UCSD’s New SCIDS School: All About the Apps
Some research universities excel in exploring intellectual possibilities and postulating abstract theories. Those terms aren’t likely to be associated with the University of California, San Diego’s new School of Computing, Information and Data Sciences (SCIDS), which is all about taking the power of data and computer science and applying it to solve problems in the real world.
SCIDS was officially formed last month by the union of the San Diego Supercomputer Center (SDSC) and the Halıcıoğlu Data Science Institute (HDSI). As the fourth school on the rapidly expanding UC campus, SCIDS is generating quite a bit of excitement from the HPC and data science communities, particularly as it relates to AI research.
SCIDS will be researching and applying techniques that are so new that they’re not even in textbooks yet, according to Rajesh Gupta, the interim dean of SCIDS and a distinguished professor of computer science and engineering at the university.
“AI and generative AI fall into that [category], where the most recent advances are maybe six months old,” he said during an address at Solix Technologies’ SolixEmpower conference held at UCSD’s Qualcomm Auditorium.
“Attention [i.e. the “Attention Is All You Need” paper] is two or three years old, Transformers a bit more, CNNs a bit older, but reborn,” Gupta continued. “And the topics that are being taught to our students or must be taught to our students, in addition to probability, statistics, and computer science and others are rapidly evolving so fast that the books are not going to be there.”
Even more importantly, he added, the people who are advancing the state-of-the-art in AI and GenAI aren’t in the university. Some don’t even have college degrees, Gupta said.
“I have been a professor for 30 years. We are constantly keeping in touch with what’s going on,” Gupta said. “But if you talk to me about the latest cloud stack or the GenAI stack, or a blockchain with authentication built in or KYC [know your customer] built in and so on–that’s not in any university.”
Keeping up with the rapid pace of change is a challenge for every research university. SCIDS intends to stay on the cutting edge by opening itself up to work with anyone who’s seeking solutions to tough challenges, said Frank Würthwein, the director of the SDSC and a professor at HDSI.
“The SDSC was founded about 40 years ago,” he told the audience at the SolixEmpower conference. “Throughout this 40 year history, we’ve become and have always been excellent in translating computational science, data science into practical applications. We’re fairly agnostic about the domain. We’re very, very broad across… all domains.”
The merger of SDSC and HDSI into SCIDS is timely, as the potential for translational science has never been higher, Würthwein said. And the potential is high for two main reasons: The end of Moore’s Law, and the diversification of scientific funding beyond government and into the private sector.
“Just give you one example, the total volume of money spent by the Gates Foundation is roughly the same, within 10% or so, of the National Science Foundation,” he told his audience. “That’s a staggering number.”
So how does a university fit into this? Würthwein continued:
“I believe that in addition to the traditional vehicles of engagement with industry, which are philanthropy and workforce, we should explore what else we can do together. Are there things that you have pain points, that we can help solve those pain points, that are in our mutual interest? Are there ways that we can build consortia with you that can solve your pain points, our pain points, and move humanity forwards?”
Würthwein said he’s ready to have discussions about these pain ponits “with anybody who wants it.” Collaboration “is in our DNA,” he said. There’s not a single unit on the UCSD campus that the SDSC has not collaborated with over its 40-year history, he said.
“So if you are looking to work on some kind of computational problem, we might be able to bring in other people from campus that have the necessary domain expertise in addition to our expertise, which lies squarely in translating,” he said. “We don’t do foundational research as much at SDSC. We translate foundational research into applications.”
One of the groups tapping into the expertise of UCSD and SCIDS is Solix Technologies. The Silicon Valley software company, which is transitioning from providing data archiving solution for large enterprises to providing a full data platform to power AI development, is collaborating with UCSD and helping to fund several research projects, according to John Ottman, Solix’s executive chairman.
The collaboration between Solix and UCSD began when the folks at Solix realized that generative AI was going to fundamentally change the nature of how humans interact with computers. While tools like ChatGPT could unlock incredible capabilities, they also exposed a fundamental challenge and barrier: the lack of data governance.
“So we went on a search and we tried to find the best place to do that [research] and we ended up here at UC San Diego,” Ottman said during his keynote address at SolixEmpower last week. “And what we found here is tremendous skill and expertise, particularly in the area of data at both the San Diego Supercomputer Center and the Halicioglu Data Science Institute, and now, of course, the new School of Computing, Information and Data Science.”
Solix has three projects ongoing with UCSD. The first project revolves around devising a system for minimizing privilege for AI apps, and is being run by Haojian Jin, an assistant professor at HDSI. The second involves research into AI safety and security, and is being led by James Short, the lead scientist and director at the SPARK AI Consortium. And the third involves using AI to predict cancer in X-rays.
“What’s happening here at UC San Diego is extremely exciting for us,” Ottman said. “This is becoming a center of AI expertise and we’re really excited to be part of it.”
Related Items:
Tips on Building a Winning Data and AI Strategy from JPMC
Solix Hosting Data and AI Conference at UCSD
Top Five Reasons Why ChatGPT is Not Ready for the Enterprise