Harvard Data Science Review Podcast Summary

Title: What Does AI Buy Us or Cost Us? Views From the Financial Industry
Host/Author: Harvard Data Science Review
Release Date: February 23, 2024

Introduction

In the February 23, 2024 episode of the Harvard Data Science Review Podcast, hosts Liberty Vittert and Shelley Mack explore the evolving landscape of artificial intelligence (AI) within the financial industry. Featuring insights from two esteemed guests—Christina Chi, CEO of Data Bento, and Victor Lo, Senior Vice President of Data Science and Artificial Intelligence at Fidelity Investments—the episode delves into how AI is reshaping investment strategies, the challenges it presents, and the ethical considerations that accompany its integration.

Guest Introductions

[00:03] Liberty Vittert opens the episode by highlighting the rich history of data usage in finance, from hedge funds employing regression analysis to fintech startups leveraging summary statistics. She introduces Christina Chi and Victor Lo as key voices in discussing the impact of AI on the financial sector.

[00:58] Shelley Mack commends Christina's remarkable journey from a rejected intern to a Forbes 30 Under 30 founder, setting the stage for an in-depth discussion on current projects and industry trends.

Current Projects and Trends in AI and Finance

[01:28] Christina Chi provides an overview of her role as co-founder and CEO of Data Bento, a market data provider serving finance, fintech, and other sectors. She emphasizes the company's role as the "data backbone" for AI applications, noting that many AI tools used in finance are trained on Data Bento's datasets.

Christina reflects on her decade-long experience running a hedge fund, observing the cyclical nature of trends like AI, cryptocurrency, NFTs, meme stocks, ESG (Environmental, Social, and Governance) criteria, and alternative data. She points out that while some trends endure, others fade, underscoring the dynamic nature of the financial industry.

Data Science Skills and the Rise of Causal AI

[02:43] Shelley Mack turns to Victor Lo, appreciating his article on the "10 Challenges of Data Science from an Industry Perspective." She seeks his insights on the rapid advancements in AI, particularly regarding the skill sets required for data scientists navigating these changes.

[03:25] Victor Lo responds by differentiating between generative AI (e.g., large language models like ChatGPT) and the broader field of data science, which encompasses both soft and hard skills. He outlines the data science pyramid:

• Descriptive Analytics: Understanding past events.
• Predictive Analytics: Forecasting future events.
• Prescriptive Analytics: Guiding future actions based on predictions.

Victor introduces Causal AI as an emerging field focused on understanding and optimizing the relationship between actions and outcomes, highlighting its potential to enhance decision-making in areas like marketing through experimental design and causal inference.

AI in Investment: Current Usage and Future Directions

[09:34] Shelley Mack prompts Christina and Victor to discuss the practical applications of AI in investment, both for individual investors and large institutions, and to envision the future trajectory of AI in finance.

[09:34] Christina Chi shares her observations from a recent financial conference, noting that while many believe AI will revolutionize finance, full-scale adoption has lagged. She explains that unlike closed systems like chess or Go, financial markets are highly dynamic and influenced by ever-changing rules and irrational behaviors. This complexity makes it challenging for AI to consistently outperform human strategies. Christina recounts her experience with high-frequency trading algorithms that required constant tweaking, as market conditions evolved and competitors adapted, demonstrating the limitations of AI in maintaining long-term superiority.

Challenges and Lessons in AI Adoption in Finance

[13:26] Shelley Mack references a quote by Jeff Bezos, emphasizing the tension between data-driven decisions and anecdotal evidence. Christina elaborates on the pitfalls of relying solely on historical data through stories of hedge funds that failed despite employing top researchers and advanced algorithms. She highlights issues like data overfitting and the discrepancy between simulated success (backtesting) and real-world performance, reinforcing the idea that reality often diverges from theoretical models.

[16:38] Victor Lo echoes Christina's sentiments, drawing parallels with academic simulation studies that rarely report failures. He underscores the importance of recognizing that "there's no free lunch," acknowledging that AI systems, like humans, can make significant mistakes with far-reaching consequences.

AI Risks and Ethical Considerations

[17:50] Victor Lo transitions the conversation to the broader risks associated with AI, such as disinformation through deep fakes and algorithmic discrimination leading to fairness and bias issues. He cites how large language models trained on imperfect historical data can perpetuate biases, such as gender or racial biases, inadvertently affecting decision-making processes.

[19:23] Victor Lo discusses industry best practices and governmental frameworks aimed at mitigating these risks. He references the White House Blueprint for an AI Bill of Rights (2022), which emphasizes safety, effectiveness, fairness, and transparency in AI systems. Victor highlights the challenges of explaining AI decisions, especially with "black-box" models, and the ongoing research to improve fairness testing and bias mitigation.

Cultural Impact on AI Ethics and Regulations

[27:08] Victor Lo addresses the influence of cultural differences on AI ethics and regulations. He observes that while countries may prioritize aspects like data privacy differently, many global regulations converge on key principles such as data privacy, intellectual property, and algorithmic discrimination. The primary variation lies in balancing AI safety with innovation, with each country determining its approach based on cultural and economic priorities.

[28:02] Christina Chi concurs, noting that in finance, regulatory standards like those from the Securities and Exchange Commission (SEC) and the CFA Institute's Ethics Handbook provide a relatively standardized ethical framework across different jurisdictions. However, she acknowledges the gray areas and varying interpretations that can lead to ethical dilemmas and legal challenges, especially as AI becomes more integrated into financial practices.

Personal Lessons and Advice from Guests

[31:56] Christina Chi shares a personal lesson emphasizing the importance of sometimes disregarding statistical probabilities in favor of gut instincts. Reflecting on her early career, she explains how dismissing discouraging data allowed her to pursue ventures that, despite eventual failures, provided valuable experiences and unique opportunities. Christina advises ambitious individuals to follow their passions even when data suggests otherwise, highlighting the intrinsic value of personal drive and intuition.

[35:28] Victor Lo complements Christina's advice by stressing the necessity of continuous learning. He advocates for a multidisciplinary approach, integrating technical skills with knowledge from fields like philosophy, law, and social sciences. Victor encourages data scientists to remain perpetual students, adapting to the evolving landscape of AI and its applications.

Magic Wand Question and Concluding Thoughts

In the finale, host Shelley Mack poses a "magic wand" question to Christina and Victor.

[36:56] Christina Chi wishes for the investor market to become more accessible and affordable to everyone, not just elite or well-funded entities. She underscores the importance of democratizing data and opportunities in finance, coupled with robust education on the associated risks.

[37:58] Victor Lo echoes the significance of education, advocating for more efficient and comprehensive learning methods to equip individuals with the necessary skills to navigate the complexities of AI and finance.

Both guests agree that addressing the challenges and ethical considerations of AI requires a balanced approach of innovation, regulation, and education to harness the benefits while mitigating the risks.

[38:16] Victor Lo concludes by reinforcing the need for proper education and informed decision-making as AI continues to influence the financial sector.

[38:50] Shelley Mack and the hosts express gratitude to Christina and Victor for their insightful contributions, wrapping up the episode with a call to stay informed and cautious in the face of AI advancements.

Notable Quotes

• Christina Chi [09:34]:
  "It's almost like sometimes in life, it's like you have to just listen to your gut instinct on, like, what wakes you up in the morning. And that might not always correspond with what the data wants you to do, but you just have to do it, and that's okay."

• Victor Lo [27:08]:
  "AI ethics involves both technical approaches like fairness testing and philosophical approaches like utilitarianism and deontology, requiring expertise from legal, risk compliance, and ethical domains."

• Christina Chi [31:56]:
  "Sometimes you have to just ignore the data and follow your gut instinct, because the human is still the most complicated machine."

Conclusion

This episode of the Harvard Data Science Review Podcast offers a comprehensive exploration of AI's benefits and challenges within the financial industry. Through candid discussions, Christina Chi and Victor Lo illuminate the nuanced interplay between technological advancements, ethical considerations, and the human element in finance. The conversation underscores the imperative for continuous learning, ethical vigilance, and embracing both data-driven insights and human intuition to navigate the future of AI in finance.