According to Chief Information Officer Marco Argenti, Goldman Sachs is introducing a generative AI assistant to its bankers, traders, and asset managers. This is the initial phase of a program that will eventually develop the characteristics of an experienced Goldman employee.
Argenti told CNBC in an exclusive interview that the bank has so far made GS AI Assistant available to roughly 10,000 employees, with the aim of having it available to all knowledge workers within the organization this year. At first, it will be useful for things like editing or summarizing emails and translating code between languages.
According to Argenti, consider all the things you would wish to accomplish in relation to a range of use cases for all those professions that are now at your fingertips. The Goldman assistant gives you access to the newest and best models through an extremely user-friendly interface.
Since ChatGPT went widespread around two years ago, Goldman’s action marks the aggressive deployment of generative AI tools to their staff, joining JPMorgan Chase and Morgan Stanley as the top three investment banks globally.
According to experts, the ability of massive language models to replicate certain parts of human cognition has led Wall Street to adopt generative artificial intelligence more quickly than any other disruptive technology in recent years.
Although it can already send emails, answer questions, and summarize long documents, there are high hopes that future iterations will demonstrate so-called agentic abilities—the ability to carry out multistep activities with minimal human involvement.
Argenti, who joined the company from Amazon in 2019, frequently compared the AI program to a new hire who will adopt Goldman culture over the next few years when discussing his vision for artificial intelligence at the company with CNBC.
According to Argenti, depending on the objective, the tool will first mostly generate responses based on Goldman data that has been fed into AI models from OpenAI sChatGPT, Google sGemini, and Meta sLlama. According to him, the bank is also considering models from businesses like Cohere, Mistral, and Anthropic.
According to Argenti, the AI assistant starts to resemble a conversation with another GS employee.
Learning the Goldman Way
According to him, the second stage is when you begin to exhibit this agentic behavior, which is when you say, “I’m doing a task on behalf of a Goldman employee, and I need to take a set of steps.” At that point, the model will begin acting like a Goldman employee rather than just saying things like one.
This clarifies why businesses have banned workers from utilizing ChatGPT for work and are instead developing their own platforms to use the technology. In addition to protecting their data, it enables businesses to create AI systems that more closely mimic the finest members of their own workforce.
It is crucial, according to Argenti, that the AI have a very distinct identity that embodies the principles, values, expertise, and mindset of the company.
According to him, in practice, the AI will learn from the same mistakes that a seasoned Goldman employee would make, such as double-checking their work using numerous data sources or using a certain algorithm for a calculation.
However, as AI models become more and more ambiguous in their ability to distinguish between human and machine thought, Argenti says he is most enthusiastic about what lies ahead, possibly in three to five years.
“The model would actually reason more and become more like the way a Goldman employee would think at this stage of AI,” he said.
According to Argenti, the AI would be able to create comprehensive plans in the same manner that a seasoned Goldman employee would, rather than being given a run book—tech industry jargon for a set of specific instructions for finishing tasks or handling crises.
Disruption risk
Employees may experience a new wave of concern due to the potential for that future and the reality that Wall Street professionals are assisting in the development of a technology that might replace certain professions while enhancing others and establishing new ones entirely.
Similar to Goldman, other large investment banks plan to provide generative AI capabilities to all of their employees within the next several months.
According to a person with knowledge of JPMorgan who spoke about internal affairs and asked not to be named, the bank currently provides in-house generative AI tools to over 200,000 employees. As of late October of last year, it was accessible to some 40,000 Morgan Stanley employees, the bank said.
Due to generative AI, which has the potential to increase revenues by billions of dollars, the finance and technology sectors are thought to be among those where workers are most vulnerable to disruption. Earlier this month, Joe Rogan, a podcaster, was informed by Mark Zuckerberg, the CEO of Meta, that this year, its AI will be able to write code just as well as mid-level software professionals.
According to a report from Bloomberg’s research arm, global investment banks may lay off up to 200,000 employees over the course of the next three to five years as they adopt AI. According to the analysis, which was based on a poll of tech executives at large banks, the back and middle office—support and operational roles—were the most vulnerable.
The official position at Goldman, however, is that AI will enable workers to accomplish more rather than necessarily lead to a reduction in the demand for human labor.
According to Argenti, the value of having an outstanding human staff will only increase.
It always comes back to people, in my opinion,” he remarked. People are going to change things because they are the ones who will be the ones who teach, empower, and evolve AI then take action.
More from CNBC:
-
Trump token plunges over 20%, bitcoin pulls back from inauguration day record of $109,000
-
Here are the products and companies most at risk from Trump s tariff plans
-
Trump says he s thinking of imposing 25% tariffs on Canada and Mexico in February