The Y in AI

Imagine a vast and ancient tree of collective human knowledge.

Its roots burrow deep into the earth, its branches reaching high into the sky, and its leaves, whispering secrets in the wind. This tree of knowledge is dense. Challenging to navigate. Full of strong foundations as well as weak twigs ready to snap. Like Norse mythology, we are emulating Odin, seeking a profound wisdom and comprehension, which led him to Yggdrasil - The World Tree. And our relentless thirst for all-knowing knowledge.

Enter AI - a tool that transforms into an extensible ladder, allowing us to climb this tree with greater ease - ascending higher, reaching new heights of understanding and predictability.

As we climb, we discover that AI not only provides access to this immense wealth of knowledge, but it services as a guide, helping us traverse the tree's intricate branches and decipher its secrets.

In exchange we take care of the ladder (AI) for our safety and its longevity. We deliver regular inspections for defects before using the ladder, we clean its moving parts, replacing and oiling them periodically to ensure its smooth operation. We keep the ladder in proper storage to avoid rust and avoid overloading it with heavy items. We invest all our money and time into the promise of new frontiers we can reach upon each step of the ladder.

And yet, it is essential to remember that AI remains a tool, a means to an end, a ladder to help us reach new heights of human potential. We must avoid Odin's hanging fate and sacrifices. The power of exploration, understanding and the pursuit of knowledge ultimately reside within ourselves, not the AI ladder. What do we really want? What do we desire? What does the world need? Why? We need to put the Why (Y) into AI.

In the context of artificial intelligence (AI) and traditional machine learning, Y typically refers to the dependent variable or the output that you want to predict or classify. Here’s a quick breakdown:

• X represents the independent variables or features used to make predictions.
• Y is the outcome you’re interested in predicting based on those features.

For example, if you’re trying to predict house prices, X might include features like square footage, number of bedrooms, and location, while Y would be the actual price of the house.

Now lets review a more complex problem. By analyzing the relationship between X and Y, machine learning models can help identify patterns and make predictions that could lead to more effective cancer treatments.

X (Independent Variables)

Features: These are the input variables used to make predictions. In cancer research, X could include:

• Genomic Data: Information about mutations, gene expressions, and other genetic factors.
• Clinical Data: Patient demographics, treatment history, and response to previous therapies.
• Histopathology Data: Characteristics observed in tissue samples, such as cell morphology and tumor micro-environment.
• Lifestyle factors: Information about diet, exercise, and environmental exposures.

Y (Dependent Variable)

Outcome: As a reminder this is the variable you want to predict. In the context of cancer treatment, Y could represent:

• Treatment Response: Whether a patient responds positively to a specific treatment.
• Survival Rates: The likelihood of patient survival over a certain period after treatment.
• Cure Rates: The probability of complete remission or cure based on various treatment regimens.

Whether you are trying to cure cancer, helping your business grow or simply trying to be more productive, the reasons for using a tool such as Artificial Intelligence are important. You need to understand what variable, outcome or hypothesis you are trying to predict first. You need to consider the Y's because this time X does not simply mark the spot.

But can't AI now generate the Y (Dependent Variable) I seek?

In generative AI, like ChatGPT, the concept of Y can be a bit different compared to traditional machine learning. While Y in generative AI may not always represent a target variable or outcome in the same way as in predictive modeling, it still plays a crucial role in defining what the model aims to produce and how it learns from data. In simplified terms:

GenAI Output Generation: In generative models, Y often represents the output that the model generates, such as text, images, or music. For example, in a text generation model, Y would be the generated sentences or paragraphs based on the input prompt (X).

Training Data: During training, generative models learn from pairs of input (X) and output (Y). For instance, in a large language model, the input might be a sequence of words (X), and the output (Y) would be the next word or phrase the model predicts.

Conditional Generation: In conditional generative models, Y can represent specific conditions or attributes that guide the generation process. For example, if you want to generate images of cats, Y could specify the breed or color of the cat.

Evaluation: Just like in traditional ML, evaluating the quality of the generated output (Y) against some ground truth or expected outcome is essential. This helps in fine-tuning the model and improving its performance. Fine-tuning in machine learning is the process of adapting a pre-trained model for specific tasks or use cases.

Conclusion: Y matters

Why AI? Why we choose to use AI as a tool matters. What solidified Yggdrasil’s importance in the Norse cosmology was the tree's purpose - the axis connecting all the Norse realms of reality together. Y represents the outcome(s) you want to predict. Y is the backbone of the learning process in machine learning, shaping how models are built, evaluated, and applied.

If you’re interested in specific machine learning techniques or applications in research, let me know!

*Disclaimer: All the views and opinions expressed are those of the authors alone.