Fabio Casati

University of Trento

Date: 20 March 2023

Time: 16:00 (CET)

Title: Trustworthy Enterprise AI: crossing the chasm among LLMs, AI services, and Enterprise AI systems

Abstract: Companies are devoting significant effort -and budget- into powering their processes with AI. The trend and enthusiasm towards AI-powered processes has skyrocketed in recent months with the advent of powerful large language models (LLMs), that open new possibilities and drive users interest and expectations. 

Leveraging the benefits of AI in an enterprise context however is far from trivial. AI are still probabilistic systems that operate in somewhat mysterious ways, often with a business value that is hard to measure, that may fall into major blunders, and that may be harmful to people of not used properly.  

The distance that still exists between the power of modern AI and the impact it has had so far on the enterprise is surprising to many, given that recent models report super-human performance in many tasks seemingly related to what AI is asked to do in the enterprise.  

In this talk, I will present some of the reasons for the disconnect between the (perceived) quality of AI services on one side, and the transformational impact they can have on AI-powered enterprise processes on the other.  

I will focus specifically on two key challenges. The first is the way we measure and communicate the quality of AI services: the way many companies approach this problem is fundamentally wrong, and this leads to wrong process design decisions and to an improper use of AI in business processes. 

The second is that AI service and technology providers often focus on the benefits of AI services once in production, but underestimate the importance of simplifying the journey that process owners have to follow to get there and the perceived risk - or lack of trust - in AI. I will present concrete examples of both issues, and show (i) how we can rethink the notion of AI quality in enterprise processes and (ii) how we can simplify the journey to adoption and deliver trustworthy AI-powered processes.

Speaker Biography: Fabio Casati is a Principal Machine Learning Architect at Servicenow as well as technical lead for the AI Trust and Governance group in Servicenow research. Fabio focuses on designing, architecting and deploying AI-powered workflows for enterprise customers. On the research side, he is working on AI applied to workflows and on quality in AI.  Fabio is also Professor at the University of Trento. In that role, he started research lines on crowdsourcing and hybrid human-machine computations, focusing on applications that have direct positive impact on society through tangible artefacts adopted by the community. Prior to that, he was technical lead for the research program on business process intelligence in Hewlett-Packard USA, where he contributed to several HP commercial products in the area of web services and business process management. He co-authored a best-selling book on Web services and is author of over 250 peer-reviewed papers and dozens of patents.

Homepage: https://www.linkedin.com/in/sphoebs/

Video Recording

Due to copyright limitations, the recording of this talk is not publicly available.