On Friday, Nov. 3, the Connecticut Advanced Computing Center hosted a workshop on the current uses and future prospects of artificial intelligence from both academic and industry perspectives. For four hours in the Innovation Partnership Building, various speakers represented widely-recognized companies such as Pfizer and Travelers to discuss their companies’ application of AI within an ever-increasing market and the challenges that arise as a result.
At this point in time, bio-pharma, aerospace and defense companies are reluctant to heavily use AI due to the humanistic services their industries provide.
Dr. Kishore Kumar Reddy spoke on behalf of RTX Technologies, an industry leader in aerospace and defense. Having more than 25 peer-reviewed publications under his belt, he is well-respected within the field. Reddy is always looking ahead in his head role at RTX’s research center in East Hartford. He highlights the potential for AI to be used for cockpit automation, spacesuits, cybersecurity, fighter piloting and fleet and constellation management. He juxtaposes this potential with the safety risks involved with these tasks and machines. In one way or another, humans must negotiate these challenges that AI presents, hence Reddy’s remark: “It’s amazing technology, but it comes with a lot of baggage.”
Great minds come together at this research center to create a hub of collaboration, a business-centric focus and an impact on innovation in the industry. Part of the center’s reluctance to use AI stems from the fact that researchers are currently working at a smaller scale than other industries. For example, Tesla’s autopilot technology works with 15 million miles of data per day and processes this data with a supercomputer containing 5,760 graphics processing units. Comparing this with the 80 GPUs that make up the RTXRC’s supercomputer, you can see the technological gaps between different companies, and why Tesla’s autopilot is being used out on the road while aerospace and defense tasks still largely use humans. Reddy finished his segment by saying that he would want an autonomous aircraft he’s flying on to be certified. Safety is undoubtedly RTX’s top priority.
Dr. Seda Arat, a computational toxicologist for Pfizer, spent her time detailing the laborious process of how a drug is manifested from an idea into a market-viable drug. The journey is comparable to getting a law passed in Congress, except the regulations are even stricter.
The desire to use AI to help make this process more efficient is definitely present following the pandemic, as it can help screen thousands of drugs a week without removing scientists from their other work obligations. Also, 10,000 components or more are often synthesized into one drug; the possibility of automating this would be game-changing.
In terms of regulatory approval for drugs, she notes how there is developing AI for target classification and side-effect prediction for drugs, although currently in its “kindergarten stages.” The hesitation to use this AI in its current state was felt — as it was with aerospace and defense — because there is no room for error when people are involved. While AI can often supplement human intervention, it cannot be relied on in the same capacity that Tesla autopilot can, at least not yet.
In addition to RTX and Pfizer, there were sponsored segments from Cyberdyne Systems led by speakers featured throughout the workshop. Overall, this event proved optimistic by those who will be guiding AI use in human-centered industries with patience and caution.