From: mk_thisisit
The Royal Swedish Academy of Sciences, located in a building where Nobel Prizes in physics, chemistry, and economics are awarded, recently hosted a press conference where the future of Artificial Intelligence (AI) was a central topic [00:00:36]. The fundamental question addressed was whether strong Artificial Intelligence will ever come into being [00:01:00].
The Dawn of Superintelligence
Both Professor Jeff Hinton, considered the “father of neural networks,” and Demis Hassabis of DeepMind, believe that superintelligence will be achieved [02:26]. This is not seen as an exaggeration or a marketing ploy, but a long-held belief [00:00:00].
They predict that something akin to superintelligence could emerge within 5 to 20 years, with Demis Hassabis suggesting it could be as soon as 10 years [02:41]. This rapid progress means humanity must seriously consider how to maintain control over such a powerful technology [02:48].
Regrets and Risks
Professor Jeff Hinton expressed regret not in the sense of guilt, but rather unfortunate circumstances, wishing he had considered the security issues related to superintelligence earlier, as it is likely to arrive sooner than he anticipated [01:57].
Demis Hassabis, who founded DeepMind in 2010, stated that his passion has always been to use AI for scientific discoveries, particularly in areas like disease treatment, energy, and climate [03:10]. However, he acknowledges the inherent risks of such a powerful technology, emphasizing the need to take them “extremely seriously” [03:28]. There is limited time to conduct research on the interpretation and control of these systems [03:44].
Regulation and Accessibility
The challenge of regulating AI is significant due to its rapid development [05:07]. Governments and social organizations are advised to create flexible regulations, perhaps building upon existing frameworks in areas like healthcare or transport, and dynamically adapting them to technological advancements [05:16].
A direct threat of AI is autonomous lethal weapons, which current European regulations explicitly exclude from military applications [04:23]. This leads to an AI arms race among nations such as the United States, China, Russia, Great Britain, and Israel [04:43].
To promote fairness and prevent discrimination, DeepMind makes many of its tools and research results available as open source or easily accessible, making the technology increasingly democratic [06:01].
AI in Scientific Discovery
AI is described as a “new type of microscope” that offers a novel perspective on reality, with its results confirmed through experiments [06:22]. It is increasingly seen as a key stage, or even the main element, of scientific research [06:42].
Nobel Prizes, however, will not be awarded simply for using AI; they will continue to recognize actual discoveries [06:50]. For example, future Nobel Prizes might be awarded for breakthroughs in understanding cell function enabled by AI tools like AlphaFold, which predicts protein structures [07:00].
Deep learning methods, while powerful, require vast and well-prepared datasets. The success in protein structure prediction, for instance, relied on decades of experimental research and billions of dollars in investment to create detailed protein data [07:19]. The future development of AI in science will depend on the creation of similarly rich datasets in other fields [07:56].
Nobel Recognition for AI
Professor David Baker, a Nobel Prize winner in chemistry, highlighted that the recent Nobel Prizes serve as “undoubtedly evidence of the enormous influence of Artificial Intelligence” across various fields [09:21]. While his own work predates the AI era, the design methods developed in recent years are based on its use [09:28]. This year’s awards notably emphasized technology [09:44].
Professor Baker emphasized that while technology provides tools, it is people who identify problems and drive the solutions [09:56]. Identifying problems solvable by designing new proteins requires human intuition and approach, which computers cannot provide independently [10:17].
The 2023 Nobel Prize ceremony was unique because it awarded individuals like the creators of AlphaFold/DeepMind, who are young and whose research is very recent, unlike the traditional decades-long wait for recognition of scientific achievements [10:30]. This reflects a shift in how the Nobel Committee acknowledges the impact of technology on humanity’s development [10:48].