Navigating Generative AI Challenges: Insights from Infosys AI Head

Navigating Generative AI Challenges: Insights from Infosys AI Head

Simplifying the Blog Post: Infosys and Generative AI

Introduction

Companies are keen on using generative AI tools like ChatGPT and GitHub Copilot for optimizing code and automating business processes. Gary Bhattacharjee, VP of Data Strategy and AI at Infosys, shared insights in an exclusive interview with Analytics India Magazine about the challenges associated with adopting generative AI.

Challenges in Generative AI

  • Predictability in Code Generation:
  • Bhattacharjee emphasizes the challenge of predictability in code generation. While AI tools can write code, the most optimized version often requires human decision-making, especially in areas like Python for Fibonacci’s sequence.
  • Code IP Concerns:
  • Bhattacharjee discusses the challenge of code intellectual property (IP). Different perspectives exist on this, with alternatives like CodeGen from Hugging Face offering better options for companies concerned with code IP.
  • Complexity in Code Translation:
  • Translating code, especially in languages like COBOL to C, poses challenges in efficiently handling code fragments and variable structures. Different approaches, such as those used by IBM, involve unique algorithms and rules-based computational search methods.
  • Generative AI in Indic Languages:
  • Infosys is actively working on generative AI in Indic languages.
  • Tokenization Challenges:
  • Addressing concerns about tokenization costs, Bhattacharjee suggests alternatives like Eigenvectors and Eigenvalues to handle token estimation differently, preventing language models from becoming overly large.

AI at Infosys

  • Operational Use of AI:
  • Infosys leverages AI for operational and administrative improvements, including expense management, accounts receivables payments, and client solutions that can reduce costs significantly.
  • AI in Workforce Transformation:
  • Bhattacharjee shares Infosys' focus on transforming the workforce's skill sets from coding to algorithms, emphasizing higher-end mathematical understanding and developing algorithms for complex quant models.
  • Concerns and Transformations:
  • While acknowledging the ongoing transformation, Bhattacharjee believes Infosys will continue hiring fresh talent, partnering with institutions like IITs and IIMs to build a robust tech talent pool.

Conclusion

In navigating the challenges of generative AI, Infosys remains at the forefront, using AI not only to enhance its own operations but also to offer innovative solutions to clients, emphasizing a shift from coding to advanced algorithmic skills.

Read more