Universities Adapt to Keep Pace with AI’s Rapid Development
In the rapidly evolving domain of artificial intelligence (AI), academic institutions find themselves at a crossroads. Struggling to keep up with the significant computing resources of Silicon Valley’s giants, universities are innovating their approach to AI research. They are increasingly focusing on areas that require less computational power while also striving to expand their computing capacities to support larger, more complex models.
“Academic institutions are scrambling to get access to compute,” highlighted Hod Lipson, chair of the mechanical engineering department at Columbia University. The stakes are indeed high as universities endeavor not to be left behind by private companies that have dominated generative AI research, thanks to their considerable access to data and financial resources. Companies like OpenAI and Google have been at the forefront, developing models like GPT-4 and Gemini, with some projects costing upwards of $100 million.
This surge in private sector activity comes at a time when the tech industry continues to experience a shortage of qualified professionals, underscoring the critical role universities play not only in educating the next generation of AI talent but also in shaping the future path of AI development. “I think it’s important that industry is involved in this. It’s important that government is involved in this. But to balance both of these forces, we also have to have other people, open source, academia, who have a say in where and how this technology is used,” Lipson explained, emphasizing the broader societal benefits of AI from healthcare to energy.
To combat these challenges, universities are not only investing heavily in their own computing resources but are also exploring collaborative efforts. The University at Buffalo, for instance, will host the Empire AI computing center, part of a state-wide initiative involving several universities, including Columbia and Cornell. New York Gov. Kathy Hochul’s office remarked on the increasing concentration of AI resources in the hands of large corporations, highlighting the need for broader access to foster safety and societal benefit in AI development.
Moreover, academic institutions are leveraging partnerships with national laboratories and industry to enhance their computational capabilities. The University of Chicago, for example, has expanded its resources through a partnership with Argonne National Laboratory, with further plans to grow its computing infrastructure. “I definitely think it’s something we should be planning for,” stated Hank Hoffmann, chair of the university’s computer science department.
As the boundary between academia and industry becomes increasingly porous, universities in tech hubs are finding new ways to nurture their talent while accessing greater resources. In some cases, academic researchers also engage in industry work, blending the best of both worlds. This symbiosis is evident at institutions like the University of Washington, where programs facilitate such dual engagements, according to Dan Grossman, vice director of the computer science and engineering school.
Despite these efforts, the challenge of competing with industry’s deep pockets remains. Universities are therefore becoming more strategic in their research focus, aiming for innovative applications of AI rather than attempting to build the next big model. From Cornell University’s emphasis on leveraging large language models (LLMs) for specific use cases to the Massachusetts Institute of Technology’s work on AI for code development, the academic world is finding its niche in applying AI to solve real-world problems.
This strategic pivot does not come without its challenges, including the procurement of cutting-edge hardware. “Even if you have the money, just getting your hands on top-of-the-line GPUs is actually quite, quite hard,” shared Armando Solar-Lezama, a professor and associate director at MIT’s Computer Science & Artificial Intelligence Laboratory.
As universities continue to adapt to the rapidly changing landscape of AI, their role in developing the technology responsibly and equitably becomes even more critical. Through strategic investment, partnerships, and a focus on application-driven research, academia is finding ways to ensure it not only keeps pace with but also significantly contributes to the future of AI.
