Innovation: Alumnus Filippo Pavone on Adopting AI at Scale

Professor Silvia Pulino welcomed alumnus Filippo Pavone to her Early Entrepreneurship class on March 9, 2023. Filippo graduated from JCU in 2016 with a B.A. in Business Administration and a minor in Marketing. While still at JCU he founded Emaut, an e-commerce platform for astronomers in Europe, which he built to over 2500 members. After graduation, Filippo managed portfolios of companies at Skyscanner, at G2 and at V7 Labs, a leader in AI machine training. In parallel, he invests in AI companies as part of an angel investor group.

Filippo Pavone

Filippo’s presentation provided a valuable structure for students to develop their understanding of the Artificial Intelligence world. He first described the field and its various subfields: cognitive computing (large data), computer vision (CV), machine learning (ML), neural networks, deep learning (DL) and natural language processing (NLP, “home” to ChatGPT).

He then explained how a model is trained: data is collected, then annotated (or labeled), and used to train the machine; repeated iterations of the process lead to building a model catalogue that can then be used in testing. He then focused specifically on his area of expertise, computer vision, and used images to demonstrate how the computer learns to recognize them using Convolution Neural Networks.

An ideal candidate for CV adoption is the healthcare sector, where computer vision can greatly assist doctors, for instance, in early detection of tumors. Other applications are in manufacturing, where CV can replace the routine, repetitive visual controls currently carried out by humans, or in the agricultural sector, where machines can be trained to recognize fruit that is ready to be picked.

This brief glimpse was enough to give students an understanding of the myriad of possible applications, and a sense of the deep revolution that has just begun. Equally interesting are the questions raised. What will be the impact on employment? Many jobs will disappear, and new ones will be created; the balance may be a favorable one, but the likely fact that the new jobs will not be picked up by the displaced workers raises important social issues.

Another interesting question relates to the impact on global development. In developed nations there will surely be a rush to deploy AI applications designed to replace low skilled human labor, but in countries where labor is cheap, the incentive will be much less pressing. Will this mean a widening of the technological gap between developed and developing economies?

Questions are also raised on the regulatory front. AI applications are built on vast amounts of data, but the collection of such data can be problematic. In the healthcare world, for instance, each piece of information contains metadata which makes it impossible to share across hospitals, so hospitals have to build their own data sets and cannot benefit from a more widely spread pooling of information.