Geoffrey Hinton, a prominent figure known as one of the founding fathers of artificial intelligence, called upon governments on Wednesday to intervene and ensure that machines do not seize control of society.
Hinton gained significant attention in May when he resigned from Google after a decade of service to speak more openly about the risks associated with AI, following the widespread fascination with the release of ChatGPT.
Speaking at the Collision tech conference in Toronto, Canada, Hinton addressed a captivated audience comprising over 30,000 startup founders, investors, and tech professionals, most of whom sought insights on how to navigate the AI wave rather than a discourse on its perils.
“Prior to AI surpassing human intelligence, I believe the developers behind it should be encouraged to dedicate substantial effort to comprehending how it may attempt to assume dominance,” stated Hinton.
“Currently, there are 99 highly intelligent individuals striving to improve AI, and only one equally brilliant person working to understand how to prevent it from overpowering. Perhaps we should aim for a more balanced approach,” he suggested.
Hinton cautioned that the risks associated with AI should not be taken lightly, countering critics who believe he exaggerates these risks.
“It is crucial for people to recognize that this is not a work of science fiction or mere fearmongering,” he insisted. “It is a genuine risk that demands thoughtful consideration, and we must proactively determine how to address it.”
Expressing concerns about exacerbating inequality, Hinton warned that AI could deepen the divide, as the significant productivity gains resulting from its implementation primarily benefit the wealthy rather than the labor force.
“The fruits of this progress will not be shared equitably among workers; instead, they will further enrich the already affluent, which is highly detrimental to society,” he added.
Moreover, Hinton highlighted the peril of fabricated news propagated by ChatGPT-like bots and expressed hope that AI-generated content could be labeled, similar to the way central banks watermark currency.
“It is of utmost importance to attempt to flag every piece of falsified information as such. Whether this can be achieved technically remains uncertain,” he remarked.
The European Union is currently contemplating this approach as part of its AI Act, a legislative proposal that aims to establish regulations for AI in Europe and is presently under negotiation by policymakers.
Concerns Amid Mars Colonization Discourse –
Hinton’s apprehensions about AI dangers diverged from the discussions at the conference, which were less focused on safety and threats and more centered around capitalizing on the opportunities presented by ChatGPT.
Venture capitalist Sarah Guo asserted that apocalyptic predictions about AI as an existential menace were premature, likening them to “discussing overpopulation on Mars,” quoting the views of another AI expert, Andrew Ng.
Guo also cautioned against “regulatory capture,” where government intervention shields established players before other sectors, such as healthcare, education, or science, have the chance to benefit.
Opinions varied regarding the future of dominant generative AI giants, primarily Microsoft-backed OpenAI and Google, with some speculating that new actors would introduce their own models and innovations, expanding the field.
“In five years, I still envision that if you seek the best, most accurate, and advanced general model, you will likely have to turn to one of the few companies with the necessary resources,” remarked Leigh Marie Braswell of venture capital firm Kleiner Perkins.
Zachary Bratun-Glennon from Gradient Ventures envisioned a future in which “there will be millions of models interconnected, much like the current network of websites we have today.”
Agence France-Presse
