When Wilson Pang graduated from Zhejiang University in China, where he grew up, with Bachelor’s and Master degrees in electrical engineering, his career plans didn’t include working in artificial intelligence.
Mostly, that’s because artificial intelligence was still the stuff of science fiction. From his early days, Wilson fell in love with technology. In 1993 while in high school, he wrote a program on an Apple computer and felt like he had created a “small world.” This passion for technology led him to start his career as a software developer at a startup and later on, IBM and eBay.
From IT to AI
For Wilson, making the transition from IT to AI was smooth. While working with eBay, he started to work on search science, making sure customers were getting the right results based on their search terms. While not AI as we think of it today, search science was a pre-cursor to today’s AI by using machine learning and algorithms to help people find what they’re looking for.
Part of what Wilson found fascinating about search science was the math. “There’s no kind of, no absolute answer like why this item has to rank number one. Or why the other item is ranking number two. It’s all about probability. It’s all about what you really want there.”
In order to focus on this new career path, Wilson spent the next two years taking weekend and evening classes and learning everything he could about machine learning and AI. Looking back on it, he says it’s one of the best decisions he’s made in his career.
Another aspect of his transition from software developer to CTO of Appen came in the form of data management. At eBay, Wilson began building cross-platform data solutions to make it easier for any internal team member to access the data they needed. A few years later, he became the Chief Data Officer at CTrip, where he managed the analytics, data reporting, and machine learning for the entire corporate group.
After transitioning to Appen in 2018, Wilson got a whole new perspective on data. No longer was he managing data from a single industry, but he was observing how data was being used across a wide variety of industries and use cases and seeing the broader effects.
While a passion for data led him to Appen, Wilson doesn’t take the leadership role he’s in lightly. How he views his leadership style is as a servant leader who’s there to empower his team to achieve the best business outcomes.
Instead of issuing orders, Wilson sees his job as CTO of Appen as “how you can inspire people for the overall vision, while also empowering them to take ownership and be accountable. At the same time, as the leader, you’re there to provide support and resources while unblocking pathways that are obstructed.” When talking to his team, Wilson always reminds them to “treat me like a resource to help you succeed.”
Wilson goes on to recognize that hiring the right people is a crucial part of team success. As a leader, his job is to not only be there to support his people, but to help them grow and transform. That starts with hiring the right people. Once you have the right people, you simply have to nurture and support them in their growth as they help the company to grow.
Wilson’s View: The Future of AI
Another aspect of Wilson’s job as CTO of Appen is to be thinking about the future of AI and machine learning technology. In this rapidly changing, quick-moving industry, Wilson has to be looking towards the future at the same time as managing the present.
Exciting New Initiative: Workflow
While the newest, cutting edge technology is always exciting, what Wilson is most excited about this year is Appen’s Workflow program. Many AI and machine learning projects involve multiple steps and a lot of back and forth. Currently, most project platforms require that each step of an AI project be a unique project.
Appen saw how this overcomplicated projects for customers and wasted time and money by ramping up again and again. With Workflows, Appen customers can import multi-step projects to create a single project.
For example, a customer who is building a voice recognition engine will need only one project to collect voice data, transcribe the data, and annotate the data. At other companies, that would be three separate projects, even though it’s all the same data for a single goal.
With Workflows, projects become more efficient and cost-effective. A win-win for suppliers and customers.
Overhyped Tech: AI-Annotated Training Data
One of the exciting recent initiatives within AI is the ability to annotate training data without human intervention by using AI. In Wilson’s opinion, this won’t be possible, at least for a very long time. AI is great at pattern recognition and repetitive tasks. But, it still can’t match a human mind for intelligence and understanding.
Training data that’s annotated by AI may reduce the cost of creating training data, but it doesn’t create higher quality data. Instead, Wilson believes in a combination of human and AI annotation. Let AI do what it does best: annotate the data through pattern recognition. Then, bring in human annotators to complete a quality assurance check.
This combination of using human and AI annotation creates an efficient, accurate process that puts out the highest quality possible AI training data.
In the wide world of AI, Wilson also points to GPT-3 as over-hyped AI technology. While GOT-3 and other language AI can write an article, it’s not going to put writers out of business. AI is logical. While humans are logical and creative. AI is just learning from the data that it’s fed. It can’t access the type of thinking that humans have. While the technology is innovative and exciting, it’s not going to be replacing humans anytime soon.
With his attitude of servant leadership and drive to think critically about the future of AI technology, we feel so lucky to have Wilson on the Appen team.