Nobel Prize scientists on AI, democracy and critical thinking

Nobel prize-winning scientists and a world-leading AI researcher highlighted the dazzling potential of AI to support research, the contributions of science to democracy and the importance of critical thinking in the age of AI, at a Nobel Prize Dialogue held in Brussels this week.

The three scientists are: Ben Feringa, a Dutch professor who received the 2016 Nobel Prize in Chemistry for his research on molecular machines; Sir Paul Nurse, British winner of the 2001 Nobel Prize in Physiology or Medicine for his work in genetics; and Demis Hassabis CBE, a British AI researcher and co-founder and CEO of Google DeepMind.

“We’re now at an incredible inflexion point,” said Hassabis. “We’re about to enter, maybe in the next 10 years, a new golden era of scientific discovery, helped by AI in many fields.” His lab is working on a large language model that could work like a research assistant.

The scientists joined two Nobel Peace Prize winners and three top democracy researchers at the first Nobel Prize Dialogue held in Brussels to discuss “Fact & Fiction: The future of democracy”. Filipino journalist Maria Ressa received the Nobel Peace Prize in 2021 and Oleksandra Matviichuk is head of the Center for Civil Liberties in Ukraine, which won the 2022 Nobel Peace Prize.

The event

The two-hour event was held on 5 March at the Palais des Beaux-Arts in Brussels and online, and was produced by the Nobel Prize Outreach in partnership with the European Research Council (ERC), under the auspices of the Belgian Presidency of the European Union.

Its premises included that understanding science is key to understanding the world, and that AI-generated misinformation affects democracies, which rely on fact-based world views and science as well as on narratives that can bring together large and diverse communities.

This year half of the world’s population will experience elections – but 80% of those people live in non-free or partially free societies, said Matviichuk.

“This means we are losing freedom. The problem is not just that in authoritarian countries the space for freedom is shrinking to the size of a prison cell. The problem is that even in well-developed democracies, there are forces putting into doubt the Universal Declaration of Human Rights,” Matviichuk stated.

The Brussels dialogue is part of a series being held around the world, said Anna Sjöström Douagi, acting CEO of the Nobel Foundation and executive director of the Nobel Prize Summit, last convened in May 2023 in Washington DC in partnership with the United States National Academy of Sciences with the theme “Truth, Trust and Hope”.

The Nobel organisation and the European Research Council (ERC) share an interest in science and discovery, said ERC President Maria Leptin. The European Union won the Nobel Peace Prize in 2012, and many ERC-nurtured researchers have gone on to win Nobel prizes. The ERC has a massive budget of €16 billion (US$17.49 billion) from 2021 to 2027 to fund basic research across all fields.

AI is accelerating research

For a session on AI, science and society, Nobel prize winners Feringa and Nurse and Google DeepMind CEO Hassabis were joined by Vera Jourová, vice-president for values and transparency at the European Commission, with Leptin moderating.

“I’ve spent my whole life working on AI because I always regarded it as potentially the ultimate tool for helping us to do science and accelerate scientific discovery,” said Hassabis. “If you think about AI at an abstract level, it’s a system or a tool that can make sense of a lot of data – find patterns, insights and structure in data.”

Hassabis is the developer of AlphaFold, a deep learning-based algorithm for accurately predicting protein structure. AlphaFold is solving the problem of protein folding, trying to find the 3D structure of proteins just from their amino acid sequence.

“That has all sorts of implications in drug discovery and disease understanding. Over the last couple of years, we’ve predicted the structures of over 200 million proteins, pretty much every protein known to science, which would have taken many, many – actually millions –of years of experimental work,” Hassabis told the dialogue.

“So we’re seeing a revolution in biology. It’s going to apply to other areas too, like chemistry, material science, physics and mathematics. All these scientific disciplines will benefit from AI.”

Biologist Nurse said AlphaFold had a major impact on his lab, which uses it all the time. The algorithm enables scientists to turn sequences into biochemistry. “It isn’t always right, but it is sufficiently right to be a fantastic tool,” he said.

While AlphaFold has been very useful, for researchers ChatGPT has been “pretty hopeless”, said Nurse, who is director of the Francis Crick Institute in London and has served as president of the Royal Society, as head of Cancer Research UK and president of Rockefeller University. “What we get back is something like a middle quality, high school student assessment of the area. Maybe it will get better.”

Feringa, of the University of Groningen in the Netherlands, who is on the ERC’s Scientific Council, gave an example of how AI is being used. “If you want to make a new drug to treat cancer, for instance, you need sometimes 35 or 40 different chemical conversions, chemical steps – like Lego – to build a complex molecule that treats breast cancer, for instance.

“To design these routes, there are hundreds or maybe thousands of possibilities. So from all the collective information that we have in the chemical literature and in the physical literature, we then use these programmes to design routes,” he explained. Humans decide on the best routes to take. Making a drug becomes easier with AI.

Another example is hair shampoo, which is surprisingly complex. “We are building robot systems in the lab. We feed data to the robots, and they do a selected experiment. So we don’t have to do thousands of experiments, but maybe only 100, to get the right combination and get the information we need to make a good hair shampoo,” Feringa said.

AI as research assistant

Jourová, a Czech politician and lawyer, joined the scientists in saying that while currently flawed, generative AI will improve to become a very useful tool. “I would really like ChatGPT to contain wisdom.” Feringa said the way around the generative AI flaw of ‘garbage in, garbage out’ is to train AI models on quality data. For his part, Nurse wants “ChatGPT with spark. Something more lively and imaginative.”

While “everyone’s gone crazy” over chatbots, the most interesting AI systems are specialised for scientific endeavours – such as AlphaFold, said Hassabis. “We’re going to see a lot more of those types of systems.”

At the same time, there is also a lot of work being done on chatbots. “A core problem we have to solve with AI systems in general is to make sure they don’t hallucinate and they stick to facts. Then they could be quite useful at a research assistant level, in terms of summarising an area or a series of papers, and then senior scientists can help make connections.”

Hassabis continued: “Actually, we’re working on a large language model that could be like a research assistant, and maybe help predict things like the outcome of an experiment. You could imagine describing an experiment, and then maybe it gives you some view on that. These sorts of things are coming, maybe in the next few years.”

Further, what software systems such as AlphaFold are really doing is helping people to navigate possibilities. Whether in chemistry or mathematics, in physics or biology, researchers are trying to search, “like a needle in a haystack”, to find a solution to a question, he said.

“That’s really what a lot of science boils down to. If you can capture your problem in that way, then these types of AI systems we’re building now can be very useful. There’s huge potential,” said Hassabis.

The critical thinking imperative

Leptin reminded the audience of some of the risks of algorithms discussed in earlier sessions – algorithms based on getting clicks through fear, hate, anger and mistrust.

Feringa is positive about AI and other technologies, but stressed: “A critical attitude is crucial.” A primary job of university scientists is to train students to be critical. “To think, what does this mean? On what data sets is this analysis based, and what is the quality of its output.”

Opportunity is what Nurse sees. “Because in the past, if you wanted to look at a topic, you would go to the library, you would trawl through the journals, you would take days, maybe weeks, to put things together that [new] methods can produce in minutes. Now, frankly, that was rather dull. So reducing this work to minutes is a good thing.”

He agreed that the next step required is critical thinking. “So the time we liberate from trawling through the library is then applied to critical thinking that has educational consequences.”

Academics must ensure that data used to generate analysis is good. Further, it is crucial to ensure that algorithms used perform in the ways that researchers want. Finally, scientists must understand what is going on: “We can’t be stupid and just press the button on the computer.”

Jourová believes the critical thinking of autonomous individuals is extremely important. “Because what we witness in this digital era is that more and more there is a tendency to create a crowd and individual people are disappearing.”

When working, Jourová thinks about the person – the voter, the citizen – who should be able to recognise what is false and what is true. In democratic elections, it is important that voters know that a real person who wants to be elected is speaking to them.

Under the AI rules that Europe is testing, there is one that says AI-produced texts or images or other material must be labelled, so that citizens can know if something is an AI production. It is important that social media platforms will proactively and systematically remove deep fakes.

All of the big technological companies except Twitter, now X, have committed to either label an AI production – or remove it – “so that people can be sure of what they read,” Jourová said. “Without that, we will see hidden manipulation. Then we can forget about free and fair elections.”

Hope, science and democracy

Regarding AI and democracy, Hassabis said that if used cleverly, AI could be part of a solution. Actions such as watermarking, and synthetic IDs, which use AI to watermark material so that governments, journalists and others can detect it. “It can be flagged automatically, and it's hard for a hacker or a bad actor to remove.”

AI can also be used defensively in cybersecurity and other areas. “I think AI is going to be a very important tool for our democracies to defend themselves,” he said.

Used responsibly, AI could accelerate creativity and innovation, and could help tackle many of the challenges facing society today, such as healthcare systems and climate change.

“AI could be an incredible strength for democracies to accelerate what we already do best, which is innovate and do science and discover new knowledge, including truth,” said Hassabis.

Feringa cited historian Professor Yuval Noah Harari of the Hebrew University of Jerusalem, who said that while in the past, power was with people who had information – in future, power will lie with those who know how to find valuable information amidst the flood.

Jourová opted for a much cited anonymous quote, that “artificial intelligence is no match for natural stupidity”. She called for some sort of permanent, triangular collaboration between the three worlds of technologists (tech companies), politics or the public sphere, and research.

Asked to connect science and democracy, Nurse said: “Science is critical for democracy. Science fundamentally is the pursuit of truth. Democracy is built on truth and trust. Science provides a trusted way at generating knowledge. It’s critical for the democratic process.

“And not only that, science is influencing society increasingly as generations go on, and we have to produce democratic institutions and ways of working that can accommodate and take on board the complexities of science. I'm not sure we’ve got there yet, but we have to do it.”

Nurse said science has the power of diplomacy. “Science provides a common language across all nations. Science breaks down barriers. Even when it’s difficult, it is possible to talk – particularly with discovery science, which is open and somehow beyond the politics.”

When political relations break down, such as between the United States and China, contacts can be rebuilt through science. “We mustn't turn that off,” said Nurse: “I really think science has an important diplomatic role which will lead to increased peace.”

Email Karen MacGregor at macgregor.karen@gmail.com.