Published on February 27, 2024, 6:17 pm

Title: “Citi’S Strategic Shift: Embracing Generative Ai For Enhanced Efficiency And Productivity”

Citi, like many major financial institutions, is strategically evaluating numerous use cases for generative AI to determine their business impact and associated risks. However, the bank is swiftly progressing on specific initiatives. One such project involves implementing GitHub Copilot across all developers within the organization, totaling approximately 40,000 employees, by mid-April. According to Shadman Zafar, CIO of personal banking and wealth management at Citi and the lead for generative AI initiatives, this move is expected to streamline processes significantly by leveraging reusable code from the bank’s repository.

Moreover, Citi is harnessing generative AI technology to revamp outdated systems and automate initial drafts of compliance evaluations. Zafar expressed his optimism about the sustained influence of this technology on reshaping work methods over the forthcoming decades.

Notably, Citi isn’t alone in its swift adoption of generative AI solutions among financial institutions aimed at enhancing developer productivity. Concurrently, banks such as Goldman Sachs are experimenting with similar technologies like providing autocomplete-style suggestions using generative AI during coding processes.

Zafar detailed how Citi transitioned from a pilot program involving 250 developers to broader implementation encompassing 2,000 programmers adopting GitHub Copilot daily. The choice of GitHub’s Copilot Enterprise was motivated by its enhanced control features over inputs and outputs compared to other options.

While embracing generative AI offers substantial benefits, one concern among developers is the potential risk of plagiarism when utilizing algorithm-generated content sourced externally. To mitigate this issue, GitHub Copilot incorporates guardrails that restrict such occurrences within Citi’s development environment.

Furthermore, Citi employs retrieval augmented generation methodology to ensure accuracy by retrieving pertinent data and documents while utilizing its extensive code repository for guidance. This approach has been effective in maintaining content authenticity and aligning with industry standards.

Beyond these applications, Citi looks into various other generative AI use cases spanning operations automation, fraud detection, customer service enhancement, and efficiency improvements in office tasks. The bank has established an AI lab dedicated to scrutinizing proposed ideas through technical feasibility evaluations followed by rigorous risk assessments concerning explainability and transparency levels.

Training employees on leveraging generative AI tools showcases Citi’s commitment to upskilling its workforce in readiness for future technological advancements in this domain. As Zafar acknowledges the prevalent hype surrounding generative AI technologies presently continuing through an evolution phase towards more efficient models in the near future.


Comments are closed.