(149c) External Risks Created By Reliance on Open-Source Artificial Intelligence Libraries

The most powerful open-source AI toolkits such as Pytorch and Tensorflow are developed by Meta (formerly Facebook) and Alphabet (formerly Google) respectively. The open-source AI community, including the chemical industry, have largely come to rely on these toolkits to develop high impact AI pipelines today. It is clear why these tech companies would want to develop these tools for in house use but not so clear as to how long they plan to make the toolkits available to the public or why they do in the first place.

The questions I want to explore are these:

1. What risks are we facing by relying so heavily on these toolkits?
2. Should we mitigate those risks?
3. How can we mitigate those risks?