The Future of Everything: Kathryn Hume

Jan 26, 2018

8

min read

K

athryn is a Venture Partner at ff Venture Capital. She serves as an advisor to numerous ffVC portfolio companies and brings extensive expertise in applying artificial intelligence and data analytics to a wide variety of business models and industries, including Saas software, e-commerce and social media, cybersecurity, and big data. Kathryn also serves as Vice President, Product & Strategy for integrate.ai, a SaaS platform company that enables B2C enterprises to apply artificial intelligence to a unique combination of social media, behavioral, and first-party transaction data.

Previously, Kathryn worked at Fast Forward Labs advising Fortune 500 enterprises on accelerating artificial intelligence, data science, and machine intelligence capabilities to solve real-world business problems. Prior to that, Kathryn was a Principal Consultant in Intapp’s Risk Practice, advising on issues of privacy, cybersecurity, and regulatory compliance. She also served as a Visiting Professor on technology and law at the University of Calgary Faculty of Law. Kathryn is a frequent writer and lecturer on applied artificial intelligence and data science at industry conferences and universities — including Harvard Business School, Michigan State University Law School, and the MIT Media Lab — and is a respected thought leader on issues involving the intersections of technology and society. Kathryn received a PhD in comparative literature from Stanford University and a BA in mathematics from the University of Chicago. She speaks seven languages and resides in Toronto.

1. How did you get started in the wonderful world of VC?

As with many of my achievements, my path into VC was more serendipitous than intentional: I didn’t set a goal and develop plans to achieve it, I embraced an opportunity that presented itself to me, the surprising, emergent result of work I’d done in another context.

My path into VC germinated from my path into machine learning. I did a PhD in comparative literature at Stanford, specializing in the history of mathematics, philosophy, and literature in 17th- and 18th-century France. After deciding to leave academia, I began my industry career at Intapp, a legaltech software company in Palo Alto. I’m a firm believer in micro-optimization, in excelling in a local environment as opposed to working on 5- or 10-year horizons.

At Intapp, I dug in deeply to the subject matter at hand, which was applying software to help law firms comply with professional responsibility requirements. I went deep, and became regarded highly enough in the space to get a job as an adjunct law school professor at the University of Calgary.

But I was restless. The subject matter didn’t excite me to my bones. I eventually stepped back and asked myself what I was truly interested in. The answer I came up with was machine learning: I wanted to understand how systems could use linear algebra and statistics to mimic complex human capabilities like making meaning or perceiving beauty. This led to a job leading business development for an amazing AI research lab called Fast Forward Labs (now part of Cloudera), founded by data science pioneer Hilary Mason. She took a chance on me; I am eternally grateful to her. While at Fast Forward Labs, I gave a talk at the Future Labs AI Summit, sponsored by ffVC. The partners were impressed by my talk and knowledge of enterprise AI, and offered me a role as a venture partner. I said yes in a heartbeat.

2. What is your methodology for defining and selecting the companies and entrepreneurs who are going to create the future?

I’ve been following the methodology of the partners at ffVC, primarily learning from founding partner John Frankel. There are a few parameters to evaluating companies. The first is to appreciate tectonic forces at work in the balance between hardware, software, and computing power, which leads to long-term seesaw trends pushing towards and away from centralization, towards and away from platforms or vertical focus. The second is to evaluate the business model, to see if the company’s approach to applying a new capability has the scale and focus to be valuable to a market with a problem. And naturally there’s the team, which needs to have both gumption and charity to build a company. It’s amazing how frequently personalities ruin companies.

3. From an investors standpoint, being the one who gets to invest in the “future of everything”, what can we expect to see in a 5–10–15 year timeline?

As I mentioned above, I’m skeptical about long-term timelines. Our predictions are normally linear. We focus on one class of technology and overfit our predictions to our recent past. Our imaginations are profoundly limited.

In 5-, 10- and 15-years, AI will have had serious impacts in products researchers today have not anticipated, because observant entrepreneurs focused on value will pivot applications to satisfy unanticipated market potential. I’ll give you an example of how this shifts play out.

When researchers at Northwestern first studied natural language generation (NLG), software that automatically writes sentences and communicates insights from data, they thought the killer app would be automated journalism, things like earnings reports, weather reports, or sports scores. NLG vendors ended up pivoting to focus on narrative business intelligence, partnering with data visualization vendors like Qlik and Tableau, to present findings in company transactional data to non-technical business stakeholders.

The researchers didn’t have that in mind when they first built the capability. We see this time and again. That said, machine learning really is taking off, and it will impact almost everything. The ability to automate processes and tasks that are fuzzier, that can’t be clearly described with if-then rules, is very powerful.

4. What themes, trends, or sectors are you primarily interested in and why?

I focus on machine learning, which I find interesting for countless reasons. The big revolution underway with deep learning is enabled by the fact that deep neural networks can abstract out hierarchical, general features in data sets that other statistical techniques fail to capture: the math can approximate general functions as opposed to being constrained by certain types of functions. In practice, that unlocks the ability to use computers on data that has historically been intractable, on images, video, text, literature, art, speech.

Most progress in data science over the past 15 years has been on structured transaction data, widgets sold, actions taken as proxies for interest or personality. Things get much more interesting (and potentially disturbing) when we can analyze what was said in a conversation, not just how long a conversation lasted. Forcing the world into the parameters of classification, moreover, is bringing profound social questions to the fore. We have some big questions to resolve in the next few years.

5. What moonshot are you most excited to see solved?

Quantum computing. I find the premise to be extraordinary: instead of fitting the mathematical properties of a probability distribution to a series of digits, we observe the relationship between entangled q-bits as an analogue for mathematical meaning.

Currently these systems are too delicate to perform at scale — I’m not up to date on how many q-bits they can get to perform reliably these days, but I think the numbers are still small, like 4 q-bit systems which enable us to observe 2⁴, or 16, combinations. But the growth here is exponential, like the old Paal Paysam allegory with grains of rice. Imagine what we could build if we could observe a complex probability distribution within one operation?

6. What problem or challenge is under-funded and why? What sector needs more attention?

Good question. I’d say education technology. The world has many urgent problems that affect basic human rights and our ability to live on Earth: energy, water, global warming, inequality. It is a travesty that place of birth has the effect that it does on quality of life. The difficulty in tackling educational inequality in a country like the US is that it’s a thorny problem fundamentally addressed by policy, not just technology. The sector is conservative and slow to change. It breaks my heart to think about the compound impact access to better technology will have on income and social inequality in the future.

7. What is the role of venture capital in creating the future? What responsibility does VC have to help improve the lives of others, and ultimately the world?

VC is at the top of the pyramid of value creation in the technology world. People use products; startups create the products people use; VCs fund the startups that create the products people use; LPs supply the funds that the VCs invest into the startups that create the products people use. It is our ethical imperative to enable the help shape and enable the world others want to live in. And to do that, we need to actively embrace diversity and inclusion, to broaden our perspective on what constitutes value. I hope to play my small part in doing that, if only by serving as a role model to inspire other women to work in the space.

8. What company or investment in your portfolio are you most excited about and why?

I’ve been working closely with the team at Mt. Cleverest, part of the NYU AI Nexus Lab. I love the team: they are smart, hard-working, humble, inquisitive, and motivated to solve a problem they care about. They are applying sophisticated natural language processing algorithms to automatically generate quizzes from any content on the web, and have cool features on their product roadmap to automate grading of free-form essay responses. This enables teachers and students to engage with the internet in a way that solidifies our ability to turn information into knowledge.