On October 12th, the CMU Summit hosted a fireside chat featuring Jason Costa, an Entrepreneur In Residence (EIR) at GGV Capital, and Eric Xing, who is a currently a professor at the Machine Learning Department, Language Technology Institute, and School of Computer Science at Carnegie Mellon University (CMU) and the co-founder of Petuum, Inc., an company that aims to provide AI and ML services across industries.
Professor Xing and Mr. Costa engaged in an interesting and in-depth conversation about both the history of and recent breakthroughs in the field of Artificial Intelligence (AI) and Machine Learning (ML). They also commented on the role of AI in human society and addressed people’s concerns for this rapidly developing technology. In addition, Professor Xing talked about his motivation for founding Petuum and his experience as an entrepreneur. Finally, Professor Xing shared his thoughts on the future of the AI industry and his expectation of CMU’s roles in shaping it.
A Brief Introduction to AI and ML
What is Artificial Intelligence? To many people, Artificial Intelligence is Alpha Go or Androids that have emotion. However, as Professor Xing pointed out, robotics and playing games is only one aspect of AI. Although he does not have a clear-cut definition of AI, Professor Xing is interested in the AI that is useful to human beings and to society.
Regarding the recent development in the field of AI and Machine Learning, Professor Xing commented that the fundamental ideas are actually not very different from those developed a few decades ago, a time when even a calculator was deemed as an AI because it fulfill the tasks that are supposed to be done by humans. In addition to innovations in model design, Professor Xing introduced that the major differences are the computing power and the data size. Although we are still using previous algorithms, the dramatic increase in computer’s memory size allows for transformation of massive amount of data into useful models.
Another change Professor Xing mentioned is that the barrier to enter the AI/ML industry is becoming lower. As a result, many people are drawn to this field and are only interested in learning a particular technique or a specific programming language. He was not particularly optimistic about this because he observed that the number of people who actually understand the concepts behind ML is actually decreasing, which is an upsetting reality to a professor.
AI’s Impact on Human Society: Should We Be Concerned?
Will AI will initiate a WWIII? Professor Xing reassured many people’s concerns. In his opinion, AI is never meant to be a threat to people. In fact, he thinks that we should embrace the change because we are currently in an age of technological revolution, just as how humans went from fire to electricity, from candle to light bulbs.
Professor Xing also does not consider AI as a direct threat to people’s jobs but thinks of it as a remedy that emancipates humans from monotonous and repetitive tasks while still giving them opportunities to do the jobs creatively. For instance, although photographs taken by digital cameras are better in many aspects, there are still a lot of people who prefer portraits painted by painters.
The Motivation Behind Founding Petuum
At the fireside chat, Professor Xing revealed his motivation behind founding Petuum. Despite the high demands for AI/ML solutions, current AI/ML products are just like the very first light bulbs: they are unaffordable, inaccessible, and are not mass-produced. Thus, Petuum is created to fill the industry gap. Professor Xing said at Petuum, “we are not trying to invent computer algorithms that are 5% better than [those of] others, but we aim to transform them into extremely robust products that will benefit all industries.”
Life As An Entrepreneur
Professor Xing admitted that every aspect of being an entrepreneur is challenging. Intellectually, what entrepreneurs do for companies is dramatically different from what researchers do in labs. Besides producing actual results, you have to convince other buyers that the result is consistent and reliable. Socially, entrepreneurs have to be able to obtain resources, to communicate ideas to investors and employees effectively and patiently, and to be prepared to answer a lot of questions. Psychologically, as the leaders, entrepreneurs have to ensure that everything is done perfectly, even if they are not responsible for that particular task. Moreover, being an entrepreneur is even a challenge to stamina and physical power.
As the CEO of Petuum, Professor Xing told us that, because they solve cross-disciplinary problems, he is looking for employees that are extremely knowledgeable and well-trained. They have to be responsible and have team spirits, which is very important for completing tasks efficiently and effectively. For students interested in entering the AI/ML industry, Professor Xing considers the ability to independently identify problems very important. Because at the end of the day, it is not enough to just have hardcore skills like the ability to program in Tensor Flow — new programming languages and new methods are quickly emerging so you have to be prepared to change gears.
What’s Next for AI and for CMU?
Professor Xing personally does not think that there will be an “Age of AI” in the near future because there are still some technical bottlenecks and many concerns over the use of AI. Moreover, every company in the AI industry is developing their own AI products in their own way. This is problematic since there is little transparency in the field and the works are being done in a highly repetitive and inefficient manner. Although the current “AI Solution” industry is dominated by large firms with massive datasets like Google and Facebook, Professor Xing predicts that once AI/ML solutions become a standardized commodity, the entry barrier will be lowered and many new companies will emerge and join the competition. That will be the start of the “Age of AI”.
Professor Xing also reminded us that there will not be a single technology that leads to a revolution in the AI/ML field after Deep Learning but rather, a collection of innovations. For example, when people think of the technology behind Alpha Go, they might only recall Deep Learning, which is massively publicized for marketing reasons. However, there is actually a lot more technologies and innovations that make Alpha Go possible.
As for CMU, Professor Xing thinks that although CMU is a leader in the revolution of AI technology, it is underappreciated due to its ability to push for AI entrepreneurship. This is partially because the students here are the best engineers in the world but not yet the best thinkers or opinion leaders. Thus, he encouraged CMU students to think about AI not only in terms of technology but also in a more socially-responsible way.