Catholic Tech

Dr. Kevin Greenman on Harnessing AI for Scientific Discovery

Jun 10, 2024
News

Dr. Kevin Greenman, CatholicTech professor, recently presented at The Society of Catholic Scientists at their 2024 Conference: “Critical Issues at the Interface of Science and Faith.” His topic, “Harnessing AI for Scientific Discovery,” focused on the achievements, opportunities, and ethical implications of AI development.

Greenman began by noting that AI has become much more mainstream in society; whether it’s creating images or using large-language models, AI tools are being used in daily life without any need of technical training for people to use. There are many good implications, like using ChatGPT to create a grocery list, or using Magisterium AI to answer a question about the Eucharist.

But is there an existential risk in AI development? There is a diversity of opinion about whether to stop the development of AI. Many have voiced not a pause in research in general, but “merely a stepping back from the dangerous race to ever-larger unpredictable black-box models with emergent capabilities.” Since not all AI research is the same, there are many practical applications of AI that do not need to be paused at this time, such as programs that help in business, customer service, marketing, and more.

Dr. Greenman noted that in Stanford’s yearly AI Index report, they have recently made an additional chapter specifically for AI in science and medicine. There have been many applications in these fields that have been very beneficial and practical. AI is now being used in almost every field at the moment, making it relevant for the area of science as a whole.

One of the recent programs that has helped in the field of medicine that Dr. Greenman mentioned was AlphaFold, a program that is said to have solved the problem of protein folding. Greenman explained, “In biology, you have DNA that encodes for proteins . . . but knowing the amnion acid sequence is not know what the protein is going to do, because what the protein does is very dependent on its structure . . . and it’s a very non-trivial problem to go from that amino acid sequence to the structure.” AlphaFold was able to solve this problem for the first time, achieving high accuracy after just a few years. AI programs are also being used to model for new drugs in research and development, using the actual 3D coordinates of the atoms in a molecule. GNoME AI has also been beneficial in accelerating the discovery of solid electrolytes, and millions of new materials were discovered in material science.

What does this mean for how we now do science? Dr. Greenman posed this question with an article from Microsoft Research Blog, which set out five paradigms of science: (1) empirical, direct observation; (2) equation, theoretical models; (3) numerical simulation; (4) data-intensive discovery; and now (5) AI emulation of simulations. These paradigms have been building upon each other throughout the history of scientific research, and we can use AI now to learn from previous research and simulations to continue to advance science. Greenman remarked, “I don’t think these paradigms are mutually exclusive in any way. I think they all build on one another, they all co-exist, and they all complement one another. This isn’t to say that AI is going to replace any previous parts of science, but rather that it’s going to build on it and complement it.” AI has the ability to aid in all stages of the scientific method, whether it is formulating a question, or predicting experiments to refine results.

However, there are potential shortcomings in the development of AI, large-language models can be used and then not reviewed enough, leading to scientific misconduct, or potential ethical misuses of AI. Greenman addressed this, noting how important it is for scientists to uphold their conduct and the authenticity of their work, saying, “Right now, I think we live in a time in society where there is potentially degrading trust in science, and things like this only contribute to that degradation, so it’s very important for us, as scientists, to take all of our duties seriously.”

Dr. Greenman concluded his talk by discussing the role of AI in the work of science alongside our human dignity: “A human contribution to science is necessary . . . Science is very important to us as humans and what it means to be human, and it is important for us to be participating in that process of science rather than giving over that responsibility to an AI.” We need to be able to question ourselves about what the most important concerns in our society are to work on, and it is part of our human experience to be able to wonder, and science satisfies this wonder. Additionally, he said, “being able to understand the world or try to understand God’s creation has certainly been a way for me to experience and enhance my human dignity, and I don’t think that’s something I should try to give away to an AI tool.” Work, according to the Catholic Church, is something that man must do because “the Creator has commanded it and because of his own humanity, which requires work in order to be maintained and developed” (JPII, Laborem Exercens, 16).

There is great potential for the practical uses of AI, in our daily lives but also to aid in scientific advancements, but we need to use it responsibly. Catholic scientists, Greenman concludes, should be leading this discussion.

For the full presentation, watch it here.