UF and other research universities will fuel AI. This is the reason

In the global AI race between small and medium-sized competitors and major competitors, established companies and new players, and the use of ubiquitous and niches, the next huge leap isn’t about faster chips or improved algorithms. If AI agents are already filling up much of the information on the Internet with a vacuum cleaner, the next big uncertainty is where they find the next big data.

The answer is not Silicon Valley. The key to maintaining a global competitiveness against China is across the nation of our major research universities.

To teach AI systems to “think,” you need to draw a huge amount of data to build a model. At a recent conference, Ilya Sutskever, former Chief Scientist at Openai, is called Data. Just as fossil fuels run out of money because they are not renewable, he said we are running out of new data to continue to drive AI benefits.

But much of this idea assumes that AI was created by private Silicon Valley startups, for example. The history of AI is actually deeply rooted in American universities dating back to the 1940s, with early research laying the foundations for the algorithms and tools used today. Computing power using these tools was created very recently, but the foundation was laid in universities rather than in the private sector after World War II.

Contrary to the “fossil fuel problem,” I believe AI has its own source of renewable fuel. Data and expertise generated from comprehensive public academic institutions. In fact, at major AI conferences that drive the field, most papers come from academic institutions. Our AI systems learn about our world only from the data we provide.

Current AI models like ChatGPT have scraped information from several academic journal articles in open access repositories, but there is a huge earth of undeveloped academic data that can be used to make all of these models more meaningful. The way beyond the data shortage is to develop new AI methods that utilize all our knowledge in all forms. Our research institutions have a wide range of expertise in all aspects of society to do this.

One example is: We are creating the next generation of “digital twin” technology. Digital Twin is a virtual recreation of our world’s locations and systems. Using AI, we have developed a digital twin that collects all the data and knowledge about the system, whether it’s a city, a community, or even a person, in one place, allowing users to ask questions “if any.”

For example, the University of Florida is building a digital twin for the city of Jacksonville. The city of Jacksonville includes the profiles of each building, elevation data for the city, and even the location of septic tanks. The twins also embed detailed cutting edge water flow models. In that virtual world, you can test all sorts of ideas for improving hurricane evacuation plans and water quality in Jacksonville.

As more data continues to be layered with twins, the ability to deploy urban resources, including real-time traffic information, road conditions scanning, and more is more informed and driven by real-time practical data and modeling. Using this digital twin-backed AI system, city leaders said, “How will new roads in downtown Jacksonville affect evacuation times? How will additional roads change the water runoff?” and more.

The possibilities in this emerging realm of AI are endless. You can create digital human twins and layer human biology knowledge with personalized medical history and imaging scans to understand how individuals respond to specific treatments.

The university is also gaining more and more powerful supercomputers, as it has recently acquired innovations such as the University of Florida Hypergator from NVIDIA. This is used in all areas for problems. For example, Oregon State University and the University of Missouri are using their own access to supercomputers to advance marine science discoveries and improve elderly care.

In short, don’t look to Silicon Valley right away to see the next big leap in AI. Starts scanning the horizon of research universities with computing hose power and launches a unique ability to continuously update data and knowledge beyond the next big thing in AI.

Dr. Alina Zare teaches and conducts machine learning and AI research as a professor in the Department of Electrical and Computer Engineering at the University of Florida. She is also the director of UF’s Institute for Artificial Intelligence and Informatics.