This Goldman Sachs coder does not sleep or get paid
What's the story
Goldman Sachs is integrating an artificial intelligence (AI) coder, Devin, developed by London-based start-up Cognition. Marco Argenti, Goldman Sachs's Chief Information Officer, said they will start using Devin to perform tasks on behalf of their developers. The bank plans to scale usage of the AI tool, which could boost productivity by three to four times (compared to older assistants) for software teams.
Strategy
It will handle mundane tasks under human supervision
The AI coder will take care of mundane tasks like updating legacy code under human supervision. This comes as Goldman Sachs is also rolling out its GS AI Assistant, a company-wide generative AI platform already used in investment banking, wealth management and research roles.
Profile
A look at Cognition
Devin is the brainchild of Cognition, a start-up founded in late 2023 and already valued at nearly $4 billion. The company is backed by Palantir's Peter Thiel and Joe Lonsdale among others. Unlike generative AI systems that assist with emails or research, Devin represents a new class of "agentic AI" designed to work autonomously and execute entire workflows from start to finish.
Job impact
Adoption of AI coding tools raises job loss concerns
The adoption of AI coding tools like Devin has raised concerns over potential job losses in the banking sector. Bloomberg Intelligence had predicted that banks could cut as many as 200,000 jobs worldwide over the next three to five years due to AI adoption. However, Argenti envisions a future where humans and AIs work together in a "hybrid workforce."
Productivity gains
Here's what Goldman Sachs's president said
Goldman Sachs President, John Waldron, had earlier said that developer teams using AI tools have seen a 20% increase in productivity. This suggests that while these tools are making workers more efficient, they could also lead to fewer employees in the future. MarketWatch has warned that traditional analyst roles could shrink dramatically due to automation of routine coding and research tasks.