You make valid points about the common understanding of "data-driven" decisions. I want to intentionally broaden the definition of data to include a wider range of inputs influencing our choices.
Let’s take the hiring process example. While it feels like a "gut feeling," I'd argue it's still rooted in data - just not the clearly quantifiable kind. This "data" includes past experiences, cultural conditioning, and even evolutionary predispositions.
I think in the current age of LLMs and multi-modal models, you can consider a podcast as data. For example, you can use the podcast to train a model. We would consider it as ‘data’ that the model is trained on, in terms of LLMs. So why not consider it as data, when we humans listen to it?
The brain is like a neural network, so the networks have some weights and biases. Every time we add new data to the brain, through various forms like sight, hearing, smell, taste, touch, etc., we modify the weights and biases of our brain. Regardless of what the data is or its relevance, it changes. So, arguably, anything we consume could be data.
I don’t think you can make the argument that Earth is data-driven. The Earth follows the laws of physics. It’s just an object without the ability to process information, so just because it is following a mathematical path, it doesn’t mean it is data-driven.
The core argument isn't that all decisions are based on spreadsheet-style data, but that what we perceive as intuition is complex processing of various inputs accumulated over time.
By expanding our concept of "data," we can gain deeper insights into human decision-making, including seemingly instinctual or emotional choices. I want to complement the current colloquial understanding of data by recognizing the full spectrum of information influencing our decisions.
Let’s take the hiring process example. While it feels like a "gut feeling," I'd argue it's still rooted in data - just not the clearly quantifiable kind. This "data" includes past experiences, cultural conditioning, and even evolutionary predispositions.
I think in the current age of LLMs and multi-modal models, you can consider a podcast as data. For example, you can use the podcast to train a model. We would consider it as ‘data’ that the model is trained on, in terms of LLMs. So why not consider it as data, when we humans listen to it?
The brain is like a neural network, so the networks have some weights and biases. Every time we add new data to the brain, through various forms like sight, hearing, smell, taste, touch, etc., we modify the weights and biases of our brain. Regardless of what the data is or its relevance, it changes. So, arguably, anything we consume could be data.
I don’t think you can make the argument that Earth is data-driven. The Earth follows the laws of physics. It’s just an object without the ability to process information, so just because it is following a mathematical path, it doesn’t mean it is data-driven.
The core argument isn't that all decisions are based on spreadsheet-style data, but that what we perceive as intuition is complex processing of various inputs accumulated over time.
By expanding our concept of "data," we can gain deeper insights into human decision-making, including seemingly instinctual or emotional choices. I want to complement the current colloquial understanding of data by recognizing the full spectrum of information influencing our decisions.