Quantum and AI Are Our Next Problem-Solvers, Says IBM’s Research Director
Alumnus Darío Gil ’98 returns to campus to give his perspective on bits, neurons, qubits and the future of AI and quantum technologies in a packed lecture
AI and quantum science technologies will continue to evolve rapidly, helping us address the critical challenges of our time more quickly than ever before possible — while also bringing new technical and societal challenges with them as they grow.
That was the optimistic message Stevens alumnus Darío Gil ’98 returned to campus October 9 to deliver as the 17th installment of the university’s President’s Distinguished Lecture Series.
Stevens President Nariman Farvardin prefaced the lecture by explaining how Gil has built an impressive career since leaving Castle Point with his electrical engineering bachelor’s degree in 1998.
In addition to serving as an IBM Senior Vice President and Director of Research, Gil has advised U.S. presidents, currently chairs the National Science Board (which oversees the National Science Foundation), chairs the MIT-IBM Watson AI Lab and serves on numerous other educational and industry boards.
AI: Rapid advances, escalating power demands
After a hearty round of applause from a packed Tech Flex Auditorium, Gil began by describing the history of digital information — crediting the German philosopher-mathematician Gottfried Wilhelm Leibniz with the original idea of encoding information in binary form.
American scientist Claude Shannon picked up the idea much later, Gil continued, formalizing the idea of information theory as we know it today.
“We came to understand that objects as different as a punch card or DNA, that look so different, are all carriers of information,” explained Gil. “It’s a really fundamental idea.”
A new technology then enabled the concept to scale up rapidly, in a way that would soon transform society globally: the invention and development of the transistor.
“We have made bits free, almost,” Gil said, noting that by the end of the decade we will be able to pack 1 trillion transistors into a space roughly the size of a fingernail.
The development of artificial intelligence began with a similar leap. Gil (who is originally from Spain) noted the early influence of the Spanish medical researcher Santiago Ramón y Cajal, one of the first to characterize the physical details of neurons as well as their interactive processes within the body’s nervous system.
This is the story that is unfolding. We’re living ... in the opportunity not only to build these technologies, but to unleash them to actually solve problems that we could not solve before.
Computer scientists, continued Gil, eventually borrowed these ideas from neuroscience to develop so-called neural networks, which interconnect with each other and encode and reinforce certain outcomes and predictions in systems that can be automated to “learn” autonomously.
“Once that happens, and it’s inside this large-scale network, you can actually connect [disparate things] to one another,” he explained. “You can do this for almost anything.”
However, Gil cautioned, there is at least one key challenge of AI and large-language models: the power demanded of and consumed by these systems is rapidly escalating — possibly even threatening to overwhelm the world’s energy supply.
“We have a lot of work ahead of us as a community to radically lower the amount of energy consumption to deliver these functions,” Gil said, noting that technology firms will need to develop both hardware innovations (such as improved chip architecture) and novel approaches within AI itself to make it more efficient.
Before departing the topic of AI, Gil also stressed the importance of making its principles as transparent, open-source and diversely accessible as possible, noting how IBM recently co-founded The AI Alliance with Meta to explore information-sharing and efficiency with approximately 50 industry and academic partners.
Quantum: Super complex, super cool (literally)
Next Gil moved on to quantum science, which is beginning to disrupt and transform computing technology in bold new ways.
“What is the relationship between physics and information?” asked Gil, before answering that they are actually no longer separated — quantum computing has brought them together spectacularly.
“It is the wrong thinking to say that a quantum computer is a faster computer,” he said. “What it is, is a fundamentally different type of computer,” he said. “What’s the difference? As amazing as our computers are… they can only solve ‘easy’ problems.”
However, many real-world challenges — such as modeling natural phenomenon, for example — are far too complex to attack and solve quickly or efficiently with classical computers.
Quantum computers are superior for those types of problems thanks to three “superpowers,” Gil explained, which he defined as interference, superposition (mathematical tools that increase the information that can be carried by physical pieces of material) and entanglement.
That last principle, entanglement, is especially remarkable: two entangled, separated entities can essentially “twin” each other, always telling us everything about the other when we’re looking at one — even if that seems both impossible due to their physical distance apart and also just unintuitive.
To build quantum systems that leverage these quantum effects, however, computing design has to evolve rapidly. IBM has been at the forefront of that effort, Gil explained: The company first placed a small, five-qubit quantum computer online in the cloud in 2016.
To make a quantum computer work, he continued, superconducting technology within the machine converts information flowing into the system into microwave pulses that travel into a cryostat — basically, a deep-freezer — while the entanglement, superposition and interference operations take place.
“One of the coldest places in the university is actually the bottom of a quantum computer,” noted Gil, and IBM’s machines are cooled to near absolute zero (about -460 degrees Fahrenheit, which is the coldest theoretical temperature anything can be).
Up next: Solving the biggest challenges
As demand continues to explode for these new computers, IBM has chosen to double down on its own technologies, Gil said, and is now developing, testing and building newer versions of quantum computers that will incorporate multiple quantum processors within individual cryostats, as well as multiple cryostats connected to one another.
The ultimate goal of more powerful quantum computing? Societal betterment. Gil listed a few potential uses: These computers could help iterate new antibiotics, battery technologies, energy catalysts, fertilizer chemistries, stronger aircraft materials, as well as address other challenges.
“This is the story that is unfolding,” he concluded. “We’re living, this decade, in the opportunity not only to build these technologies but to unleash them to actually solve problems that we could not solve before… and we could use the help.”
A spirited question-and-answer session with the Stevens audience followed, during which Gil affably touched on everything from technical metrics to the emerging interconnections between AI and quantum and the economics of designing and building quantum computers (tens of billions of dollars, at minimum).
He also discussed quantum partnerships to address health challenges; the types of problems quantum computers will be best suited to address going forward; and the unique quirks and challenges of quantum systems — they require their own specialized computing programs, for instance, and also do not possess ‘memory’ in the conventional sense that stores data perfectly and can bring it back up later at the touch of a button.
“You cannot clone quantum information,” he explained. “You measure… and then it’s gone.”
President Farvardin concluded the event by conferring the ninth President’s Medal upon Gil and announcing a new first-year Stevens scholarship in the speaker’s name.
Gil also held a private roundtable with Stevens students prior to his talk.