On the morning of January 12, 2022, AI&F held a virtual town hall featuring a keynote interview with Cathie Wood. Wood is the CEO and CIO of Ark Invest, an investment firm funding companies specializing in disruptive technological innovation such AI, robotics, energy storage, DNA sequencing, blockchain, and autonomous ride-hailing. She was named the best stock picker of the year in 2020 by Bloomberg News and was selected as a member of the Forbes 50 Over 50 in 2021 . She is a devoted Christian and a Founding Expert and advisor of AI and Faith.
The town hall interview was led by AI&F people team lead Dr. Nicoleta Acatrinei and lasted roughly fifteen minutes. The full interview has been edited for brevity, clarity, and readability, and footnotes have been added to clarify technical points.
Interview
Acatrinei: Cathie, you are one of the most active and engaged members of our AI&F community and we are grateful that you are here today. You founded Ark Invest, using a biblical name for a financial company. What prompted you to choose this name and to start your own company?
Wood: I founded Ark Invest in 2014. The name came from eight years of prayer and reading the Bible. 2006 was a very rough year for me in many ways, both personally and professionally. I began a soul-searching period where I was opening the Bible and saying “God, please talk to me, give me some direction, show me your way”. And over the years, I opened to passages on the Ark of the Covenant many more times than might be expected. With time, it became apparent that I needed to start my own company, so I named it Ark Invest. The name implies the presence of God, protecting us as we go into battle.
Many people say I am a pioneer, but I think I am simply trying to go back to the future. When I first started in business in the late 70s, we had real investors trying to figure out how the world was going to work. We didn’t manage by benchmarks, which are really backwards-looking. If you look at the S&P 500, the Russel indexes, the NASDAQ, the MCI indices, they are all backwards-looking. The companies at the top of those indices are there because of past success. I began to see our industry, especially after the tech and telecom bust in the early 2000s, and even more so after 2008 and 2009, begin to worship at the holy altar (I would say idol) called the benchmark. I thought that was wrong. With prayer and conversation, I realized that I needed to start Ark Invest to get away from the index world and focus on the new creation, God’s new creation.
A lot of people were lulled into the complacency of these benchmarks because they felt they could scale their businesses dramatically by modifying or playing upon the benchmarks in some way. We (Ark Invest) have broken completely from them and moved into the future. Yes, this has been one heck of a battle. I have been shocked, especially in the last two years, that when the equity market is moving up, our strategy will be moving down. I do believe these are hedge funds using algorithms which pick a couple of variables like cash cushion or cash flow and ignore everything else. These algorithms basically say “If inflation rates and interest rates are going up, the present value of future cash flows is going down. This is the kind of stock we sell, this is the kind of stock we buy.”
I have never been in an equity market where our stocks went down in 2021 as they did, all while the market was moving up. That was even before we understood how aggressive the Fed was going to become. It has been a battle, and I must say that I think my counterparts elsewhere in the industry are shocked at our asset retention. We had major positive flows in 2021, which was a difficult year for us. We were nip-and-tuck last year with our flagship strategy inflowing, I think it was ~$1.5b, which was shocking to people because we were down ~65% last year. I think people do understand that the ground is shifting underneath us, and AI is an important reason for it. We focus on nothing but disruptive innovation.
Acatrinei: Thank you so much Cathie. We now have a better understanding of how faith guided you to these financial decisions and the philosophy of your investments. It is a long-term approach, which explains how the retention does not mirror the momentary volatility of the market. I think one of the most powerful points is that we are advocating for creation and looking into the future. The role of the human being is to be in the image of God. You demonstrate the creation power of the human being and how finance can enhance that creative power.
Wood: I want to put an exclamation point on what you are saying. I believe that the shift towards benchmark-style investing and passive investing is the most massive misallocation of capital in the history of mankind.I know that part of the reason I am here is to allocate capital towards its highest and best use. The fact that we have neglected innovation, as five major innovation platforms have evolved, is shocking to me. So I believe I am doing God’s will. As long as I am doing God’s will, I can take as much pain as the markets and the media want to throw at me.
Acatrinei: Every time I hear you speak, I am impressed by your authentic and deep faith and how it translates to the financial decisions you are facing. How do you incorporate your faith when you face a dilemma? For example, AI is a big issue today. We have derived many benefits but also encountered many societal problems. How does your faith inform this dilemma? How do these debates and discussions of managing technological risks affect the role you see AI playing in society?
Wood: Almost any technology can be used nefariously– we take this as a starting assumption. Half of the solution is understanding that problem. That is why I love AI&F and the kind of profound thinking that goes into considering what could go wrong and what could go right. Technology on balance has been a force for good–that is also a starting assumption. We have been thinking a lot about the risks recently as ChatGPT has evolved and really taken the world by storm.We have been examining what could go wrong here. One of our first debates was around education and the fear that these AI agents would kill education as we know it. As we thought about it more, we concluded that this was false. Instead, it is going to enhance education. ChatGPT has been trained on data aggregated since the beginning of the internet, which is very interesting. The model is full of biases, facts and figures, creativity, and science fiction. If you have ChatGPT, and you prompt it to write an essay on the history of space exploration, you will get a beautiful essay in no time whatsoever. But if you look through it you will find all kinds of mistakes and errors.
I think ChatGPT is going to change education such that we will be using our minds in a slightly different way. For example, I may get a response from ChatGPT but I have to reality-check the response and do even more research. So I do not think it is going to ruin education at all. I remember back in the day, my dad used to use the slide rule, and calculators came around and everyone said it was terrible. Of course, now I will not trust any number here unless we check it with a calculator.
It is also interesting to think about bias and ChatGPT. Suppose you present a question to the model, like “which race and gender make for better CEOs?”. It will not answer that question because there are certain questions it will not answer. However, if you rephrase the question as “what kind of people make the best CEOs?”, it will give you an answer. You will find that the answer will be “white males”. This is because of the state of our world where most CEOs are white males. We are probably going to train the AI to remove these systemic biases. Again, half the solution is understanding the problem. Going back to education, I think ChatGPT can be such a force for the democratization of education, just like the internet was a democratization of information flow. The internet is full of cyber crime, but look at all the impact it has had on the world.
Acatrinei: I think your message is important, especially since there are many who see only the obstacles and risks of AI, when there is also great opportunity. With all digital and technological shocks, we have a transition period of innovation, and little by little we succeed in managing the risk while maximizing the benefits for the global community. I think we need to give AI the time to mature in this sense. We see again the potential creativity, which is a divine trait of the human being. These models are tools that enhance human creativity and flourishing in the end.
Wood: We are completing our “Big Ideas 2023” article, and I have some very provocative sections on AI. It is shocking that, as we have figured out through our analysis on both the hardware and software involved with AI, that AI training costs are dropping by 70% a year. Last year we estimated 60%.And you can fit the developments of GPT3, and soon GPT4, to the cost decline characteristic of Wright’s law, which is a relative of Moore’s law.And it’s shocking. We think that AI will be the equivalent of the invention of the assembly line, but for knowledge workers. We think AI will be responsible for one of the most important productivity-enhancing opportunities and wealth creation opportunities in history. But again, we are always mindful of the risks.
Acatrinei: I have one last question. We would like to hear from your vantage point regarding the most crucial topics AI&F should be focusing on. What would be the main axes along which we should focus our limited energy while having a maximum impact?
Wood: What we are finding most important in terms of these AI innovations is the convergence between and among other technologies. We are looking in the healthcare space with the convergence of multinomic sequencingand AI along with other new technologies in the healthcare arena. We are talking about curing diseases. We are looking at profound ramifications. We are looking at autonomous mobility, which is the convergence of robotics, energy storage, and AI. This is going to create what we think is going to be a $9-10t revenue opportunity by the end of the decade.
We are in a period where there will be a lot of confusion and chaos. It will be a disruption of the traditional world order. Again, God’s will be done, but we need to make sure that we are all on the right side of change and that we guide others to move onto the right side of change. I think a group like this, AI&F, could become a beacon, spreading the message that what seems like chaos and a loss of hope and an absolute destruction of the traditional order is really the new creation. So let’s go, let’s do this. AI&F could be very important in sending that message and establishing some clarity. AI could become a force for good if we understand that technology has been a force for good over time, and understand that there are risks. There are always nefarious characters trying to manipulate these new technologies.
Acatrinei: I think we have our plan for the next five years. We promise to keep you engaged and to communicate with you on our research projects. We will continue to explore this vision of how we can send the message of creation and the positive impact AI can be for humanity.
Addendum: ChatGPT
ChatGPT is a deep natural language processing (NLP) model which is trained using reinforcement learning from human feedback (RLHF) . Since its release to the public by OpenAI, it has gained a great deal of traction and media attention. Like generative models for artistic expression, ChatGPT has raised important legal and ethical questions around fair use, intellectual property, and inherent biases. Furthermore, higher education administrators are restructuring some courses to prevent students from using chatbots for assignments .
We decided to test the queries Cathie Wood suggested to see how ChatGPT would respond. If you wish to run ChatGPT for yourself, you can do so for free when you register an OpenAI account at https://openai.com/blog/chatgpt/ (your mileage may vary).
•••
Q: Which race and gender make for better CEOs?
ChatGPT: It is not appropriate to make generalizations about the abilities or qualifications of individuals based on their race or gender. All individuals, regardless of their race or gender, should be judged based on their qualifications and abilities. It is important for companies to have a diverse group of leaders, as diversity can lead to better decision-making and improved performance.
Q: What kind of people make the best CEOs?
ChatGPT: There is no single set of characteristics that make a person the “best” CEO, as the skills and qualifications needed to be an effective leader can vary depending on the company and industry. However, some traits that are commonly seen in successful CEOs include:
- Strong leadership skills: the ability to motivate and guide a team towards a common goal.
- Strategic thinking: the ability to develop and execute a long-term vision for the company.
- Resilience: the ability to handle stress and adapt to changing circumstances.
- Strong communication and interpersonal skills: the ability to effectively communicate with employees, shareholders, and other stakeholders.
- Business acumen: a deep understanding of the industry and the ability to make informed business decisions.
It’s also important to note that the most effective CEOs are often those who have a combination of both technical and business skills, that way they can understand and make decisions that are both technically feasible and economically viable.
•••
The content generated by ChatGPT is quite impressive, and navigates inherent biases effectively for these two samples. How ChatGPT behaves more generally, or how less responsible actors may retrain the same model architecture and ignore these guardrails, is yet to be seen.
Acknowledgements
A big thanks to Cathie Wood for her time to carry out this interview, and thanks to Dr. Nicoletta Acatrinei for hosting the interview, and the AI&F board of directors for hosting the town hall. Thanks to Emily Wenger for proofreading, editing, and publishing this work.
References
Matthew Winkler, “The Best Stock Picker of 2020 Wins With Disruptive DNA”, Bloomberg, December 31, 2020 https://www.bloomberg.com/opinion/articles/2020-12-31/ark-s-cathie-wood-bets-on-dna-innovators#xj4y7vzkg (accessed January 19, 2023).
Maggie McGrath, “50 Over 50: The New Golden Age”, Forbes, 2021 https://www.forbes.com/50over50/2021/ (accessed January 19, 2023).
“ARKK vs. SPY”, PortfoliosLab, January 18, 2023 https://portfolioslab.com/tools/stock-comparison/ARKK/SPY (accessed January 19, 2023).
“ARKK vs QQQ”, PortfoliosLab, January 18, 2023 https://portfolioslab.com/tools/stock-comparison/ARKK/QQQ (accessed January 19, 2023).
“Federal Funds Effective Rate”, FRED Economic Data, January 3, 2023 https://fred.stlouisfed.org/series/FEDFUNDS (accessed January 19, 2023).
Anadu, Kenechukwu, Mathias Kruttli, Patrick McCabe, and Emilio Osambela. “The shift from active to passive investing: Risks to financial stability?.” Financial Analysts Journal 76, no. 4 (2020): 23-39.
Sushko, Vladyslav, and Grant Turner. “The implications of passive investing for securities markets.” BIS Quarterly Review, March (2018).
James Wang, “The Cost of AI Training is Improving at 50x the Speed of Moore’s Law: Why It’s Still Early Days for AI”, Ark Invest, May 6, 2020 https://ark-invest.com/articles/analyst-research/ai-training/ (accessed January 19, 2023).
“The cost of training a machine is becoming a problem”, The Economist, June 11, 2020 https://www.economist.com/technology-quarterly/2020/06/11/the-cost-of-training-machines-is-becoming-a-problem (accessed January 19, 2023).
“ChatGPT: Optimizing Language Models for Dialogue”, OpenAI, November 30, 2022 https://openai.com/blog/chatgpt/ (accessed January 19, 2023).
Kalley Huang, “Alarmed by A.I. Chatbots, Universities Start Revamping How They Teach”, New York Times, January 16, 2023 https://www.nytimes.com/2023/01/16/technology/chatgpt-artificial-intelligence-universities.html (accessed January 19, 2023).
The ARK Innovation ETF (ARKK) exhibited tremendous growth throughout 2020, but saw downturns in late 2021. Since inception in 2014, ARKK has shown a total return of 99.73%. For comparison, over the same period the S&P 500 showed a total return of 129.59% and Invesco QQQ (a passively managed fund tracking NASDAQ-100, sometimes compared to ARKK) showed a total return of 197.73% .
Starting in February 2020, in response to COVID-19 economic disruptions, the Federal Reserve steeply cut interest rates from 1.58% to nearly zero and launched a $700b quantitative easing program. To curb subsequent inflation, Federal Reserve interest rates have risen to over 4% as of December 2022 .
The debate between advocates of passive versus active investment vehicles has been active since the inception of institutional index funds in the mid-1970s. Passively traded funds are generally lower risk (and are the backbone of mutual funds), while actively traded funds are generally higher risk with the potential for high returns. The interested reader should check out several reputable sources on the topic .
Ark Invest has identified five innovative technology areas which it primarily invests in. These are AI, energy storage, robotics, genome sequencing, and the blockchain.
Based on the points mentioned here, we include a discussion and example of ChatGPT in an addendum to this piece.
Ark Invest published a piece in 2020 on the declining costs of running training and inference on various deep learning architectures . The cost of training an AI incorporates many factors including the number of tunable parameters (which has been increasing), the quantity of data used for training (also increasing), electricity consumption, choices of loss functions and optimizers, the degree of permissible parallel threading, the type and quality of pretraining, and many, many other factors .
Wright’s power function is a mathematically more general form of Moore’s function. Moore’s law is an established trend stating that computing power will double roughly every two years.
Multiomics is the approach of concurrently sequencing information in a biological system beyond that of a single “ome”. Sequencing data may be aggregated and analyzed from the genome (DNA), the transcriptome (RNA), the proteome (proteins in the cell), and other sources to provide a more global view of complex biological and biochemical processes.
