10/28/2024
“The best supercomputer is still the size of an apple and runs on sugar between our ears.”
10/28/2024
“The best supercomputer is still the size of an apple and runs on sugar between our ears.”
“Today, on behalf of the student community, the alumni community, and the world at large, I am here to thank you all for what you do. You are changing the world by changing lives like mine one at a time.
I take great pride in my close association with the university over the past many decades, and I can't thank you enough for that. To the extent that I've been able to make some small contributions to what is and what will be the greatness of this august institution, the University of Illinois Urbana-Champaign, it is my great privilege and honor.”
“I am very much a product of the University of Illinois Urbana-Champaign.
To the extent that I learned to read and understand history, I learned that at Illinois.
To the extent that I learned to write effectively, think creatively, work diligently to question everything, and think out of the box, I learned that at Illinois.
To the extent that I learned to overcome adversity, to be doggedly persistent, to strive for excellence every day, and to try to attain the unattainable, I learned that at Illinois.”
Thomas M. Siebel
So said Thomas Siebel when he visited his alma mater on Friday, October 18, for the dedication of The Grainger College of Engineering’s Siebel School of Computing and Data Science. He graciously accepted expressions of praise and gratitude from UI system, campus, and college leaders gathered along with faculty, staff, and students at the NCSA auditorium.
Before the gathering, Siebel joined the 2025 Illinois Siebel Scholars for a discussion. One onlooker said that he sat down and got right into an animated discussion with Tanay Dixit , Hao-Yu (Max) Hsu, Baoyu Li, Ashutosh Sharma and Alan Wang. The Siebel Scholars program annually recognizes nearly 80 exceptional students from leading graduate schools of business, computer science, and bioengineering in the United States, China, France, Italy and Japan. Siebel also attended a meeting of The Grainger College Board of Visitors.
Siebel was candid and informative in his fireside chat with Grainger Engineering Dean Rashid Bashir. The pair discussed Siebel’s career journey from Oracle to founding Siebel Systems and selling it to Oracle, as well as his current endeavors at C3 AI. Siebel addressed the costs and benefits of entrepreneurship, his approach to philanthropy, how C3 AI deploys artificial intelligence across its cloud-based services, the present and future of AI and LLMs, AI and the stock market and AI and the future of jobs as he took on questions from the Dean and those submitted by students. What follows are some selections from the discussion. See the dedication and fireside chat here.
Advice to students about what to study
“You are at the University of Illinois Urbana-Champaign. You have died and gone to heaven, but you may not know it. The resources that you have at your disposal are unlimited. If they tell you can't get there, figure out how to break the rules and get there. There is nothing that is not at your disposal. If you need a nuclear reactor, you get access to a nuclear reactor. There's nothing they don't have here. Find out what you're interested in and become a domain expert. And I'm not saying that the right thing to do is computer science. If it's philosophy, become an expert in philosophy. If it's religious studies, do that. Economics, fine arts, it doesn't matter. Become a domain expert. The resources instantly available to you in this institution are limitless, and if you can't access it, you just haven't asked the right person yet.”
Siebel’s philosophy of philanthropy
“I want to leave my neighborhood in a better place than I found it. I've had enormous good fortune in my professional career, so I try to give back in ways that are highly leveraged. I think the greatest gift there is education. Do you want to fight poverty, do you want to fight homelessness, or do you want to fight hunger? What is the key? I think there needs to be a safety net. But what's the real long-term solution? It's education, and that’s what has been the motivation.”
C3 AI and Enterprise AI
“They're out-innovating each other like crazy, and you want to take advantage of that innovation. You want to be LLM agnostic, so whoever's got the hot ticket that day you can use it. Reference the $2 billion of work we did in the first decade to solve these fundamental problems: the problems of access control and factor authentication, encryption, and attribution. We turned the temperature down to zero, so it's hallucination- free. We separated the language model from the data. The language model cannot access the data, so it can’t data exfiltrate. We have hundreds of installations today at places that are considered secure, like the Intel agencies and the United States Air Force and the United States Marine Corps and a number of banks, so you know we've solved those problems.”
AI and the stock market
“What's going on in AI now, particularly in generative AI? You ask, is there a bubble here? Yes, holey moley! In generative AI it's just truly crazy out there right now. I think John Maynard Keynes said, ‘Markets can remain irrational longer than you can remain solvent.’ What we're seeing in generative AI is wild. Is this technology important? It is hugely important it changes everything, but the company that makes Chat GPT loses about 5 billion dollars annually. This company could disappear next Monday, and it wouldn't matter. It wouldn't affect anybody. This company just got valued at $150 billion. Some of you can remember when that was a lot of money. That would be the valuation of, maybe, the second or third biggest company on the planet, but if it disappeared next Monday, nobody here would be that surprised, and it wouldn't make any difference at all to anybody. You see what's going on in my world. Right now, in the venture capital community on Sand Hill Road and in Boston, we have scores of companies being funded every day. Three people from a university want to build a generative AI model for law office automation. They're going to use it for generative AI for divorce lawyers, generative AI for doctor's offices, or generative AI for dental offices. These companies are being valued every day at greater than billion-dollar valuations. Does this sound like some tragedy about to play itself out? Well, it is.
There are enormous problems with IP liability. Large language models, as you're aware, are trained on the corpus of the internet. Somebody owns the copyright on all that information, be it a novel by Stephen King, the coming weather data from the weather company that's owned by IBM, or stock prices that were generated by Bloomberg. Those people want to be paid. Over the next ten years, there will be a massive amount of litigation.”
AI and risk
“What is the primary exchange for the sex trade? The documented fact is Facebook. The right to be forgotten - everybody’s for that. Why can't you send Google, Facebook, whoever it is, a letter and say, ‘I want to be forgotten in 24 hours’ and 24 hours have all your records expunged? Because it's not in their best business interest, that's why. So, I don't understand why there is outrage about the problems associated with AI today. Can it get worse from here? Absolutely, and it's going to get worse, but I don't think we need to worry about sentient computing. Okay, there's some nonzero probability that sometime in the future, there'll be a sentient computer. I don't think it's anytime soon.
Let's recalibrate and put this into perspective. The largest computer cluster we spool up for training these models might be 20,000 GPUs. It's $200 million worth of computer capacity; it takes 20 megawatts of power, and these learning models do almost nothing compared to the human brain. We have 100 billion neurons and 100 trillion simultaneous connections, and all of those connections are analog. These are neurotransmitters and neuromodulators like serotonin and dopamine, and we're really underestimating their power. The relative capacity of the Blue Waters machine or this 20,000 GPU cluster we put up might have the intellectual capacity of a mealworm compared to what the human mind can do. People in data science believe, and this is hard to imagine, that these neural networks or deep learning models somehow emulate the structure and functions of the human brain. Let's get a grip here. All this conversation that we need to be worried about the sentient computing that's going to take over, that our smart refrigerator is going to take over your house like HAL in the movie let's get over it.
There are some real problems out there. How about social media?”
AI and jobs
“It's going to make us enormously more productive at what we do, and that's a good thing. Now, there's some dark side to this, too. We can get into it if you want, and we should go there. Supply chain, demand forecasting, fraud detection, production optimization and things like these are absolute job creators. This is going to be a big economic boost, and it'll boost the GDP. Some things will be automated that have to be automated, but it'll be a major job creator. That's what many of us believe, but there is this public concern around what will be automated.”
Students leaving school to start a company
“Everybody wants to leave college and go and be an entrepreneur. I've built four successful companies, and if I had left here and gone and started a company, the probability of success would have been very low. Because, like every graduate student and undergraduate student here, I didn't know anything. I didn't know how to sell. I didn't know how to market. I didn't know how to account. I didn't know how to do customer service. I didn't know how to hire. I didn't know how to deal with human capital. So, I went to work for a little startup and watched it grow from 20 people to a thousand. I learned from them. We made mistakes every day, but I learned from what we did right. I learned from what we did wrong. I would go to work for a company that has bright people and learn something as it relates to your field. Then, if you have a drive to be an entrepreneur, go do it. I'm telling you, it's harder than you think. I think this idea of leaving school and going out and doing startups a day after school is candidly crazy.”
On being a tech CEO
“The idea that you can go to soccer games every Thursday night and be Andy Grove or Bill Gates, it's not going to work that way. If you decide to be that person, you can be that person, but you're going to pay a price because you're not going to be there for birthdays and soccer practices; you're not going to be home for dinner, so be careful what you ask for. I'm not certain that's the right answer. I think it's okay to be home for birthdays; I think it's okay to be at soccer games. I think that might be the right answer, but this idea that you can do all that and be Steve Jobs? It cannot be done. If you want to be Steve Jobs, Bill Gates, Andy Grove, or Larry Ellison, you will pay a price, so you need to be willing to pay it. It's straight from where I've been; I've done and paid for it. I'm very fortunate. I've had my wife of 38 years, four wonderful children, and four great grandchildren, so I am the luckiest person in the world. But you won't be there for the birthdays.”