For the last two decades, the life of neuroscientist David Cox, 40, has revolved around one single theme: artificial intelligence (AI), the branch of computer science that aims to create machines capable of emulating the human ability to reason, make decisions, and solve problems. After graduating from Harvard University with a degree in biology and psychology, studying a PhD in neuroscience at the Massachusetts Institute of Technology (MIT), and becoming a professor at Harvard’s Center for Brain Science, he last year took over as director of the MIT-IBM Watson Artificial Intelligence Lab.
“This project, a collaboration between IBM and MIT, represents an entirely new partnership between business and academia. Its focus is on fundamental research in artificial intelligence,” explains Cox. One of the lab’s objectives is to develop algorithms, devices, and architectures that enable the creation of new artificial intelligence solutions. “It’s hard work and it requires a lot of inspiration,” he says.
Considered one of the world’s leading experts on AI, Cox recently participated at the event Colloquium 2018 – Artificial Intelligence Today: Advances and Opportunities in the Industry, held in São Paulo by IBM. Between conferences, he spoke to Pesquisa FAPESP about the research being conducted at the MIT-IBM lab, recent advances in AI, and what we should expect from this field in the future.
What is artificial intelligence?
Artificial intelligence [AI] means different things to different people—there is no single accepted definition. For me it means giving computers the ability to make decisions that help us. AI is a way of augmenting our lives. It is about automating skills.
Is an in-depth understanding of how the human brain functions important to creating more sophisticated AI solutions?
I do not think it is essential, but it is one way of doing it, and it is a great source of inspiration. It is helpful to keep an eye on studies of the brain when it comes to artificial intelligence, but it is not essential. Inspiration can come from a wider range of sources than you might imagine, including medicine, physics, and biology. There are many opportunities to learn and they all intersect in various ways.
AI advances have the potential to change the world and bring real benefits to humanity
Are we close to creating a machine that thinks like a human?
That is an interesting question: What are the goals of artificial intelligence? Are we trying to create an artificial intelligence that is exactly like us? That might be a long-term goal. Today, however, what we are really trying to do is to build AI systems that can help us, not that can be exactly like us, but that work in the same way our minds do. We are not trying to create something to replace us. We would love to have artificial intelligence with human capabilities, but the goal is not to replicate humans.
Many people think the aim of AI is to copy the human brain. What you are saying is that the goal is actually to make it even better. Is that possible?
Yes, and there are already some good examples. Computers are better at math and at performing complex calculations than us. Self-driving cars, too, are a major current application of artificial intelligence. When humans drive, we use our hands and our eyes. Cars, though, can be much better equipped. They can have multiple cameras and sensors that provide information on direction and speed. These systems process information faster and have more sophisticated capabilities than the human mind. You do not necessarily have to teach the car to drive the same way as a human.
Is human behavior taken into account when creating technologies based on AI?
There are many applications of AI where understanding human behavior is very important, not least because we are building AI for humans, to be consumed by humans. In the case of autonomous vehicles, for example, we must take into account several human factors, such as the presence of pedestrians crossing the street in varied and sometimes adverse situations.
In your opinion, what are the main contributions to society of applied research into AI?
There are already many services and products available thanks to advances in AI, especially those that are part of the IT [Information Technology] infrastructure behind the scenes. These systems are invisible to the public but help make decisions and make life more efficient in various sectors of society. IBM has a whole division dedicated to AI for social good. With the support of artificial intelligence solutions, applied science and technology can solve many difficult social challenges. There will definitely be important advances in health, as well as in agribusiness, where many applications designed to increase efficiency are already emerging. In education, too. As AI systems improve, they will be better able to interact with and teach people. There are very interesting opportunities on the horizon.
During the 2018 Colloquium on AI in São Paulo, you mentioned that two of IBM’s main research targets are health and security.
Health and cybersecurity are high-priority issues with important needs. Individuals, businesses, and governments have all been targeted by cybercriminals and artificial intelligence can really make a difference. In health, the aim is to help doctors make better decisions when analyzing medical data and images. AI has the ability to encapsulate human knowledge into a form that can be used to support doctors. One of our dreams is to help make the best doctors even better than they are today. There are also studies into AI that are helping us to better understand molecules and biology, and to develop better drugs. It is a hard journey, and nothing happens overnight.
Uploading our brains is a theme that has captured the imagination of many people, is that something based in reality, or is it just science fiction?
Uploading our consciousness [copying a person’s mind onto a computer] is purely science fiction. There are people interested in the subject, but it is not real. It is fun to talk about, but it is something we are still very, very far away from. This topic was actually mentioned when I was giving a talk at the World Economic Forum in 2017, and I immediately dismissed it. It is not reality. It is just Hollywood.
How much is being invested in AI around the world today?
I don’t have that answer off the top of my head, but governments and industry are investing billions of dollars in AI projects worldwide, from China to the USA and Europe. Every government and every industry can see that this is going to be strategically important for the future. We are only at the beginning of the journey, but there is already a lot that artificial intelligence can do. It is exciting to see all the possibilities ahead.
What industry demands could AI advances help to address in the coming years?
There are a number of problems faced by companies that AI is going to be able to resolve. One of them is what we call small data. We talk about big data all the time, but even when you have a lot of data, it often may not be the best data or may not be in the right form. I think there will be breakthroughs in the next five years around being able to do more with less data—without needing to organize and interpret huge volumes of information. With AI, a lot of progress is being made in this direction, and AI systems will help curate data, allowing companies and their employees to accomplish more with less.
What is the focus of the research being conducted at the MIT-IBM Watson Artificial Intelligence Lab, established in 2017?
Our focus is on fundamental research. The lab was created as a result of IBM and MIT getting together and asking what we really need to do to unlock the potential of AI. The idea is to take AI to the next level, so a wide range of research is conducted by the lab. What we are asking is what technologies do we not have today that would enable us to tackle the problems of tomorrow. We try to invent new approaches to AI that allow us to develop applications and address unsolved problems.
How much is being invested in the laboratory? And where are the funds being used?
A total investment of US$240 million has been committed for the next 10 years. This money is spent on machines and processing, software development, and personnel. At all organizations, people are the most expensive resource.
What kind of projects does the laboratory work on?
We organize our projects into four pillars. The first is the fundamental development of advanced algorithms to expand the capabilities of machine learning and reasoning. The focus is very broad, not aimed at any specific task, and addresses the most complex problems. New algorithms that not only process huge amounts of information, but also learn from limited data to augment human intelligence. At the same time, we investigate new materials, devices, and architectures that could support future approaches to AI training and deployment. The second pillar is the physics of AI. Research into quantum computing, for example, to optimize and accelerate machine-learning algorithms and other applications of AI.
What are the other two pillars?
The third is application of AI to industry. We look at AI advances through the lens of industry problems in order to understand their demands and needs, as already happens in the areas of healthcare and cybersecurity. The last pillar is what we call shared prosperity. We explore how AI can provide economic and social benefits to a wider range of people, countries, and businesses. We study the economic implications of AI and investigate how it can bring prosperity and help individuals improve their lives.
What are the biggest challenges faced by the lab?
It is hard work and it takes a lot of inspiration and new ideas. Many other professionals, from large universities and companies, are investing in this industry, so it is becoming very competitive. We try to work on things in a slightly different way from our competitors, but we are all striving to achieve the same goal. Artificial intelligence has the potential to change the world and bring real benefits to humanity. It can make our lives easier, healthier, and safer.