The MQM program offers many opportunities for learning and engagement in the classroom, including guest speakers. One of those speakers, which I heard from during Professor Jiaming Xu’s Modern Analytics class, has played a pivotal role in shaping my data science journey.

Jordan Meyer, founder and CEO at Spawning and a Kaggle Zillow Prize winner, presented on the topic “The Applied Science Grind – Building and Maintaining Multimodal Models”.

He posed a simple question: “Should we build this?” Crucial to the initial stages of machine learning model development, this question can be interpreted in four distinct ways, reflecting its depth and significance in the decision-making process. More than just a question, it’s a way of thinking that encourages us to stop and consider the outcomes before investing time and resources into building a model.

‘SHOULD’ we build this?

The emphasis on ‘should’ is intentional, compelling us to contemplate whether it’s ethical and practical to embark on building the model. In this stage, it’s important to evaluate the model’s purpose, potential impact, and possible consequences. It’s not just about the technical feasibility but also about the societal implications and whether the outcome justifies the effort and resources invested.

Should ‘WE’ build this?

This question focuses on the capabilities and suitability of the team or organization. We should consider whether we have the right expertise, resources, and understanding to successfully undertake this project or if we should seek external assistance, like from a third-party consultancy. This also includes examining our strengths and weaknesses, and understanding whether or not the project aligns with our goals and capabilities.

Not every project is fit for everyone. The ability to recognize this should be viewed as a strength, not a limitation.

Should we ‘BUILD’ this?

This focuses on the action of building the model itself. It questions the feasibility and the approach. At this stage, we are considering if building a new model is the best solution or if existing models could be adapted or improved. This stage also includes evaluating the technology, methodology, time, and resources required for building the model.

Should we build ‘THIS’?

In an age where innovation is often equated with success, this question grounds us in the reality of need and impact. Focusing on the model’s relevance and utility, this step is crucial for determining if this particular model is the right solution, considering the available data and how the model integrates into the broader system or context.

Meyer’s question has left a lasting impact and has influenced how I approach every project. The power of the question, “Should we build this?”, lies in its simplicity and depth. It shows that just changing how you look at things, or the way you say something, can bring out really important ideas and help you make smart choices.

As technology continues to rapidly evolve, I’m encouraged to continue grappling with and embracing this question. It’s a constant reminder to pause and look at the bigger picture of what we’re doing. This will help us make sure we’re not just moving forward with our technical abilities, but also with careful thought and a clear understanding of why we’re building this model.

While my current focus is on applying these principles to solve problems within the business domain, I’m eager to explore the ethical implications of machine learning on a broader societal level. Perhaps in the future, I can contribute to the development of frameworks that ensure responsible and impactful applications of machine learning across various industries.