Why AI MattersIn conversation with Jensen Huang and President Jayathi Murthy.

In the spring of 2024, Oregon State’s first Global Futures Forum on artificial intelligence explored the technology’s potential economic, scientific and creative impacts on the university and the world. Here’s an excerpt from a panel discussion with OSU President Jayathi Murthy and alumnus and NVIDIA CEO Jensen Huang, moderated by Provost Ed Feser.

Feser: My first question to Jensen and then President Murthy is “so what” about AI?

Huang: Let me take a step back first and explain why NVIDIA is at the center of the AI revolution.

We started the company 31 years ago to pioneer a new way of doing computing. Our observation was that in many important applications — it could be scientific, it could be computer graphics, it could be artificial intelligence or robotics — 5% of the code consumes 99.9% of the time to run. And if that’s the case, why would we compute using a general-purpose computer that does everything the same way? It’s not really sensible.

So we invented a programming model called CUDA that has effectively driven down the cost of computing over the past decade or so by a mil­lion times. Just do that thought experiment: If the cost of doing something — weather simulations, molecu­lar dynamics simulations — goes down by a million times, how would that change what you do? And it turns out because computers are such a foundational part of al­most everything we do, and the computer is such an important instrument in nearly every field of science, this changed everything.

For example, look at a few sample data points. Make an observation and apply the scientific method. Write some software to process that data and try to make that prediction that you’ve observed and then test your hypothesis against more experimentation. Go through that loop over and over again. That’s called the software development method.

Instead of doing it that way, why not just give the computer as many examples as you could and let it go figure out what the program is by itself? Let it write the program. Let the computer observe all of this mountain of data and find the patterns of relationships within the data, to extract, if you will, the defining features by itself, the predictive features by itself, and then write the program.

That is the gigantic breakthrough. And what I just described is the foundational technology called machine learning, which led to deep learning, which led to the breakthroughs that we know of today called artificial intelligence. That’s the “so what.”

When something of great value, something that’s hard to do, becomes a million times or a billion times or a trillion times faster or cheaper, how would that change behavior? We observed that it was going to change the industry completely, that software programing would be revolutionized, that the type of software we could write, unimaginable in the past, we would do on a routine basis now and a whole bunch of new applications would be created.

An example is applying artificial intelligence to one of the largest multi-physics problems: climate science. Or an incredibly tough problem for computers — easy for us, but a tough problem for computers — the articulation and manipulation of things, which is called robotics.

These types of problems are now within the practical imagination of developers around the world. That’s the “so what.”

Feser: President Murthy, how would you respond from a higher education and Oregon State University standpoint?

Murthy: If you look at the problems facing the universe or the world — the climate crisis is such a big part of it — our ability to handle big datasets give us tools that we’ve never had before. And this is particularly relevant for OSU: big on oceanography, on forests, on agriculture. AI gives us tools to address these problems, to create new models, to predict the future, to control the future. That’s a huge, big deal.

Enabling our graduates to be able to use these tools to address climate science, to address biotech, to address human health, to address robotics — that’s the big “so what” for me. Just expanding the possibilities of the problems that we can solve and enabling our students to address these big problems.