many of those procedures can also be automated, requiring only a few humans to supervise the process. In the past, Mortonson’s position as a tech strategist required a group of five to ten individuals to put in a considerable amount of overtime in order to produce a thorough presentation that addressed every aspect of technology. He claimed that he now only need two or three workers.
Former Wall Street data scientist turned venture capitalist Matt Ober told MarketWatch that it used to take him weeks to put together the pitch decks and investment memos he writes for firms. ChatGPT now handles it automatically. He may instruct the chatbot to generate a message after training it with previous memoranda from his company by uploading corporate financials, meeting transcripts, emails, and even his texts with startup founders.
The firmwide launch of GS AI Assistant, an internal conversational platform that connects to several large language models—the name for artificial intelligence tools that can produce and communicate with human language—was announced by Goldman Sachs on Monday. Although it’s a recent development, since January, roughly 10,000 employees have had access to AI technologies.
Goldman is now conducting experiments and implementing more AI tools. According to a person familiar with the situation, MarketWatch has learned that the company’s AI capabilities thus far include tools that help its developers with coding. Additionally, the AI technologies can help staff members with information retrieval, first draft and summary creation, proofreading, language translation, and email drafting. Additionally, Goldman is expanding what it refers to as Banker Copilot, which gathers pertinent material and offers investment bankers a dialogue-based interface. Before it is used more widely, a small group is now using it and optimizing it, according to a person familiar with the situation.
LLMs are also cutting down on research time for testing trading strategies, said to Alexander Fleiss, CEO of Rebellion Research, an AI financial consultant and hedge fund. For instance, strategies based on mean variance reversions—the quantified probability that an asset’s price will return to its moving average within a specific time frame after deviating—are highly profitable. Researchers can ask the LLM for the probability measures surrounding an asset’s price movements rather than performing what is called as a back test for each question they have, which involves replicating trades using historical data. Then, it would be worthwhile for a researcher to perform a back test on those with higher likelihoods. As a result, fewer people and more ground may be covered in less time.
AI can be applied to stock selection as well. Using a filter that identifies names with specific financial ratios or equities that have been rising higher is a frequent method used by investment funds to locate stocks. These filters are often strict, adhering to the same standards for decades. By enabling money managers to ask more specific and qualitative questions, LLMs can expand the filtering choices, according to Fleiss. For instance, among other things, AI can be requested to retrieve stocks of companies with expanding profit margins and earnings calls that use the most upbeat rhetoric. The stock-picking process can benefit from the addition of that information, even though it might not be traded on. According to Fleiss, LLMs are currently more useful for value investment strategies or mutual funds, as well as for any company where stockbrokers rely more on a company’s financial information to select stocks that will be held for extended periods of time.
Who will be left, then?
Who you ask will determine how many jobs could be eliminated and who will still be employed on Wall Street.
According to Tambe, it’s too early to predict with certainty how much of an impact it will have on the workforce on Wall Street. Businesses must determine the AI tools’ dependability, how to incorporate them into the decision-making process, and any potential regulatory ramifications for their use in investment banking. It will be challenging to predict with any degree of accuracy the entire extent of the impact on the financial-services sector until more of this is known.
However, it goes without saying that some talents will be less in demand as LLMs can handle a lot of administrative and research duties. AI tools might perform what Mortonson calls the “busy body work”—research duties like assembling financial models and notes or analyzing a company’s yearly 10-K financial report. Wall Street bosses will need to change how they keep people moving up the ranks at their companies because such task is typically associated with junior personnel.
Furthermore, sector or industry analysts who concentrate on fields like technology, banking, healthcare, or industry may find that a large portion of their work is performed by agentic AI, which has the capacity to make decisions and conduct actions that can be tailored. According to Mortonson, these agents may eventually be paired with a sector specialist.
“It’s very taxing on the brain to try to absorb all the sector data coming out,” Mortonson stated. An analyst can spend the majority of the day trying to identify a needle in a haystack because there is so much information, yet at least 60% of what is crucial is still overlooked, he said. The amount of labor needed to locate and process data can be decreased by using an AI agent, which can search for sector-relevant data around-the-clock. He predicts that this enormous benefit will fully materialize in two to three years.
