Goldman Sachs is strategically integrating artificial intelligence across its organization, recognizing the continuing evolution and uncertainties associated with this technology. Currently, 50% of the firm’s 46,000 employees have access to AI tools. Marco Argenti, the Chief Information Officer, emphasized the need for the entire organization to adapt to AI’s potential, remarking, “We have the entire organization that needs to somehow re-tune and re-tool itself for AI.” He noted that the company is being deliberate in managing the transition and fostering change.
The firm is exploring the use of agentic AI, though it has not yet fully implemented it across all operations despite its potential benefits in automating crucial tasks such as compliance checks and customer transaction processing. Recent challenges and setbacks from hastily executed projects have highlighted the necessity for these AI systems to undergo thorough training and utilize quality data to avoid inaccuracies and “hallucinations” in their outputs. Goldman Sachs is still identifying what additional safeguards are needed to ensure the responsible use of agentic AI.
Approximately 25% of Goldman Sachs employees are engineers, and this group was prioritized when initiating the rollout of generative AI tools. Argenti made AI coding assistants like GitHub Copilot and Gemini Code Assist available to these engineers, encouraging creativity through competitions reminiscent of the reality television show Shark Tank. He assesses the return on investment from these tools based on factors such as usage frequency and the acceptance rate of AI-generated code.
The implementation of the GS AI Assistant last year marked a broader initiative to integrate generative AI within the company, eventually expanding its access to 10,000 employees, including bankers, traders, and asset managers. This assistant is designed to summarize documents, draft emails, analyze data, and create tailored content, with plans to make it accessible to nearly all employees by the end of 2025. Research from Vlerick Business School lends credence to Goldman’s cautious position on AI, particularly regarding its role in corporate budgeting. The research indicates that while AI can manage tactical budgeting effectively, human oversight is imperative for strategic planning to ensure that immediate financial decisions align with long-term business goals.
The study conducted simulations where experienced managers allocated budgets for a fictional automotive parts manufacturer, comparing their decisions to those made by an AI algorithm. Results revealed that AI excelled in budget optimization when strategic frameworks were clearly defined. However, when key performance indicators diverged from strategic objectives, AI’s effectiveness diminished. The authors conclude that firms leveraging AI’s strengths in tactical budgeting while preserving human oversight in strategic areas will likely maintain a competitive advantage.
Data quality is integral to Goldman’s AI strategy, which Argenti describes as a three-legged framework encompassing AI technology, data integrity, and the users. High-quality data is essential for delivering accurate results from large language models (LLMs), but the adaptation of employee behavior is equally vital. Argenti states, “It’s about amplifying capabilities, and in the hands of the best people, I think you’re going to get the best results.”