Thank you.
My name is David Duvenaud. I'm a professor of computer science at the University of Toronto, where I used to specialize in deep learning and generative models. In 2023 and 2024, I led Anthropic's alignment evaluation team. Our task was to test whether the company's AI was capable of pursuing hidden agendas—for instance, by subverting human oversight or decision-making.
I'm also an author on the “International AI Safety Report” and a member of the safe and secure AI advisory group for the federal Advisory Council on Artificial Intelligence. I am a co-chair at the Schwartz Reisman Institute for Technology and Society.
Today, I'm speaking in a personal capacity. I want to concur with Mr. Tessari L'Allié, in that very capable models raise serious security and loss of control risks. However, I want to address another challenge.
Large language models in particular, and their successors, are on track to becoming a competitive or superior replacement to humans in almost all of our important white-collar and decision-making roles over the next five to 10 years, roughly. In the slightly longer term, we're on track to make almost all humans economically obsolete, permanently. This will, in turn, cause a permanent loss of bargaining power for workers. Citizens will switch from being necessary for growth to becoming troublesome wards of the state and will have little recourse if they are then further marginalized and disempowered. We face a much larger problem than simply managing a temporary labour disruption.
I realize that this sounds similar to incorrect predictions made about previous labour disruptions, like the Industrial Revolution. Much wealth and many new jobs will be created as a consequence of improving AI capabilities. However, AI will, at some point, also be able to fill these new jobs better than humans. Eventually, each of us will have to compete with machine workers at least as capable as us but also faster, more responsive, more reliable and cheaper. This is the stated goal of the largest artificial general intelligence, AGI, companies, and they're well on their way to achieving it. Many of my former students are working at these companies, making fortunes, and many of them also believe that this is probably their last chance to have a real job. Teaching at the university is becoming depressing as students see the value of their skills decreasing month by month.
You might expect the major AI companies to have an answer to the question of how AGI development is supposed to ultimately benefit the average person economically, even if indirectly. However, their consistent stance has been that this is a huge problem they don't know how to address. For example, Dario Amodei, the CEO of Anthropic, said, “in the long run AI will become so broadly effective and so cheap that [comparative advantage] will no longer apply” and “At that point, our current economic setup will no longer make sense”. OpenAI's CEO, Sam Altman, was recently asked, “How will people survive economically?”, to which he replied, “I don't know; neither does anyone else.”
Over the past few years, I've systematically asked my colleagues in industry labs, research institutes and other academic disciplines for any coherent vision of how our civilization could robustly serve human interests once we are no longer competitive. The only consensus is that the window for individuals to compete and contribute is closing.
What does this mean for you, as MPs?
The main thing I'd like you to keep in mind, going forward, is that people are right to fear being replaced. This isn't just a period of disruption, after which things will return to something like business as usual. The default path is that we all become unemployable, except in mandated make-work contexts, and then marginalized in favour of a machine economy oriented towards growth for the sake of competitiveness.
The second thing to keep in mind is that we should expect governments to generally become much less responsive to the needs of their citizens after this happens. The need for human labour by the state aligns the state with its citizens. Right now, investment in education, and human capital more broadly, pays off for everyone in the long run. However, soon, fiduciary duty will require investing instead—mainly in data centres, power plants and robotic factories.
Finally, there is no way to address these problems without global coordination. Human replacement can happen even if everyone involved would prefer to prioritize human interests. It's just going to be the only way to remain competitive. Every country, industry and worker faces a choice between adapting to AI as fast as possible or being out-competed. No one can unilaterally do much to slow or soften the blow of eventual human irrelevance.
Thank you.
