CF Su Of Hyperscience On The Future Of Artificial Intelligence

An Interview With David Leichner

Limitless Possibilities: We’ve yet to see an AI tech ceiling, especially around the power of LLM. I like to say that I’m at the top of the roller coaster looking down, unable to comprehend what’s next or how far this LLM technology can take us.

As a part of our series about the future of Artificial Intelligence, I had the pleasure of interviewing CF Su, VP of Machine Learning, Hyperscience.

CF brings over 15 years of R&D experience in the tech industries. He’s led engineering teams in fast-paced start-ups as well as big Internet giants. His expertise includes areas of search ranking, content classification, online advertisement, and data analytics. Most recently, CF was the Head of Machine Learning at Quora, where his teams developed ML applications of recommendation systems, content understanding, and text classification models. Before that, he held technical leadership positions at Polyvore (acquired by Yahoo), Shanda Innovations America, and Yahoo Search and was a senior researcher at the Fujitsu Lab of America. To date, CF’s industry contributions include 14 U.S. patents and more than 20 technical papers.

Thank you so much for joining us in this interview series! Can you share with us the ‘backstory” of how you decided to pursue this career path in AI?

Like many others, my career path happened by accident. After completing my Ph.D. from the University of Texas at Austin in Electrical and Computer Engineering, I worked at the Fujitsu Lab of America in Silicon Valley for a few years. It was a corporate research lab for conducting research, publishing papers, and transferring technology to product development in the telecommunication domain.

I sat on the fence for a while, trying to decide between an academic research career and a software industry career. Ultimately, I put my academic dream on hold and moved to Yahoo’s web search engine team, which applied machine learning (ML) capabilities to search results ranking. It was an excellent first taste of machine learning, and I quickly found my desire to keep working with remarkable technologies.

From the web search engine, I moved into adjacent areas, like online ads, e-commerce, recommendation systems, and content moderation, where ML is applied to serve content to optimize user engagement on websites and applications. Most recently, I moved to enterprise software to work on digital transformation and automation in the business world using ML. I firmly believe that many applications of AI/ML can benefit society beyond increasing user clicks on websites, especially those areas currently underserved by technology, such as public sectors, financial institutions, insurance, and health care.

What lessons can others learn from your story?

Don’t be afraid to move outside your comfort zone to explore new opportunities in a different industry. Consider exploring an adjacent field first — as I did after Yahoo when moving from web search engine to online ads — and strive to develop additional skills and grow existing ones to help you advance. It will pay off in the long run.

Perhaps more importantly, always watch out for new technology and assess its long-term impact. Some, but not all, lead to a true paradigm shift. Several great examples include the beginning of the Internet and web applications, the change from PC to mobile computing, the transition to cloud-based infrastructure, the rise of social network apps, and the rapid growth of ML/AI. If you see a significant technological shift that aligns with your interest, it is time to learn as much as possible and get in on the ground floor.

Be wary of all shiny new technology, though. There will always be a new and exciting technology that promises to change the world, but approach it with skepticism and be mindful of its implications. I’ve worked with many technologies throughout my career and believe it is important to take the time to understand their capabilities, limitations, and potential before falling in love with them.

Can you tell our readers about the most interesting projects you are working on now?

At Hyperscience, I spend a lot of time working on Intelligent Document Processing (IDP), which involves transforming human-readable content into actionable data. It requires applying ML to extract data from documents, understand the context, and make decisions. We look to automate the most process-heavy and error-prone workflows to improve business and customer outcomes with higher accuracy and lower cost.

One unique use case is the benefit claim processing at a federal agency and our efforts to turn information into actionable data to accelerate the approval process. As many readers know, most government activities are driven by forms. Need to file your taxes or get a passport? Fill out a form. Thus agencies need to process millions of applications, claims forms, and tax documents, which come in various formats with messy handwritten text or low image quality. It takes lots of time and cost to process them.

Our IDP product captures critical information and processes claims more quickly, so this agency isn’t faced with severe backlogs, and citizens receive the benefits sooner. As a result, the average processing time of a claim was reduced from 3 weeks to half a day. This increased (accurate) speed ensures that individuals who need financial relief in a difficult time aren’t left waiting for long. While this is a small use case, we see the incredible, real-world impact each day of how our technology is deployed, and it’s gratifying.

None of us are able to achieve success without some help along the way. Is there a particular person who you are grateful towards who helped get you to where you are? Can you share a story about that?

I received my bachelor’s degree in Taiwan before heading to the United States for my doctorate. I was new to the country and looking for a source of inspiration, which I found in my Ph.D. advisor, Professor Gustavo de Veciana, at the University of Texas at Austin. My advisor taught me many things throughout my program, from rigorous research frameworks to “writing well.” I was very fortunate to be his first Ph.D. student, and we spent a lot of time working together. My time at Austin got me acclimated to the thrilling tech industry. My research was about the intelligent management system of telecommunication networks, and he was the first person who exposed me to the emerging Internet (it was still the early days). He asked me to set up the first-ever website for our research group, which was an eye-opening experience as it impressed upon me how powerful the internet could be. It planted the seed in my mind to work in Silicon Valley after my Ph.D. to observe the Internet revolution and the dot com bubble later.

Looking back, the interest he took and the skills he helped me develop were why I stayed in Silicon Valley instead of returning to Taiwan to become a professor. If not for him, I may not have had a front-row seat to the technological evolution we’ve witnessed during the past few decades, and I’ll be forever grateful.

What are the five things that most excite you about the AI industry? Why?

There’s so much to be excited about. AI technology has gotten so good that users don’t necessarily need an advanced degree to master it; instead, they just need experience working with these various platforms. Specifically, here are five things I’m looking out for:

1. Major Steps Forward: In the past few years, the industry had major AI and Natural Language Processing (NLP) breakthroughs around a technology called Large Language Model (LLM). A recent example is the widely popular (and hyped) ChatGPT. Now lots of investments are going into building similar platforms and new applications. Though they’re still in the early stages, these models have so far provided reasonable responses to user prompts — and with fine-tuning, the potential is limitless.
2. Creative Expression: We have been using AI to classify images and recognize objects with great success since the 2000s. The newer Diffusion Models (e.g., Open AI’s Dall-E 2, Google’s Imagen) allow us to generate pictures and artwork based on text input. This was a groundbreaking development for creative art and expression.
3. Limitless Possibilities: We’ve yet to see an AI tech ceiling, especially around the power of LLM. I like to say that I’m at the top of the roller coaster looking down, unable to comprehend what’s next or how far this LLM technology can take us.
4. Human-Machine Compatibility: Advances in NLP have made it more likely to have a natural language interface between humans and machines. The ability to understand spoken and written language will significantly change how we use our tech gadgets, and how humans interact with technology in general.
5. Growing Public Awareness: Society has reached a point where AI technology has become mainstream and part of everyone’s vocabulary. From self-driving cars to search engines that can converse, we’re becoming more comfortable with AI, which will help accelerate future growth. I expect a productivity boost like those accelerated by the arrival of PCs and the Internet in the past.

What concerns you about the AI industry? Why?

Unsurprisingly, there are plenty of areas of concern regarding AI and ensuring it is developed and deployed ethically and thoughtfully. There are four in particular that come to mind:

1. Transparency and Explainability. AI and machine learning have become very complex, making oversight increasingly challenging. We must overcome the public’s concerns about transparency and explainability to increase AI’s adoption. The White House’s AI Bill of Rights touches upon ethical guidelines that AI leaders should consider applying to AI technology. It is a good start, and it remains to be seen how the industry progresses in this direction.
2. The Winners Take All. Data is the new oil in the 21st century, and big technology companies are often the only players with access to large amounts of user data, giving them a unique competitive advantage. LLMs require enormous computing power (in addition to data) and cost lots of money to develop. We must find a way to lower the entry barrier to ensure large tech companies don’t monopolize AI’s future innovation and capabilities.
3. Boom and Bust Cycle. Recent months have seen generative AI hype and high levels of investment. However, in the past decade, we’ve experienced several ‘AI Winters,’ where AI technology interest decreased cyclically. I wonder whether we’ll face another AI Winter or if it will be AI Summer forever. We need to have realistic expectations of AI to avoid the boom-and-bust cycles.
4. The Trust Gap. With generative AI in the public’s reach, we face the danger of trust destruction. Think of it like robocalling: many of us are now skeptical about picking up a phone call out of caution that a malevolent program is on the other end. As generative AI permeates daily life, will humans be able to trust who’s on the other end of digital communication? This could be the case for online discussions, video clips on the evening news, or students’ term papers.

As you know, there is an ongoing debate between prominent scientists, (personified as a debate between Elon Musk and Mark Zuckerberg,) about whether advanced AI has the future potential to pose a danger to humanity. What is your position about this?

We need to be vigilant in using AI technology. I’m not concerned about the singularity where technology controls human life. Advanced AI is a tool with improved capabilities, but humans will always be in control. However, I am concerned about how humans will leverage advanced AI in interactions with each other in the future. People with access to great AI powers could potentially pose a danger to each other. As mentioned above, these generative AI tools could threaten to erode public trust if not used properly and ethically and could hinder interpersonal communication. If one is unfamiliar with this ethical concern, I would recommend looking at the AI Bill of Rights (or its summary) to develop awareness about the challenges we face in an AI-powered digital society.

What can be done to prevent such concerns from materializing? And what can be done to assure the public that there is nothing to be concerned about?

It’s impossible to stop the wheel of innovation, so strict regulation of technology development will be fruitless. Instead, we just need to increase awareness among the public as the first step., after which we can develop solutions collectively as the technology progresses. I don’t expect a single silver-bullet solution since the impact is so broad. The trust gap is the one that I am most concerned about.

I don’t have a crystal ball to see what the future will hold, but I am interested in finding similar lessons from history. For example, it’s inspiring how people in post-Gutenberg Europe (after he introduced the printing press) coped with an explosion of information and opinion, and how it led to societal changes. I am hopeful that better technology and collective wisdom will help address these concerns. Learning and embracing technologies is the best way for people to understand their capabilities.

How have you used your success to bring goodness to the world? Can you share a story?

Many government agencies use Hyperscience’s AI/ML-based product to automate manual processes, which helps speed them up and promptly deliver much-needed services to citizens. One partnership we’re incredibly proud of is supporting the International Rescue Committee (IRC) to apply ML to its data collection in health clinics treating malnourished children.

Globally, 50 million children are acutely malnourished at any given time, with only 20% receiving treatment. To combat the complex and costly processes that limited how many malnourished children could receive treatment, the IRC implemented a simpler system that could scale to help treat more children. IRC teams use Hyperscience’s human-centered automation technology to capture handwritten data that can be photographed and digitized to automate information extraction. This process accelerates the speed at which critical information on patient outcomes can be analyzed, allowing IRC teams to speed up treatment and help more children.

Instances like these are countless and showcase the real-world impacts AI technology can have.

As you know, there are not that many women in your industry. Can you advise what is needed to engage more women into the AI industry?

We must start at the top of the funnel as a society to increase young women’s educational opportunities in STEM. The harsh reality is that roughly 20% of US computer science undergraduate students are women — a number far too low. It was not always like that, as many computing pioneers were women, and the number of women studying computer science grew faster than men for decades until the 1980s.

To get more women involved in AI and other technological fields, we need to get more young women interested in STEM and admitted to computer science programs at universities. It could be achieved by creating a more welcoming learning environment and providing more resources for young women to participate in STEM. Organizations such as Girls Who Code and Women in Machine Learning do great work to build interest in the field. Some computer science programs, such as the track at Carnegie Mellon University, set a good example of gender diversity by admitting roughly equal numbers of young women and men in the first-year class.

What is your favorite “Life Lesson Quote”? Can you share a story of how that had relevance to your own life?

I like Albert Einstein’s mantra that “insanity is doing the same thing over and over again and expecting a different result.” This mantra helped me jump out of my comfort zone — I wasn’t incredibly passionate about my work after a few years at a research lab writing papers, and I decided to take the risk rather than expecting things would simply work out in the long run. If you’re doing something that isn’t working as well as you like it to be, don’t just keep doing it: investigate the issue, devise a plan to adjust and take action. Inertia is a very dangerous enemy!

You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people, what would that be? You never know what your idea can trigger. :-)

Education is still near and dear to my heart. Though I have somewhat failed my childhood dream of an academic career at a university, I would love to return to teaching one day. I’m passionate about helping young people get involved in careers or initiatives they’re passionate about.

The unfortunate reality today is that education is still hugely underleveraged to unlock the potential of human talents and connect people to more economic opportunities. I’m excited about the potential AI has to help address this. We saw several waves of new online education platforms started after the popularity of the Internet and Web paradigm. I am hoping for another new wave resulting from the advancement in AI. My current AI projects will help boost the productivity of knowledge workers in the business world. AI could provide similar productivity benefits for both students and teachers. We may see a truly personalized curriculum and delivery customized to meet every student’s unique need.

I’m not sure exactly what that will look like, but I can’t wait to find out!

How can our readers further follow your work online?

We post tons of great content on the Hyperscience blog, which shows our latest innovations and some unique use cases of intelligent document processing. Feel free to check it out at https://hyperscience.com/tech-blog/.

Thank you so much for joining us. This was very inspirational, and we wish you continued success in your important work.

About The Interviewer: David Leichner is a veteran of the Israeli high-tech industry with significant experience in the areas of cyber and security, enterprise software and communications. At Cybellum, a leading provider of Product Security Lifecycle Management, David is responsible for creating and executing the marketing strategy and managing the global marketing team that forms the foundation for Cybellum’s product and market penetration. Prior to Cybellum, David was CMO at SQream and VP Sales and Marketing at endpoint protection vendor, Cynet. David is a member of the Board of Trustees of the Jerusalem Technology College. He holds a BA in Information Systems Management and an MBA in International Business from the City University of New York.