Nobel Prizes in Physics and Chemistry Awarded to AI Researchers, Signaling a Paradigm Shift in Science

To go back to this article, go to My Profile and then to View saved stories.

Nobel Prizes in AI Might Shift Research Priorities

Demis Hassabis was unaware of his Nobel Prize in chemistry award by the Royal Swedish Academy of Sciences until his wife received numerous calls from a Swedish number on Skype.

"Hassabis mentioned during a press conference held today to commemorate the prize reception, with his Google DeepMind colleague John Jumper by his side, that she repeatedly hung up the phone, yet the calls continued relentlessly," Hassabis recounted. "Eventually, she noticed it was a call from Sweden and they wanted my contact information."

The news that he clinched the award, the most honored in the scientific community, might not have been too surprising: Just a day before, Geoffrey Hinton, frequently referred to as one of the "pioneers of artificial intelligence," along with John Hopfield from Princeton University, were honored with the Nobel Prize in Physics for their contributions to the field of machine learning. "Clearly, the committee aimed to send a message, it seems, by awarding the two of us," Hassabis remarked during a press conference held following his victory.

Just to make it unmistakable: Artificial Intelligence has arrived, and it's currently feasible to secure a Nobel Prize by delving into it and making significant contributions to various domains—be it physics, as demonstrated by Hinton and Hopfield, or chemistry, evidenced by Hassabis and Jumper's achievement alongside David Baker, a genomic scientist from the University of Washington.

"Eleanor Drage, a senior research fellow at the Leverhulme Center for the Future of Intelligence at the University of Cambridge, describes this as a significant milestone for artificial intelligence within the realm of science. She highlights the remarkable achievement of renowned computer scientists securing awards in chemistry and physics, sparking anticipation about potential recipients of a peace prize. Drage mentions that there’s playful speculation among her peers about xAI proprietor Elon Musk possibly receiving such an honor."

Drage characterizes the decision to grant physics and chemistry awards to AI scholars as a significant controversy, affecting not just those fields but also external perspectives. She posits that this trend could stem from either the increasing integration of AI across scholarly fields, altering traditional academic demarcations, or from a societal fascination with computer science, to the extent of incorporating its experts into any available niche.

She's uncertain about the direction that recent choices indicate. However, she and her colleagues are convinced that these decisions will significantly impact the course of future studies.

"Securing a Nobel Prize through AI advancements might be an opportunity that's already passed, yet it's set to shape future research paths," states Matt Hodgkinson, a freelance expert in scientific research ethics and ex-research integrity manager at the UK Research Integrity Office. The real concern is if this influence will be positive.

Baker, who clinched the Nobel Prize for chemistry this year, is renowned for pioneering the application of artificial intelligence in predicting protein structures. For many years, he dedicated himself to this challenge, achieving progressive improvements. He understood that the clear-cut nature of protein structures presented an ideal scenario for testing AI technologies. His journey to success was far from overnight; Baker has contributed over 600 scholarly articles throughout his career. Likewise, AlphaFold2, a project developed by Google DeepMind that also received accolades from the awarding body, was not a sudden breakthrough.

Hodgkinson is concerned that the focus among researchers might shift more towards the method rather than the underlying scientific principles in their efforts to understand the reasons behind the trio's award this year. "I'm concerned that this might lead to a misuse of chatbots among researchers, driven by the incorrect assumption that all AI technologies are the same," he states.

Concerns about potential outcomes stem from the surge in popularity seen with other technologies touted as revolutionary. Hodgkinson notes, "Technological advancements often go through phases of exaggerated expectations, with blockchain and graphene being recent examples." After graphene was identified in 2004, there was a notable increase in academic interest, as evidenced by 45,000 papers discussing the substance from 2005 to 2009, data from Google Scholar shows. The recognition of Andre Geim and Konstantin Novoselov with a Nobel Prize for their work on graphene led to an even greater spike in research, with publications soaring to 454,000 from 2010 to 2014 and surpassing a million from 2015 to 2020. Despite this flurry of academic activity, the practical impact of these efforts remains relatively limited so far.

Hodgkinson is of the opinion that the acknowledgment of several researchers by the Nobel Prize committee for their contributions to artificial intelligence might inspire more individuals to gravitate towards the domain. This influx could lead to a variability in the quality of scientific work. "The real value and efficacy of the proposed ideas and uses of AI remain to be seen," he notes.

The influence of media focus and public interest in AI on the scholarly sector has been evident. Research from Stanford University highlights that from 2010 to 2022, academic output on AI surged threefold. In 2022, the field saw an explosion of activity with around 250,000 papers being released, equating to over 660 daily publications. This surge occurred even before the generative AI movement was propelled forward by the launch of ChatGPT in November 2022.

The issue of whether scholars are inclined to chase after media spotlight, financial incentives, and accolades from the Nobel Prize committee troubles Julian Togelius, an associate professor specializing in AI at the Tandon School of Engineering, New York University. Togelius observes that researchers often navigate towards the paths that offer the least resistance and highest rewards. Considering the fierce competition within the academic sphere, where the availability of funds is diminishing and intimately connected to the career opportunities of researchers, it appears probable that the allure of engaging with a currently popular subject, which now holds the promise of Nobel Prize recognition for outstanding contributors, might be irresistibly appealing.

There's a danger that this approach could hinder fresh, innovative ideas. Togelius points out, “Extracting more basic data from nature and developing new theories that are comprehensible to humans are challenging tasks.” These endeavors necessitate profound contemplation. It is significantly more advantageous for scientists to perform simulations supported by AI, which reinforce current theories and utilize existing data, resulting in minor advancements in knowledge rather than groundbreaking discoveries. Togelius anticipates that future scientists will gravitate towards this method since it presents a path of lesser resistance.

There's a concern that computer scientists, who have been pivotal in developing AI, might be tempted to branch into other scientific disciplines like physics and chemistry if those areas start rewarding AI contributions with Nobel Prizes. This could lead to them overstepping into domains where they lack expertise. "Computer scientists are known for their tendency to delve into unfamiliar fields, apply some algorithms, and claim it as progress, whether it's beneficial or not," Togelius notes. He confesses he once considered applying deep learning to a different scientific area to "improve" it but ultimately decided against it due to his limited knowledge in physics, biology, or geology.

Hassabis serves as a prime illustration of effectively applying AI to propel scientific progress. With a foundation in neuroscience, earning his PhD in 2009, he attributes his success in advancing AI through Google DeepMind to his academic background. However, he has noted a shift in the industry towards a greater focus on engineering. During his Nobel Prize press conference, he remarked, "Nowadays, the field has leaned more towards an engineering approach," pointing out that current advancements are increasingly based on algorithmic improvements, moving away from brain-based models.

This could also affect the nature of research conducted, the profiles of the individuals undertaking it, their expertise in the subject matter, and the motivations for their involvement. Instead of specialists who have dedicated their careers to a particular area, there might be an increase in studies carried out by computer scientists who may not have a deep understanding of the practical aspects of their research topics.

However, the festivities for Hassabis, Jumper, and their acknowledged team members for their Nobel Prize win this week are expected to overshadow that. Earlier today, he mentioned, "We are on the verge of finalizing the [AlphaFold3] code to make it available for the academic circle to use without any cost.” He added, “Following that, our advancements will continue.”

Recommended for You …

Delivered to your inbox: A collection of the most fascinating and peculiar tales from the archives of WIRED

Elon Musk poses a threat to national security

Discussion: Meredith Whittaker Aims to Disprove Capitalist Ideals

What's the solution for a challenge such as Polestar?

Gathering Alert: Don't miss out on The Major Interview happening on December 3, located in San

Additional Content from WIRED

Advice and Tutorials

© 2024 Condé Nast. All rights reserved. WIRED could receive a share of revenue from the sale of goods bought via our website, as a result of our affiliate agreements with retail partners. Content from this site is prohibited from being copied, shared, broadcast, stored, or used in any form without explicit consent from Condé Nast. Ad Choices

Choose a global website