Dave Harrison Smith, CFA, Executive Vice President, Domestic Equities, and Head of Technology Research delves into—both massive and more mundane—potential advancements, impacts, and risks as generative AI leaps forward.

 

“ChatGPT just saved me,” my wife exclaimed earlier this week. As a legal manager she is often tasked with communicating complex issues, and in the past has found that she too often can come off as brusque and overly direct. She found herself spending hours wordsmithing memo introductions to strike a warmer tone while still maintaining a professional manner. This week, she dropped a draft introduction into the prompt box for the generative AI superstar ChatGPT and asked it to soften the tone and spin up a novel, friendlier introduction. Thirty seconds later her memo was complete, and she was thrilled.

While the Artificial Intelligence (AI) field has been under development since the 1950s, over the past year we have witnessed a monumental leap forward in a sub-branch of AI called generative AI. Specifically, the private company OpenAI released a revolutionary interface, ChatGPT, which comprised a powerful large language model (LLM) and a modern chat-based user interface. The readily accessible, easy to use interface—combined with the power of a learning model with hundreds of billions of parameters—had one AI expert saying it was akin to “looking into the face of God.” What has followed has been nothing short of remarkable: ChatGPT became the fastest consumer product to reach 100 million users (in a mere two months), academic scholars and open-source developers have made incredible strides with almost daily breakthroughs in generative AI research, and massive, trillion-dollar companies and well-known start-ups have rapidly pivoted strategies to embrace the potential of generative AI.

ChatGPT saving my wife a half hour of work writing an email introduction may not sound groundbreaking, but it’s our belief that generative AI has the capacity to make a powerful difference in the world. The average individual’s perception of economic progress is unfortunately frequently tied to the business cycle or stock market returns. Ask a person on the street how humanity has done over the past 40 years and the answer may well depend on the state of their local economy. Economists would argue that, in the long run, prosperity is tied directly to productivity growth. An increase in the average output per hour of labor worked means that the economy is growing, with our per capita income similarly increasing. Unfortunately, by many measures, productivity growth has stalled since the early 2000s. Disappointed economists have offered a myriad of reasons for the slowdown, including demographics and the lack of life-changing inventions, like the internal combustion engine, but the growth stagnation has continued unabated.

Enter Generative AI
Through our research, we have come to believe generative AI technology is key to unlocking a step function increase in productivity growth and, longer term, our quality of life. News and social media feeds alike are already flooded with peers using generative AI to save immense amounts of time. Professors are generating the framework of a new paper in hours rather than days. Programmers are debugging code in minutes rather than hours. Marketers are creating content drafts in seconds rather than minutes or hours.

These productivity gains are showing in academic studies, as well. Scholars Brynjolfsson, Li, and Raymond found that access to a generative AI tool increased call center worker productivity by 14% and by 30% for lower skilled and newer employees.1 Kazemitabaar et al found that the productivity of programmers was increased by 1.8x when they were paired with a code generating AI co-pilot.2 Incredibly, we believe we are just scratching the surface of what this technology can do, and we expect to see a Cambrian explosion of uses and products over the coming months and years. In a thought piece from its research team, Goldman Sachs stated that the “…proliferation of consumable machine learning and AI has the potential to dramatically shift the productivity paradigm across global industries, in a way similar to the broad scale adoption of internet technologies in the 1990s.”3 Like the early days of the internet, many of these products will sputter, but we believe the next crop of great technology products is poised to rise from this new ecosystem.

Acknowledging the Ghosts in the Machine
The benefits of generative AI have the potential to be profound. Yet, it is important to recognize the risks as well. The media is already awash with stories of students using ChatGPT to generate essays. While cheating should never be excused, these stories may one day seem trivial to larger threats that are far more dangerous. We are just beginning to grasp the impact of disinformation in the democratized media world of Twitter and Facebook. Imagine now, that bad actors can generate AI optimized content in text, image, and even video form that appears accurate and true. Unfriendly nations and terrorist organizations will gain a powerful new weapon when the cost of generating new falsehoods falls to zero.

Pull QuoteSimilarly, we are just beginning to understand the ‘ghosts in the machine’—the unanticipated errors and dangers—that are present in the current generation of AI platforms. At their very core, AI platforms are not trained to give correct answers. Instead (at a vastly oversimplified level), the technology is in effect statistically predicting strings of words that are likely to be paired together, based on unfathomable amounts of data. It is nothing short of magical when it works. And it is nightmarish when it fails. A prominent Washington law professor was recently the victim of a false ‘fact’ given by ChatGPT when it listed him as having been accused of sexual harassment. The ChatGPT story was complete with sources and looked identical to a factual statement. The only issue is that it was completely fabricated by the machine. While we are just beginning to understand how generative AI can help us in our daily lives, we are also just beginning to feel the pain from the other side of the double-edged sword.

Potential Effects of Productivity Change
The impact on employment is another topic of significant debate. We have argued in the past that technological shifts are not zero sum games. If call center employees are 15% more productive, that may mean a company needs to employ fewer call center agents. However, these agents are now able to find employment in other jobs and, from an aggregate standpoint, the overall economy grows. Yet, this line of reasoning can feel heartless and myopic; looking at aggregate numbers ignores the fact that these are real people with real families losing their jobs. “The productivity effects of generative AI are likely to go hand in hand with a significant disruption in the job market as many workers may see downward wage pressures,” according to the Brookings Institute.4

We are reminded of John Steinbeck’s famous work The Grapes of Wrath, where the breakthrough technology of the tractor enabled one farmer to do the work of 15, displacing tens of thousands of poor farmers and sharecroppers. The misery of these families is chronicled in detail, despite the long-term economic growth the technology enabled. It is critical that our society recognizes the pain that frictional unemployment can cause and ensures that proper investments in job training and safety nets are in place.

There is still a great deal of uncertainty surrounding generative AI. The common maxim that we are overestimating the impact in one year and underestimating the impact in ten years will likely hold true. As investors, we need to be aware of the hype surrounding the technology, as it can create bubbles of over-valuation, a la countless transformational technologies in the past. As users and as a society, we need to be aware of the risks and threats that the new technology can precipitate.

It’s our belief that this represents the next great technological evolution and, in the medium to long term, has the potential to meaningfully impact worker productivity and significantly enhance our daily lives, perhaps on par with some of the great technological leaps of the past century. We also believe there will be a significant opportunity set created for entrepreneurs and savvy investors over time. Bill Gates called generative AI the most revolutionary technology he has seen since the graphical user interface,5 which spawned the Windows operating system. Certainly, Gates knows a thing or two about generational technological shifts. We can’t wait to see what the future holds.

 

 

1 Brynjolfsson, E., Li, D., & Raymond, L. R. (2023, April 24). Generative AI at work. NBER. https://www.nber.org/papers/w31161.
2 Kazemitabaar, M., Chow, J., Ma, C. K., Ericson, B. J., Weintrop, D., & Grossman, T. (2023, April 19). Studying the effect of AI Code Generators on Supporting Novice Learners in Introductory Programming. ACM Digital Library. https://doi.org/10.1145/3544548.3580919.
3 Goldman Sachs Global Investment Research. (2016, November 14). Profiles in Innovation: Artificial Intelligence. https://www.gspublishing.com/content/research/en/reports/2019/09/04/a0d36f41-b16a-4788-9ac5-68ddbc941fa9.pdf.
4 David Kiron, E. J. A., David Autor, A. S., Sanjay Patnaik, J. K., Ajay Agrawal, J. S. G., Sukhi Gulati-Gilbert, R. S., & Nicol Turner Lee, A. K. (2023, June 29). Machines of mind: The case for an AI-powered productivity boom. Brookings. https://www.brookings.edu/articles/machines-of-mind-the-case-for-an-ai-powered-productivity-boom/.
5 Gates, B. (2023, March 21). The age of AI has begun. GatesNotes. https://www.gatesnotes.com/The-Age-of-AI-Has-Begun.