Medicine will become more targeted and personalised, improving patient outcomes, but this will also thrust thorny issues such as genetic privacy, editing and ethics to the forefront, said Professor Aaron Ciechanover in his opening plenary lecture for Day Three. 

He stated that there have been three broad eras of medicine to date. In the first era, medicines were discovered incidentally. Thousands of years ago, for example, ancient Egyptians noticed that willow bark had pain alleviating properties. This eventually led to the creation of aspirin. In the second era, scientists deployed brute force screening to identify chemicals and compounds that could serve as the basis for various drugs. 

Scientific advances including the sequencing of the human genome kickstarted the third and ongoing era of personalised medicine. “We are moving from one-size-fits-all medicine to medicine that fits us exactly. We are also starting to stratify diseases at the molecular level,” he said. “The more information we can get about a person’s genetics, however, the more questions will arise. If I get information about my genetics and vulnerabilities to certain diseases, do I have an obligation to tell my employer or insurance agent? Do I tell my children? Knowing the future is something we have never had to deal with.”

Mathematicians have always built on one another’s work, with the history of prime numbers in particular going back thousands of years, said Professor Ngô Bảo Châu in his plenary lecture which outlined the history of the subject and zeroed in on some of its more well-known proofs, theorems and hypotheses. 

He noted that the mathematical proof that there are infinitely many prime numbers was produced by Euclid of Alexandria more than 2,000 years ago. Euclid had arrived at this conclusion by demonstrating, through an equation, that it is impossible for there to be only a finite number of prime numbers. Later, Georg Friedrich Bernhard Riemann observed that the frequency of prime numbers is very closely related to the behaviour of an elaborate function. 

Even Professor Ngô’s breakthrough, for which he won the Fields Medal, involved prime numbers. As a young mathematician, Robert Langlands had formulated an ambitious programme that united two branches of mathematics, namely number theory, which involved prime numbers, and representation theory. His programme, however, relied on an assumption, called the fundamental lemma, which was unproven for 30 years until Professor Ngô showed it was correct. “People have been coming up with elaborate logics for thousands of years,” Professor Ngô said, adding that there are still many mysteries waiting to be solved. 

Professional advice shared the stage with personal anecdotes in Sir Tim Hunt’s lively plenary lecture about his life in science, and how he came to make the seminal discovery that a group of proteins controls cells’ growth, duplication and division. “The first problem for any young scientist is to find a problem to solve. When I was starting out, my supervisor told me to go to the library to look for a problem, and so I did,” Sir Tim shared. 

He also recounted how going to conferences and making friends with fellow researchers had prepared him for his later discoveries and opened up new opportunities. It was at his first conference that he heard two talks that would both play a role in his later Nobel Prize-winning breakthrough. It was also through lending a bicycle to a fellow scientist that he made a friend and was invited to work at the latter’s prestigious research institute. “I cannot emphasise enough the importance of personal interactions in the progress of science,” he said. 

Sir Tim further advised young scientists: “Work with people who are cleverer than you are, try to question your own prejudices, even though it may be difficult, and, most importantly, follow your nose, wherever it may lead.”

The ability to reason is the major component of cognition that needs to be added to machine learning to achieve the goals of artificial intelligence, said Professor Leslie Valiant in his closing plenary lecture for Day Three. In supervised learning, machines are given well-labelled data to teach them how to process similar data in future. This has been useful for various technologies but is not enough. To achieve true artificial intelligence, machines should also be able to examine datasets, reason and come up with rules to process further information.

Professor Valiant outlined several features of such reasoning, which he called robust logic. “If the machine develops two rules that contradict each other in some situations, it should be able to learn its way out of the situation by getting more samples,” he said. He noted that some experiments had underlined the superiority of machines that are able to reason. In one experiment, machines were tasked to fill in missing words in sentences drawn from the Wall Street Journal. Those that were able to derive rules from the sentences were more successful.

Professor Valiant said that attaining such robust logic would be difficult: “One of the challenges will be developing good teaching materials. But it will be worth it.” 

“Don’t you find it beautiful?”, asked Dr Leslie Lamport, showcasing a formula he found appealing at a public lecture at the NUS School of Computing. “It takes a while to appreciate beauty”. 

He was referring to a formula in context of Euclid’s algorithm, one of the oldest algorithms still in common use. It is named after the ancient Greek mathematician Euclid, who first described it in his Elements (c. 300 BC).  For Dr Lamport, it nicely illustrated his point that algorithms are much simpler when explained with mathematics rather than with programming language. 

It was a point that he was to make repeatedly at the lecture, which was held in conjunction with the Global Young Scientists Summit. He urged the computer scientists in the room to have an intuitive understanding of exactly what it means to apply Euclid’s algorithm. Large formulas, he said, are handled by hierarchical decomposition, yet Maths has the simplest and most powerful definition known to man – decrement.

Yet today, using formal definitions for algorithms is not something that is very widely practiced in industry, and it was revelatory to a lot of experienced people in the room. So Dr Lamport went on to describe that some of the world’s largest technology companies such as Azure, Microsoft and Amazon Web Services (AWS) are now applying formal methods (including TLA+, a language he developed to write algorithms with maths), to test and find bugs in system designs that cannot be found through any other technique.

He illustrated this by highlighting that programmers from AWS, quoted in the book Formal Development of a Network-Centric RTOS (2011), said that they found formal methods surprisingly feasible for mainstream development and that applying formal methods gives good return on investment. This is because, according to Dr Lamport, “You don’t produce 10 times less code by better coding. You do it with a better algorithm.”

Urging young computer scientists to learn maths, Dr Lamport said that he has created a series of videos to teach the use of TLA+, because he strongly feels that programming needs to be understood in context of mathematics, and that computer science teachers should teach maths rather than how to write things in words. Dr Lamport concluded the lecture by saying, “Don’t be brainwashed by programming languages. Free your mind with mathematics”.