This event is canceled for October 2021 and will be rescheduled at a later date. Please stay tuned!
The “Principles of AI” program was offered for the first time in October 2020. The success of this inaugural program means that AESA will offer an extended version of the program again at a later date.
The program will be taught virtually by Duncan Simester in a sequence of four 2-hour sessions.
Session 1 will begin by describing what AI is and why it is different. Duncan will then describe what types of problems AI can solve, and what data is required to solve them. We will conclude by proposing potential topics and identifying potential pitfalls.
Session 2 will focus on the role of experimentation in AI. We will first describe why experiments are an important part of AI, and then define what an experiment is. We will use a short exercise to demonstrate the power of experiments, before asking the participants to identify potential obstacles to experimentation in their organizations.
Session 3 will focus on Natural Language Processing (NLP) and Image Recognition. After defining what these concepts are, we will show how AI solves NLP and Image Recognition problems. Modern methods for solving these problems are both very powerful, and easy to use. Many organizations with modest AI capabilities can now use these tools to solve a wide range of problems. Moreover, although they solve different problems, NLP and Image Recognition use a very similar process. This is likely to become the standard process by which sophisticated AI methods graduate from the research lab to widespread adoption.
Session 4 will begin by focusing on two questions:
-
Can AI help ESAs disrupt, or will it disrupt ESAs?
-
What will be the first applications of AI for ESAs?
Participants will be asked to give thought to both questions before attending the session. Duncan will moderate discussions on each question and offer insights from his experiences about where other organizations have started. The discussion will extend to the challenges of developing an internal AI capability, and how organizations have been able to successfully overcome these challenges.
We will ask participants that did not attend the October 2020 program to attend all four sessions in this year’s program. The content in the later sessions will build upon the content in the earlier sessions.
For participants that did attend the October 2020 program, we offer an option to attend just Sessions 3 and 4. The content in Session 3 will be entirely new this year, and the content in Session 4 is a structured discussion that is also likely to cover new material.
