AI China Interview

YITU CIO Hao Lu on AI Startups, Innovation, and Coming Home

While a Computer Science PhD candidate at the University of Washington, Hao focused on intelligent human-computer interaction research, and R&D and evaluation of intelligent systems and their tools.

How can today’s innovative AI startup stand out in a sea of competitors? Some decide to double down on their core products, while others strategically advance into new business sectors. YITU Technology took a different tact: hiring a Chief Innovation Officer, former Google Research Scientist Hao Lu.

While a Computer Science PhD candidate at the University of Washington, Hao focused on intelligent human-computer interaction research, and R&D and evaluation of intelligent systems and their tools. During his time at Google, he was responsible for the incubation of the Android smart desktop and interactive App Store recommendations etc., published numerous papers in the HCI field, and obtained several patents.

But why would YITU hire a CIO — a position rarely seen in Chinese start-ups? What value will the new blood bring YITU, which scored US$60 million in its Series C funding round? And how does the new CIO view the current AI tide in China?

The best person to answer these questions is Hao, who recently participated in a wide-ranging discussion with Synced.

Synced: You have rich research experience in the US — you studied there, worked in academia, interned at Microsoft, and spent more than four years at Google. What was your motivation for returning to China now and joining a startup like YITU?

I’ve known YITU Co-founders Leo Zhu and Chenxi Lin for a long time: I met Leo while interning at Microsoft Research Asia, and Chenxi I know from my lab. Aside from trusting them a lot, Leo and Chenxi’s views and vision are very similar to my interests. In fact, I wanted to join YITU back in 2015, but had to shelve my plans due to other factors. Now, I feel like I can’t wait anymore, since I will miss this era if I wait more.

Currently, every aspect of YITU is picking up. Whether it’s resources or future expectations, YITU now has the ability and space to explore. There are also some patriotic reasons: I felt like I should come back and contribute more in China. So, joining YITU became a simple, natural choice.

IMG_20180123_110844-806x605.jpg

After former Google engineer Steve Yegge left Google, he wrote a blog post in which he suggested Google was no longer innovating. He later joined a ride-sharing startup in Southeast Asia called Grab, saying that was real innovation. You yourself left Google for a startup, can you express your opinion on innovation at Google?

I read that post as well, and I can understand many things he said in it. But I don’t think we can say one of the world’s biggest tech companies isn’t innovating anymore. Google’s founders both came from research backgrounds, but because Google’s products aren’t leading the industry, to a degree this could reflect Google’s less than ideal performance on product innovation, as Google is unable to turn its lead in technology into a lead in product design and vision, nor occupy the strategic high ground early. Hence, Google now has to continuously copy from other company’s products; and in this process of chasing the competition, its technical advantage doesn’t necessary fill the gaps in its products.

Startups have fewer employees and less-developed processes, while large companies have many employees and comprehensive, mature processes. From your perspective, which environment is it easier to innovate in?

I worked at Google Research for two years before moving on to Google Play. It’s difficult for large companies to reinvent their own products. One reason is that products like Google Play are deployed on a massive scale, which means innovation has little impact on performance. It’s safer to change at a slower pace and do more iterations. Everyone is accustomed to looking at solutions using the same benchmark, so it’s hard to think outside the box even for companies like YouTube. You can’t ditch a mega-sized commercial product for something entirely new.

Startups are more headstrong when it comes to innovation, not conservative like large companies. Startups must come up with new products, technologies, and corporate strategies.

Your research focuses on HCI and you have developed multi-touch gestures for Android and interactive recommendation tools, especially for combining demonstration and user deceleration. How is this research helping YITU?

My earliest research was on Computer Science theory. Then I wanted to research projects that could directly impact people’s lives, so I chose the field of HCI for its diversity and cross-disciplinary nature. Looking back the choice required quite a bit of familiarization as my background was fairly theoretical, working on abstract mathematical concepts and proofs.

I have dabbled in three fields over the past decade: traditional interaction systems, machine learning, and software engineering. What did I gain from this experience? A comprehensive perspective on products. Having dealt with hands-on design, interaction and the iterative process for many AI research projects, I place emphasis on the importance of engineering. The key is to focus on the product, which is also helping me innovate at YITU.

Having a Chief Innovation Officer (CIO) is fairly common in large companies but rare for startups. Some believe innovation is not a management task, but rather an internal evolution process. What is your take on this? What is an ideal CIO to you?

Innovation is a belief, and the role of CIO is useless without this belief. Innovation won’t happen through management, and the role of a CIO is not to innovate via managing others. Startups are innovative by default. But honestly, many Chinese start-ups just talk about innovative ideas — ideas are cheap and everyone has innovative thoughts. The difficult part is to improve upon an innovative idea and craft it into a product. I believe this is the point where many Chinese startups still need to improve. Right now they can be too impetuous and lack patience.

At YITU, my understanding of the role of CIO is that the new position demonstrates the emphasis that Leo and Chenxi place on innovation. Innovation is not only an evolutionary process, it also needs a force to push it forward. The process of innovation is usually very chaotic and filled with failures. It needs a lot engineering, a lot of research, and a lot of trying in various directions – a pursuit in uncertainty. I believe it is the CIO’s responsibility to acknowledge this, to support this, and to align with the CEO strategically.

How does YITU define the role of CIO? What effect will hiring a CIO have on YITU’s next steps?

Innovation has many layers, and is not only limited to technology, product, and strategy; for example a company’s management style and R&D direction also need innovation. I wish to propel everything on multiple fronts, and give everyone a new way of thinking.

Since joining YITU in January, I have talked to many people and worked on very front-line jobs. I have witnessed many things that we can work on, from R&D to product to design. Some product ideas even came out of these conversations. YITU is an innovative company, and have spent the past five years growing. How to break tradition and become more effective in R&D and innovation will be a very interesting task.

Will you create a new department in YITU? Will you pull in technical experts from across the company to form your team?

I like to do things step-by-step. Our current job is to realize product innovations under the existing team structure and using existing resources. We will change our way of thinking for some tasks, form some small teams, and attempt some breakthroughs. All of these are taking place now.

A lot of what we want to do doesn’t deviate much from our existing product line, so we don’t need to form a special team to innovate. What we are now doing more of is improving existing products and, on a much smaller scale, creating new product lines for our long-term goals. Hence, we barely needed any changes to staffing, instead what we need to change are our work focus and ideas.

YITU.jpg

Traditionally, product managers are responsible for the entire product design, including technical and business model selection. But in this era of AI, many product managers don’t possess the necessary skills to do so, and so a startup’s CEO often takes on this role. What is your take on this and how will you coordinate with product managers at YITU?

I am also looking for the answer to this question. In the era of AI the technical competency required of product managers is becoming higher and higher, unlike in the past, when new university grads could become product managers. I believe product interaction design is also becoming more and more important, further driving up the need for technical support and making life difficult for product managers.

A problem encountered by many startups is that their product managers are too young and lack in-depth knowledge on many issues, which means they have a hard time driving products. I’m currently working with YITU product managers on some projects.

YITU’s strength in computer vision is widely acknowledged, with its core businesses spanning security, healthcare, financial, urban data brain and intelligent hardware. How do you plan on innovating YITU’s existing technology and products? Do you have a specific area of focus? Will YITU modify its strategic roadmap?

Security, healthcare, and financials are all very solid industries, providing us with the necessary resources to support innovation. HCI is a very wide field which strives to improve the working relationship between people and machines. Hence, it can be applied in any industry: the security industry for example has police or other government agencies using the products; healthcare has patients and doctors; and the financial industry has banks and customers. You could say that I am looking at challenges in every industry we serve.

YITU’s strength lies in the many solid products we have made for these industries. But we cannot say we have solved all the problems, hence many opportunities remain for us in these core industries. By innovating with our products now, we can advance the application of new AI technology, which will allow us to do many different things in the long term.

Going forward, I have my own rough ideas now, but in general, the rate at which AI changes is very quick and so nobody can predict how the industry will look in one or two years. I do have some ideas we’re working on now, and we will soon be releasing a number of new products.

Chinese AI companies have grow at an amazing pace over the last year, with a big spike in papers published, funding raised, and talent poached from around the world. How does it feel to be a part of this rising AI tide?

On this question, I agree wholeheartedly with Leo, who says that talent, data, and the market have given China the opportunity of a generation. The most important of these elements is talent, and I believe one of the biggest reasons behind YITU’s decision to hire a CIO was to emphasize the value they place on talent, in hopes of attracting talented individuals to come and work with us.

In many companies I know of or have worked with, managers only create positions for company projects. This means many talented people in both the US and China may simply be cogs in a machine. Joining a company for a project is a very passive way to work. We want to be different, we are willing to create positions for talented people, and discuss a potential future with them based on their strengths or entrepreneurial spirit.

Of course, just creating a position for a person isn’t enough, the proper next step is to discuss with them how to realize the future. What we are doing is very robust, emphasizing engineering and research when we innovate. Talent is vital for this process, and we want YITU to be a place where talents become inspired.

From your perspective, what does a Chinese startup need to thrive? How do you determine whether a company is good or bad?

I believe first and foremost a company needs to have a solid, robust product. When I came back to China, one of my first impressions was that city streets were filled with bikes from bike sharing startups. Bike sharing did solve the last mile problem for commuters, but at the expense of turning cities into a mess. This demonstrated a key shortcoming for many Chinese start-ups: a lack of careful consideration regarding their products, which results in a lack of sustainable development for the entire industry. Sometimes, a little extra innovation can yield completely different results and make a world of a difference, but many startups don’t want to take this extra step because they are probably not patient enough.

There are many AI startups in China, and everyone is doing lots of PR. But the general public doesn’t understand the technical details and can’t tell the good from the bad. The way I see it, a thriving AI company not only needs to have good PR, it also needs to actually put AI into application and change peoples’ lives in a way they can see or feel.

Many AI companies are facing the challenge of turning theoretical research into products. Creating a product is not as simple as publishing a paper. Sometimes, research and product can be very, very far apart. To obtain research results, you may only need to make assumptions, but in real life, you will be faced with real problems. A company needs a solid engineering foundation to be able to quickly iterate its product — this is how research results can be transformed into good products and a good company.


Localization: Xiang Chen| Editor: Meghan Han, Michael Sarazen

0 comments on “YITU CIO Hao Lu on AI Startups, Innovation, and Coming Home

Leave a Reply

Your email address will not be published.

%d bloggers like this: