The taste of AI

As developers, we’re not just building AI models to analyze companies; we’re stepping into the shoes of active investors. While AI can process vast amounts of data, it misses the human side—the firsthand experiences we gain by touring production facilities or using a product. These insights help us assess a company in ways AI simply can’t.

Today, we’re asking: How important is it to directly experience a product, like tasting airline food, when making an investment decision? Can AI fully capture what’s felt, or is there something irreplaceable about the human touch in evaluating investments?

The starter

In the intricate world of investment, gathering comprehensive insights about potential investment targets requires more than surface-level observations. During a recent visit to an airline catering facility, I found myself standing in their catering kitchen, contemplating a fundamental question: How important is directly experiencing a product’s quality – such as tasting airline food – when making an investment decision?

The Unique Challenges of Airline Catering

The airline catering industry stands at the fascinating intersection of culinary craftsmanship and operational efficiency. Modern catering for airlines goes far beyond preparing simple meals—it’s a sophisticated operation tackling extraordinary challenges:

• Preparing meals hours in advance while ensuring they remain fresh.
• Withstanding temperature variations.
• Adhering to strict food safety standards.
• Catering to diverse dietary and cultural preferences.

For instance, designing a single in-flight meal involves balancing numerous variables, from airline specifications and passenger demographics to route characteristics and regulatory standards. Recipes must be scalable for mass production, reheatable, and compatible with the confined space of an aircraft galley. Each dish must embody an airline’s brand identity while meeting nutritional standards and delighting passengers—a logistical and culinary feat.

The limitations of culinary judgement

When evaluating investments, AI models rely on data—specifically, data that provides valuable insights about the problem at hand. This poses a challenge when it comes to sensory experiences like taste. After all, taste is a highly subjective, multi-dimensional human perception shaped by factors like:

• Chemical compound interactions.
• Individual genetic differences.
• Cultural influences.
• Personal memories.

This complexity makes taste difficult to quantify or evaluate through AI models. Yet, AI is making inroads into the culinary world in interesting ways, from personalizing recipes by substituting ingredients [1] to assisting chefs with creative recipe generation [2] or predicting flavor profiles [3].

For airline caterers, AI applications could focus on simpler yet impactful solutions. For example, analyzing food waste through camera images of returned trays can help identify unpopular dishes. However, whether such insights are accessible to investors remains a question.

Do Investors Need to Taste the Product?

So, does experiencing a product’s quality firsthand truly matter? Or, put another way, does the lack of this experience hinder sound investment decisions by AI?

While AI cannot “taste” a dish, it can evaluate the downstream effects of taste through objective, measurable indicators. Consumer satisfaction, though rooted in subjective experiences, reveals itself through concrete signals: repeat purchases, customer reviews, sales volumes, and long-term business performance.

Beyond Taste: Data-Driven Metrics for Consumer Satisfaction

We argue that tasting a product—or assessing its quality firsthand—is not a prerequisite for making an informed investment decision. Otherwise, how could a European investor, accustomed to Swiss chocolate all his life, come up with the idea of investing in the leading US chocolate producer, Hershey? The success of a chocolate producer or a catering company is ultimately reflected in objective metrics that AI can analyze to provide actionable insights:

• Sales Volume: A direct indicator of consumer acceptance.
• New Contract Acquisitions: Reflecting market trust and growth potential.
• Customer Retention Rates: Demonstrating consistent performance and satisfaction.
• Market Expansion Indicators: Signaling scalability and potential growth.

If the quality of the food is lacking, these metrics will inevitably suffer. By parsing these quantitative signals, AI transforms subjective experiences into measurable insights, offering a more reliable and comprehensive picture of a company’s potential than any single, culturally biased tasting experience could provide.

The Grande Finale: Key Takeaways for the Investment Community

For investors, the taste of a product, like airline food, is not the ultimate determinant of a company’s potential. Instead, the true insights lie in measurable data points that reflect consumer satisfaction and operational success. AI’s ability to parse these objective signals offers a transformative advantage:

• Skip the Subjectivity: Direct sensory experiences, such as tasting food, are subjective and not necessary for informed investment decisions.
• Follow the Metrics: Indicators like sales volumes, customer retention, and new contract acquisitions are far more reliable measures of a company’s performance and growth potential.
• Leverage AI: Advanced AI tools can bridge the gap between subjective experiences and objective insights, empowering investors with data-driven predictions and actionable intelligence.

In essence, successful investments stem not only from personal experience but from understanding the quantifiable signals of consumer satisfaction, operational excellence, and market confidence—areas where AI can shine.

In the end, the food was nothing short of extraordinary—a culinary experience that soared beyond the clouds—but true investment wisdom lies in navigating the stars of data, where metrics illuminate the path to success.

References:

1. Fazemi, B., et al. (2023). “Learning to Substitute Ingredients in Recipes”, arXiv:2302.07960.
2. Taneja, K., et al. (2023). “Monte Carlo Tree Search for Recipe Generation using GPT-2”, arXiv:2401.05199
3. Huang, T., et al. (2024). “FoodPuzzle: Developing Large Language Model Agents as Flavor Scientists”, arXiv:2409.12832