In this enlightening talk at MachineCon USA 2023, Amaresh Tripathy, Managing Partner at AuxoAI, delves into the rapidly evolving world of GenAI. With his extensive experience leading data analytics businesses and a hands-on approach to AI, Tripathy addresses the pressing question that is on every CEO’s mind: “What should we do about GenAI and can we afford to wait?”
Tripathy shares valuable insights from his work with clients, academia, and AI development, providing crucial guidance for data analytics leaders navigating the complex landscape of GenAI. His talk is a treasure trove of knowledge for any enterprise looking to make informed, no-regret moves in the world of GenAI.
About the Speaker:
Amaresh Tripathy is a seasoned leader in the field of GenAI. He is a managing partner at AuxoAI, a GenAI startup that builds business co-pilots. In addition to his role at AuxoAI, Tripathy serves as a senior advisor for Genpact, where he previously led a team of 15,000 data scientists. Before his tenure at Genpact, he led the PwC US data analytics business.
Tripathy is also deeply involved with the academic community. He is a founder and serves on the advisory board of the School of Data Science at the University of North Carolina at Charlotte. He also teaches a class on applied machine learning. Tripathy holds a graduate degree from the University of Texas at Austin.
The Challenge of Confusing Times
Amaresh begins by acknowledging the challenges of starting a venture during confusing times. He notes that all of us are in some ways in the same set of confusing circumstances, where some things are good, some things are fantastic, and some things are definitely issues that need to be addressed. Amaresh emphasizes the importance of understanding what we know, what we don’t know, and what are the “no regret” moves in such times.
The Evolution of Information Technology
Amaresh refers to a 2016 paper that looks at the evolution of information technology from an evolutionary biology framework. The paper suggests that we are in a transition phase, and it’s okay not to have all the answers. Instead, we should focus on understanding what we reasonably know, what we don’t know for sure, and what are the “no regret” moves.
The WINS Framework
Amaresh introduces the WINS framework (Words, Images, Numbers, Sound) as a way to understand the impact of technology on knowledge work. He suggests that symbolic work, where information is represented symbolically (like words, images, numbers, sound), will be more impacted by technology than other types of knowledge work.
The Disruption Sequence
Amaresh suggests that disruption will occur in a particular order: from the task level to the functional level, to the enterprise level, and finally to the industry level. He uses examples from the tech industry to illustrate this sequence, such as the task-level disruption caused by AI tools like GitHub Copilot.
The Impact on Economics
Amaresh argues that this sequence of disruption will change economics. He uses a simple model to illustrate how technology can increase revenue and reduce costs in various industries, particularly those that involve a lot of symbolic work.
The Impact on Different Industries
Not all industries will be impacted equally by this disruption. Amaresh presents a two-dimensional framework, considering the degree of digitization and the cost allocation for symbolic work. Industries like entertainment, software, professional services, financial services, and education are likely to be more disrupted, while others like agriculture, logistics, hospitals, and pharma may be less affected.
The Role of AI
Amaresh discusses the role of AI in this disruption, arguing that AI can either replace labor or amplify human potential. He uses examples from the education sector, like Khan Academy’s teacher co-pilot, to illustrate how AI can enhance human capabilities.
The Future of Work
Amaresh presents a vision of the future of work, where AI acts as a “co-pilot” for various functions. He uses the example of a med tech manufacturer to illustrate how an AI co-pilot could help with tasks like sequencing work, improving efficiency, and executing tasks.
The Technical Architecture
Amaresh discusses the technical architecture required to support this vision of the future. He suggests that it will involve a combination of traditional machine learning, generative AI, data pipelines, and a dialogue interface.
Despite the exciting potential of AI, Amaresh acknowledges that there are still many open questions. These include issues around the ownership of AI-generated content, the scalability of AI tools in an enterprise context, and the cost of using different AI models.
In conclusion, starting a venture in confusing times can be challenging, but it also presents unique opportunities. By understanding the evolving landscape of information technology and the potential of AI, entrepreneurs can navigate these challenges and seize the opportunities that lie ahead.