Goldman Sachs' Strategic AI Integration: Balancing Innovation with Caution

Goldman adopts measured approach to roll out of AI

This content has been selected, created and edited by the Finextra editorial team based upon its relevance and interest to our community. Currently, half of the 46,000 employees at Goldman Sachs now have access to artificial intelligence. "We have the entire organization that needs to somehow re-tune and re-tool itself for AI," chief information officer Marco Argenti told Fortune magazine. "But, I think we've been very, very, very intentional with regards to driving people change management." For example, the firm is currently experimenting with agentic AI, which has yet to be deployed across the firm, despite the apparent benefits in the automation of key tasks, such as compliance checks or the processing of customer transactions. However, recent studies and poor experiences in rushed projects have proved that the AI agents need specific training and quality data to prevent hallucinations and errors in the results they produce. Goldman says it is still assessing what additional controls it needs to effectively and safely use agentic AI. Roughly one out of every four Goldman Sachs employees is an engineer, and this group was the first Argenti targeted when deploying generative AI tools. Argenti gave those workers access to AI coding assistant tools, including GitHub Copilot and Gemini Code Assist. Goldman has conducted competitions inspired by reality TV show Shark Tank so that developers could share their most creative uses of AI. Argenti told Fortune that he measures the return on investment from these copilot tools in a few ways, including frequency of use and the acceptance rate of code generated by GitHub and similar tools. Broader use of generative AI within the company came with the launch of GS AI Assistant, which rolled out last year and has expanded to 10,000 employees including bankers, traders, and asset managers. This tool, which Goldman anticipates will be available to nearly all employees by the end of 2025, can summarize documents, draft emails, analyze data, and create personalized content. Recent research from Vlerick Business School, supports Goldman's cautious approach, challenging common misconceptions that AI can fully replace humans in budgeting, instead highlighting the need for a balanced approach. The study sought to understand the role AI plays in corporate budgeting and how it compares to human managers in both the effectiveness and strategic alignment of financial decision-making. The researchers focused on two key aspects of budgeting: tactical and strategic. Tactical budgeting involves data-driven decisions that optimise short-term performance. Strategic budgeting is more in-depth and focused on the long term. To examine this, the researchers conducted a management simulation in which seasoned managers were asked to allocate budgets for a hypothetical automotive parts manufacturer. Their decisions were compared to those made by an AI algorithm using the same data. The study found that AI consistently outperformed humans in optimising budget allocations when the strategic framework was clearly defined. But when key performance indicators (KPIs) were misaligned with strategic goals, AI struggled to produce the most effective results. The findings suggest that AI can replace humans in tactical budgeting, where its speed and precision lead to more efficient outcomes. However, human oversight is essential in strategic planning to ensure that short-term financial decisions align with broader business objectives. "As AI continues to evolve, companies that leverage its strengths in tactical budgeting while maintaining human oversight in strategic planning will gain a competitive edge," say the report's authors. "The key is knowing where AI should lead and where human intuition remains indispensable." Data is a key component of Goldman's AI strategy, which Argenti calls a three-leg stool that should also represent the AI technology itself and the people who use it. Good quality data is needed for the right output from LLMs, but changing people's behavior is equally important. "It's about amplifying capabilities and in the hands of the best people, I think you're going to get the best results," says Argenti.

Inside Goldman Sachs' Big Bet on AI at Scale | PYMNTS.com

Goldman Sachs is rolling out and scaling artificial intelligence (AI) capabilities across its business, as the investment bank harnesses the benefits of productivity and efficiency from this fast-moving technology. "For us, this year is a story of scale -- scale of adoption, scale of use cases," said Belinda Neal, Goldmans' chief operating officer of core engineering, head of product management and head of engineering partnerships. "With generative AI, we've been really excited about some of the new potential of the technology to deliver efficiency and productivity benefits across the firm," Neal said in an exclusive interview with PYMNTS. PYMNTS data is showing a growing return on investment (ROI) as businesses are starting to reap the benefits of AI capital investments. Based on a survey of 540 C-suite executives, 57% of information firms report very positive ROI from their GenAI deployments. Other sectors are seeing varying returns. Neal said the investment bank is taking a multi-pronged approach to AI: Neal said a key enabler of AI throughout the company is the creation of "AI champions," a team of employees within each business group tasked with finding the most effective use cases for AI in their line of business. These AI champions form the "connective tissue" across the firm, she added. "That is an absolutely critical part of how adoption is going to be driven and where we're going to see the highest value from" in its AI efforts. One of Goldman's internal AI applications is the GS AI Assistant, a generative AI chatbot rolling out to its bankers, traders and other workers. It can do things such as summarize documents, draft emails, create bullet points for a speech and more. The assistant will eventually transform into a smarter sidekick with the traits of a seasoned Goldman executive. Neal said about 10,000 employees currently have access, which will be rolled out across the company this year "subject to governance." GenAI chatbots have a tendency to hallucinate, or make things up in a convincing way, which is a security risk for any company, but especially the highly regulated financial services industry. Neal said there's still a human reviewer of the chatbot's responses, and employees are trained to be mindful of these hallucination risks. It is "absolutely something that we're looking to continue to improve over time, through the use of new models (and) new practices," she said. A Goldman priority is using AI to enhance the performance of its engineering group. Neal said the firm has more than 12,000 developers, which constitute a quarter of its total workforce. That means productivity and efficiency gains achieved in this group would have a large impact on the company. "The developer use case is one of the early use cases where we've really seen large-scale adoption and large-scale benefit," Neal said. "What we're going to start to see over time, especially over the course of the next 12 to 24 months, is more scale and adoption of those capabilities." But more than equipping its developers, Goldman wants them to stay on top of AI's advances. "We really want to be able to ... employ that best-in-class technology as quickly as possible into some of these platforms so that we can enable our developers and our individuals across the businesses to use the best technology," Neal said. A third priority use case revolves around governance and how the firm thinks about risk and control, Neal said. "We've doubled down on our focus around ensuring safe and secure AI deployment." Goldman's Chairman and CEO David Solomon has said that CEOs see the benefits of AI. "The more complicated thing that CEOs are wrestling with is how the technology can be deployed to significantly change operational processes," Solomon said at the Cisco AI Summit in Palo Alto, California, in January. "But to do that, you have to fundamentally change your processes, and that's a very hard thing for organizations to do." As Goldman rolls out its large language models, it is already preparing for the advent of AI agents. Neal said Goldman is taking a holistic approach by thinking of agentic AI as reinventing entire workflows, not just single tasks. These agents will also become reasoning models over time, she said. Looking ahead, Neal said there's a lot of interest in the firm around AI use cases for document or life cycle management. "I'd say this is just the next generation of us being able to use technology to drive value to the business in terms of how we're operating efficiently, but also how we're delivering best-in-class service for our clients," Neal said.