AINL#002 Augmented Intelligence in Investment Management Newsletter
Welcome to the 002 Edition of the Newsletter on Augmented Intelligence in Investment Management (AINL). Every two weeks, we deliver five unique insights tailored to empower investment decision-makers. Our insights are carefully curated by a seasoned team of market specialists. Unbiased, actionable and practical. They will help you navigate through the noise.
AINL#002 SYNTHESIS
What do these recent developments mean for investment decision-makers?
1. Applied behavioral finance is a young domain. Behavioral scientists initially focused on behavioral change of the individual. More recently, they are recognizing that in order for a person to adapt, it requires a supportive context creating a safe space and directive guidance. In such space, the individual can and should be challenged to strive for excellence (HI.3), including the exposure to high IQ talents of the whole team.
2. The Artificial Intelligence domain is not short of bold claims and predictions. Investment decision-makers need to cut through the noise, understanding which foundation models and related agents can be used for which kind of task. The evidence is increasingly clear that the more unstructured the situation, the more human control of the decision process is paramount. Agents can well be used in the exploration of the unstructured to create evidence for the decision point.
3. The necessity for a curated balance between human and machine becomes visible in the domain of sustainable investing. Structuring the unstructured through such balance improves the exploration and exploitation of opportunity sets.
TOP 5 ARTICLES
ARTICLE ONE
Evaluating AI R&D Agents Against Human Experts
ARTIFICIAL INTELLIGENCE | METR | November 2024
Important Development
How close are current AI agents to automating AI R&D? A new ML research engineering benchmark (RE-Bench) addresses this question by directly comparing frontier models with 50+ human experts on 7 challenging research engineering tasks. Many governments and (financial) companies have highlighted automation of AI R&D by AI agents as a key capability to monitor for when scaling/deploying frontier ML systems. However, existing evaluations tend to focus on short, narrow tasks and lack direct comparisons with human experts. This study closes that gap.
Why Relevant to You?
Human experts and AI agents have different strengths and weaknesses on RE-Bench tasks. AIs make more progress than humans at first, but improve their score more slowly over time. Current AI agents often struggle to respond to surprising evidence, or explore approaches beyond the most common or generic option. Systems for automated LLM research aren’t there yet, but there is real progress.
ARTICLE TWO
AI Can (Mostly) Outperform Human CEOs
HUMAN INTELLIGENCE | Harvard Business Review | November 2024
Important Findings
Generative AI has demonstrated the potential to significantly outperform human CEOs in strategic decision-making by excelling in data-driven tasks like product design and market optimization. In an experiment simulating the automotive industry, AI models outpaced human participants in market share and profitability but faltered in handling unpredictable disruptions, leading to faster dismissals by virtual boards.
Why Relevant to You?
The pattern repeats. While AI’s ability to analyze complex data sets and iterate rapidly could revolutionize corporate strategy, it lacks the intuition and foresight required to navigate black swan events. You need to master the difficult market phenomena yourself, with the support of others and the machine, but you staying in control of the decision making.
ARTICLE THREE
Measuring the MOAT
SUSTAINABLE INVESTING | Morgan Stanley | October 2024
Important Findings
A new Mauboussin Paper on quantifying the drivers of sustainable value creation. An assessment of a company’s strategy is arguably the most important part of the investment process for a long-term investor, as strategy explains how a company achieves sustainable value creation.
Why Relevant to You?
Mauboussin publications are a must-read for professonal investors. This one makes no difference. You can extract a checklist for measuring sustainable value creation when analyzing a stock and compare it with your own approach. New insights assured.
ARTICLE FOUR
Predictive Insights Through LLMs in Food Policy and Behavioral Interventions
HUMAN & ARTIFICIAL INTELLIGENCE | SUSTAINABLE INVESTING | Yale University & El-Erian Institute, Cambridge | November 2024
Important Findings
A new Cass Sunstein-led paper explores the potential of LLMs in aiding policy decisions related to food consumption and waste. The study fine-tuned GPT-3.5 on a dataset of food-related behavioral interventions to predict outcomes and a achieved 79% accuracy in predicting the direction of effects and low error rates.
Why Relevant to You?
This study shows once again that when it comes to climate action, there is no shortage of capital in the world. The question is how we choose to deploy it. It highlights the potential of LLMs to transform evidence-based policymaking by significantly reducing the time and resources needed to evaluate interventions while also identifying critical challenges to their practical adoption. Investors directly benefit from such use in their selection problem, when combining profit and impact motives.
ARTICLE FIVE
Does AI in ESG Reporting Solve Problems or Create New Ones?
HUMAN & ARTIFICIAL INTELLIGENCE | SUSTAINABLE INVESTING | Babson College | November 2024
Important Findings
Artificial intelligence is often touted as a game-changer for sustainability reporting, but is it the silver bullet many claim it to be? A recent article examined the pitfalls of AI in ESG reporting, drawing parallels to other overhyped technologies like blockchain.
Why Relevant to You?
While AI can streamline processes and uncover insights, its effectiveness depends on high-quality data and thoughtful human oversight. Furthermore, voluntary standards and regulations remain critical to ensuring accountability and preventing misuse.
