In conversation with the president of The Mohamed bin Zayed University of Artificial Intelligence

Wednesday, June 23, 2021

What you need to know about The Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), the world’s first graduate-level, research-based artificial intelligence (AI) university. And what you need to know about “AI epicenter research” in the region and the world, all here in HBR Arabia interview with Professor Eric, the president of MBZUAI, who is one of the world’s leading AI scientists. Before joining MBZUAI, Professor Eric was a professor of Computer Science at Carnegie Mellon University, and the Founder, Chairman, and Chief Scientist of Petuum Inc., which was recognized as a 2018 World Economic Forum Technology Pioneer that builds standardized artificial intelligence development platforms and operating systems for broad and general industrial AI applications. Professor Eric bets on MBZUAI to become a global AI hub where the world’s greatest minds gather to demonstrate to the world how a real AI future is going to look like. I know all of us have our questions in mind for this global leader and expert, that is why I tried to sum as much as I could in this detailed conversation.

What is the concept behind MBZUAI given its unique specialization?

Professor Eric: MBZUAI is the result of a very bold vision and concept. Start with the name – MBZUAI, which bears the name of His Highness Sheikh Mohammed bin Zayed Al Nahyan, Crown Prince of Abu Dhabi, is a testament to the commitment of the leaders here to higher education. Then, this is the first time we have a ‘University of Artificial Intelligence’ anywhere in the world, which highlights the importance of AI as a scientific discipline and its foundational role for society and the country.

There are not many universities in the world that specialize in a specific topic. MBZUAI is a bold idea that’s unleashed a lot of new opportunities for us to push the boundaries, I want to commend the leaders of this country for giving us the opportunity. The definition of a university comes from ancient Roman Greek terminology meaning universe or everything. So, having a university focused on just one discipline is groundbreaking.

AI is a major topic where you are trying to build a connection between AI and data science. Without exaggeration, AI touches every discipline – chemistry, physics, biology, social science, mathematics- which all produce data or principles for understanding data. AI becomes a tool for researchers in all these domains to do their jobs or a vehicle for them to ground their work on applications that can impact all walks of life and become the engine for the economy. AI deserves a university-level foundation. There is a need to investigate and explore how to nurture and further advance AI, and there is a need to train the next generation of workforce proficient with AI to advance the economy. Society needs to explore such new opportunities to secure a future for growth and development. AI has the potential to advance this function and deliver such results.

Our mission is therefore clearly defined to be the hub of AI fundamental research, teaching, innovation, and technological transformation. That is, a graduate university focusing on AI. We aim to be the epicentre of research in the region and the world and then to undertake first-class fundamental research and education.

Typically, we remember universities because of the faculty, their research and their role as a centre for learning.  So, to apply this to the concept of MBZUAI, we have a broader mission of creating an impact on the culture and also on the way people perceive research in AI and understanding what AI is and how it interacts and impacts our society and what kind of problems it solves. That’s why I think this project is extremely important and also why the university deserves the name and association with our Crown Prince.

As MBZUAI a post-grad university, do you have an MBA, master degrees, or only research-based degrees?

Professor Eric: Because this is a graduate-level research university for AI, the degrees we offer are masters and the PhDs in AI. That said, AI is a vast field and we have subdivided the field into specific disciplines such as Machine Learning, which is the methodological foundation of artificial intelligence, and then Computer Vision which uses AI technology to solve visual-related problems, and finally Natural Language Processing that lends itself to anything language-related and speech-related. And in the future, we aim to venture into systems for AI concerning the computer infrastructure and architecture for artificial intelligence. Eventually, we will venture into AI in healthcare, AI in finance and other sectors. Along the way, I envision that, say in three or five years we may see a need to branch into more disciplines beyond just AI.

We’ve got the first cohort from 29 countries and we are in the middle of the second group. The first batch was only presented with two majors – Machine Learning and Computer Vision. The second group will also include Natural Language Processing (NLP) which has been added to the majors.

What are the criteria for accepting students? What kind of characteristics and skills do you look for in an MBZUAI candidate and why?

Professor Eric: We have high criteria which are pretty standard for all university and graduate programs, for example, they need to have a certain Grade Point Average (GPA), TOEFL, Graduate Record Examinations (GRE), and certain other things. We have adopted some of the best practices from top universities in the United States. We require the students to have written a statement of research, to articulate their own career goals and their interest and the type of topics they want to work on. We also ask them for a full CV, as well as recommendation letters from their supervisors, to see whether they have ever interned in certain corporations. Other considerations include whether they have written any papers in research conferences; whether they joined specific research projects during their undergraduate studies or internships; their character in terms of teamwork, collaboration, leadership, self-discipline, and independence, etc. These are all aspects beyond what their GPA can reflect. Thus, transcripts by themselves do not carry too much weight. We are interested in their experiences, such as whether they came from a real research background and related fields, their maturity for advanced study and hard work, and their passion and goal for research in AI.

Typically, there are boxes a graduate student needs to tick, such as having basic training in mathematics, computer programming, and algorithms and theories, and MBZUAI is no different. Furthermore, students need to have a curiosity and a passion for AI as a scientific discipline and also as an engineering domain so they can find joy and satisfaction in studying and working in this area. That’s very important. People do their best if they are inspired by a sheer sense of interest and passion rather than a material goal. A material goal is important, but it shouldn’t be the only driver. I look for evidence in candidates for their passion and enthusiasm. I also look for students with independent mind. They should be comfortable and be looking for independence and freedom to do their research and to implement their solution and go out to become an enabler and a pillar in whatever service they provide in their job. The ability to be a team player, to communicate and interact with people is also very important. So academic preparation, a passion for the topic, a mindset of independence and freedom, and the ability to be a team player is what I’m looking for in MBZUAI applicants.

What do you promise students from such a unique university?

Professor Eric: We are a new programme that aspires to compete against the top schools. The first promise is that they get the opportunity to be trained by the best researchers in the world. We worked very hard to bring these great minds to MBZUAI. For example, the chairperson of our machine learning department was a star professor from the Georgia Institute of Technology (Georgia Tech), one of the best computer science schools in the United States. And we are in the process of hiring several other researchers from other top faculties in Europe and the United States and other regions in the world.

Secondly, we need to make sure that students are taken care of and they are not distracted by things like financial need, living conditions and other essentials. We, therefore, provide them with a very generous fellowship so that their cost of living and study here at the university are fully covered. And they also are provided with excellent accommodation, health insurance and even their travel to Abu Dhabi is also covered so that they can focus on learning and then innovating down the road.

It’s said that AI is not one size fits all. It’s something customized to you, to your data. Can we implement a system, software, solution of AI for individual users? And the second part of the question is about how fast is AI coming? Some observers seem to think AI will arrive suddenly and our lives will be changed dramatically.

Professor Eric: AI is like a human being, which means no one size fits all. No individual is good at everything. They could be an engineer, mathematician or artist and so you have to train them in different ways. Thus, in AI, the so-called artificial general intelligence is more of an inspiration. In my opinion, AI would adopt a “divide and conquer” approach where different problems are addressed through a specific approach and implementation both economically and scientifically. Although it doesn’t mean that you have to go to the finest possible granularity and solve every single task using a single software. Sometimes there is an aggregation of similar tasks. For example, in smart city applications, computer vision can recognize objects on streets, in buildings and in public areas, a similar cluster of problems that a particular AI solution can tackle. In a hospital setting, on the other hand, AI can read and interpret a CT scan and X-ray. This is how humans divide their tasks. Similarly, for AI, I do see a more practical and even an essential need to recognize this division of expertise and strengths. But that doesn’t mean that you have to bring or build every solution completely from scratch. I emphasize the need for standardization and reusability. When you come across two different AI systems, and you don’t look at them as a Black Box but as an assemblage of building blocks, you start to recognize that 80% or 90% of them could be reusable. Perhaps they simply require a different pair of eyes (i.e., the signal processor) to look at different pictures. Or maybe, they need to rewire the brain (i.e., the reasoning algorithm) slightly to be able to recognize both the face and also the organs in an X-ray.

There are reports that hundreds of millions of jobs will be suddenly eliminated and replaced by AI while other experts anticipate a more incremental adoption of AI incrementally. Considering the digital shift we saw during the pandemic, could we see a sudden shift in how the world implements AI, how do you see the future?

Professor Eric: AI development at least from a technological aspect is continuous. There is no sudden jump or a singularity point. and it’s all very cumulative if you look at the entire history of AI development from a technological standpoint. There has been fear as it slows down a bit and sometimes its a rapid jump, but there is this intrinsic character that you need to have accumulation before any breakthrough. So I don’t see any unpredictable shift on the horizon. For example, I remember five years ago people were predicting that by now we would be riding autonomous vehicles almost everywhere, which has not happened of course. The real experts could foresee the current situation because there are technical, hardware and infrastructure hurdles, which all take some time to resolve. But the phenomena during the pandemic with a seemingly sudden transformation into a new norm is impossible. What is behind any AI-driven transformation it is a complex combination of both technology, policy and societal needs among other aspects. For example, back to autonomous driving, I can’t imagine the possibility of creating special roads like railways for autonomous vehicle to mitigate the risk factors associated with driverless cars; but I do expect policies, legal and insurances practices, and even human perception and attitudes will graduate evolve along with the technology. Ultimately, it is not AI advancement alone that drives the adoption. But in any event, I don’t see the need for mass elimination of jobs and human involvement because of AI. I think that the positioning of AI productivity against human productivity is an imaginary one. They are not competing. I think AI and humans are coevolving and coexisting and will ultimately complement each other. What AI is very good at is what humans are not very good at while what people are very good at plays into the weakness of AI. Therefore, I don’t think there’s a need for them to target the same problem. For example, calculators are way better than people in doing Algebra and Arithmetic, but we did not see the invention of the calculator eliminating many jobs. It created more jobs to make calculators and helped humans become more productive. Vehicles did eliminate the jobs of horseman and the carriage person, but again that necessarily created more jobs for people who build cars. So I see AI as a great opportunity for mankind to make themself more productive and live a more pleasant life, even in their work environment.

Which skills should students or people reading this need to be prepared for?

Professor Eric: Not everyone needs to be an expert on AI. But I think AI literacy is an important skill or perhaps even a fundamental ability that future generations should have. They should know what AI can achieve and what’s the boundary of that and then position their career plans and their lifestyles accordingly. Otherwise, there is going to be a problem in expectation management and professional preparation, both for the self and for the community. For example, if you imagine AI as a powerful creature that can do everything, then obviously you will give up doing anything because you will be counting on AI. But if you know that AI can only do certain things but requiring others then you can position yourself to play a complementary role for example, how to be a creator of AI or an operator of AI. So in terms of job training, I think the ability to continuously learn will be the most important skill.

For future generations of the workforce.

The future of education and future of skills

During the Coronavirus crisis, universities in some countries here in the region like the UAE managed to smoothly transition online while countries’ education systems came to a complete halt. So for the education system in general, what should be done now to survive the disruption coming from online mini-courses, Moocs, etc.?

Professor Eric: From an infrastructure perspective, the university should stay ahead of the curve and be prepared, not only for the present but also for the future. I’ll give you an example — when the pandemic hit and universities were forced to go online, it took Carnegie Mellon University one week for the entire university to switch to online education. Even research activities made the switch very quickly because the university’s infrastructure and the skills of the teachers, researchers, and admins were already set up. But what’s more important for education to be essential and indispensable is the environment. It needs to provide a unique environment found nowhere else for innovation and training and research. A good university defines a unique environment that provides the resources, intellectual freedom, and also the kind of stimulating microsystem for people to cross-fertilize and to incubate new ideas and carry out various experiments that’s not just possible anywhere. Universities should take every endeavour  to implement such an environment and also to introduce the right incentives to encourage students to do that. Incentives mean how people get recognized, how people get rewarded for pursuing their passion and at the same time generating values and impact on society. As a University President, I’ve been thinking very hard about how we can create such a stimulating environment for great minds, so they can spend a few years of their career on campus to experience and contribute to this intellectual environment.

As a university specializing in AI, how do you suggest the design of the curriculum be aligned with the business communities and show them the real outcome of your research?

Professor Eric: Certainly, there is a great need from the industry, from the business community that draws talent and intellectual power from universities to solve their problems. And the university must take responsibility to address those needs. Silicon Valley is a good example of this as well as a few other areas in the US, Europe and China. I think that the best thing is a symbiosis between the university and also the startup and the innovation ecosystem around it, centred on the universities’ intellectual power. That means the university will implement the right policies and also the right kind of organizational infrastructure. There is a critical need for technology transfer and the right kind of policy to allow the faculty to actively work with industrial partners and solve problems. And we should have an environment to encourage our faculty to be entrepreneurial and build startups based on their inventions and innovations.

Is this part of your mission?

Professor Eric: This is part of our mission. Next to our basic research and education, we have put in the mechanism for incubating startups and for supporting entrepreneurs. I wouldn’t be surprised if within a few years we see some startups emerging because many of our faculty have students with great potential and promising ideas. We also have great partners, like ADNOC for example, that are looking to us to collaborate. Our University already works with Hub71 to develop incubators. We have looked into how our faculty and students can become more active on GitHub and other such platforms to encourage other aspects of innovation and entrepreneurship. I hope MBZUAI can function as a catalyst and also the epicenter of this technology transfer and innovation that is going to impact not just the home countries of our students, but also different sectors and the needs of communities everywhere.