Perimeter teams up with Montréal-based experts to delve into the past, present and future of AI

Are we on the brink of a new tipping point in society and technology, brought about by artificial intelligence (AI) coupled with quantum computing? Where will these new transformative technologies take us?  

The answer is still up in the air, but during a Perimeter Institute 25th anniversary visit to the thriving science and technology ecosystem of Montréal in June 2025, that question was explored by some leading experts best-positioned to find out. The panel discussion, titled 25 Years of Discovery: AI, Physics, and the Future, was part of a cross-country tour being undertaken by Perimeter to expand its public outreach and be a connecting thread for new collaborations that will advance Canadian science.

The sold-out event was hosted by Perimeter in conjunction with its partners, Two Small Fish Ventures, a tech venture capital fund, Institut Courtois, a new research institute at the Université de Montréal focused on the development of new quantum materials and AI, and IVADO, a research, training and knowledge mobilization consortium in artificial intelligence.

The panellists included host Emily Petroff, director of external relations at Perimeter; William Witczak-Krempa, a former Perimeter postdoctoral researcher who is now a physics professor at Université de Montréal as well as a researcher at IVADO, and research chair at Institut Courtois; Estelle Inack, a quantum matter research scientist at Perimeter and former postdoctoral affiliate at the Vector Institute for Artificial Intelligence, who is also chief technology officer at yiyaniQ, a company that provides asset management companies with quantum intelligent optimizers; Carlos Silva, director of Institut Courtois; and Allen Lau, operating partner and co-founder of Two Small Fish Ventures and tech founder who previously sold Wattpad, an online storytelling platform.  

The panellists discussed the past, present and future of these new technologies, along with both the concerns and the potential for their future.

Lau said there is a case to be made that the AI technology evolving today is unlike anything that precedes it. “It is the first technology in human history that can make decisions, can reason, and can have a cross-domain functionality.”

Just as electricity is now present in almost every technology we use, Lau says AI is becoming so pervasive, in everything from robotic brains to the chips in computing, it will be part of almost everything we do in the future. “AI is going through that evolution now, and it is very exciting.”

Witczak-Krempa, who does research into materials such as superconductors that display quantum properties at low temperatures, said AI is not a magic bullet that can solve every problem. But “it is possible to use AI to accelerate some parts of the research,” he said.

He added that AI is useful in improving quantum computing as well. “People are using AI to write better algorithms, to calibrate the machines and optimize the runtime,” he said.

Inack’s company, yiyaniQ, is using AI and quantum techniques for what are known as “optimization problems.” An optimization problem involves finding the best solution from a set of possible options, something that is often hard to do when there are many variables.

Optimization problems pop up in all kinds of different sectors, from mapping out a road through hills and valleys to find the shortest route to a destination, to filling boxes in the most efficient and optimal way, to solve protein folding problems to design new drugs, or in finance where the job is to optimally allocate assets in a portfolio.

That’s where Inack’s background as a physicist comes in handy. Physicists have a mathematical framework for dealing with optimization problems and it involves writing the problem down as something known as a “Hamiltonian” to represent the total energy state of a system. It’s a framework that helps in understanding and predicting how a system evolves over time.  

With today’s quantum computers, scientists can use a computational method known as quantum annealing, which can help find the best solution from a large set of possible solutions.

Inack and her colleagues at the Perimeter Institute Quantum Intelligence Lab (PIQuIL) and also at Vector Institute in Toronto, came up with a better procedure to do this, which was published in the journal Nature Machine Intelligence in 2021. She subsequently co-founded her company based on that work.

AI, which is fantastic at pattern recognition, can speed up the job, making it much faster to find the optimal solution.  

As Inack explained at the panel discussion, quantum physics is probabilistic, and in any system where there are many interacting particles, you need to get a good approximation of the wave function of that many-body system.

It turns out that artificial neural networks are very good at that, she said.

What Inack’s company can do in finance can also be applied in many other types of optimization problems, like understanding and predicting protein folding structures.

Meanwhile, at Institut Courtois, researchers are investigating novel material properties, and using an advanced type of spectroscopy to understand the complex ways that light interacts with particles.  

Carlos Silva, director of Institut Courtois, said there are applications for AI in this work. “We can unravel framed models in a way that we just couldn’t do otherwise,” he said.

Witczak-Krempa added that AI has led to “self-driving labs” — automated laboratory systems that use AI and robotics to accelerate scientific discoveries. AI can help predict which materials or molecules can be synthesized, and then robotics can be used to synthesize and test those materials.  

Inack said ever since ChatGPT was unveiled in November 2022, the market for AI has simply exploded.

“We now have hundred of billions of dollars being invested by the largest companies in the world, just to build AI infrastructure, including high performance computing (HPC) data centres. Something like one half of startups are building AI agents,” she said.

It is a technology that is already having an outsized impact on science and technology, and as quantum computers become better, they too will supercharge the AI capabilities. “I think the interplay between both technologies, AI and quantum computers, is certainly very, very promising,” Witczak-Krempa said.

Silva said his institute is involved in collaborations with scientists who run self-driving labs where robotics driven by AI can accelerate the synthesis of materials.  

He believes that one potential of this technology is that it could drive down the cost of discovery and help “democratize science” in places where access to high-end lasers and other expensive equipment is limited. “In terms of the way science is done, there is a great opportunity.”

Lau, too, sees opportunities emerging as “the cost of intelligence comes down.” He noted that already, a service such as marketing, that might have cost $20,000 a few years ago, might now be a $200 monthly ChatGPT Pro subscription.

It means that startups are already rethinking, out of the box, how they should be building their companies from the ground up, he said. “We are living in a very interesting time because the cost of intelligence is dropping so fast.”

There are downsides and concerns, of course.

One of them, raised by Witczak-Krempa, is the environmental cost. Right now, AI companies have a heavy energy footprint. “They are grinding on a lot of data, and it works, but it's not always optimal. Yet there is money in it, so people are doing it. The question is: how much should we run the GPUs for a given research project?” he said.

Inack said that it is a concern, but she also foresees how quantum computing might help to compress AI algorithms and help find solutions more efficiently than is possible today.

But she said her concern is around the potential loss of creativity in the student population. As an instructor, she wants her students to use AI as a tool, but she also wants them to think and put some thought behind the questions, not just depend on the bot to supply answers.

“Imagination is what changes the world, and so if you want to train the next generation of highly qualified persons, you need to train them to think on their own, to be curious, to be imaginative,” she said.

Silva added that it is “very important to teach students to use AI tools as tools.”

Lau acknowledged these and other concerns, such as the anxiety around job loss. But he ended the panel discussion on a hopeful note, by giving a number of examples from history when the fears about a new technology did not materialize, and sometimes, the opposite happened.  

One example is automated spreadsheets on computers. At the time that they were invented, people feared this technology would do away with accounting jobs. But accounting didn’t go away because of computerized spread sheets, although the skills and the requirements for the job changed over time.

He also noted that while new technologies might eliminate certain types of jobs, they usually create others. We now have entire cybersecurity industries, for example, based on keeping websites safe from hacking. We also now have companies developing “post-quantum security solutions” in advance of quantum computers that might be able to crack encryption on the Internet, he said.  

Also, the sheer volume of data that is being generated by AI is leading to more research analysts being needed to parse all this information, Lau said.

“If what I have said gives you the impression that I'm not worried, I would like to correct that. Of course, I'm concerned about AI's safety. I'm glad some of the AI godfathers are bringing awareness and focus to that. But that being said, we should not stop progressing. We should not stop inventing. We can use technology to solve technology problems,” he said.

Click here to learn more about Perimeter’s ongoing 25th anniversary celebrations.