In an hour, I built a production-grade ML app with full documentation—without writing a single line of code. If AI is the mason, students must become the Architects. Here’s why we need to stop grading the product and start assessing the process.
In 2021, Zillow shuttered its “Offers” division after a staggering $880 million loss. It wasn’t a market crash or a lack of data that did them in—it was the math itself. Explore how one of the world’s most sophisticated regression models became “confidently wrong” at scale, and why the most expensive mistake a data-driven company can make is confusing statistical confidence with situational awareness.
We are inviting educators and researchers to participate in a study at SUTD evaluating the pedagogical quality of AI-generated tutoring responses. Participants will review 24 pairs of anonymised dialogues for middle-school mathematics to help us understand how AI demonstrates qualities like clarity, scaffolding, and learner support. The online session takes approximately 45–60 minutes.
Math isn’t a gatekeeper—it’s a lens. For too many aspiring data scientists, mathematics is a wall of dry formulas to be memorized for interviews and then forgotten. But true mastery lies in moving “Beyond the Code.” In this post, we strip away the intimidating notation to reveal the intuitive elegance behind five core concepts. From the “magic” convergence of the Central Limit Theorem to the geometric “shadows” of the Dot Product, discover why understanding the mechanics of uncertainty is the precise boundary between a coder who merely executes scripts and a data scientist who truly understands meaning.
Is this the end of human coding? Anthropic’s Boris Cherny admits he hasn’t typed code in a month while building the new Claude Code tool. We analyze how agentic AI is shifting the developer’s role from writing syntax to orchestrating products.
In an hour, I built a production-grade ML app with full documentation—without writing a single line of code. If AI is the mason, students must become the Architects. Here’s why we need to stop grading the product and start assessing the process.
In 2021, Zillow shuttered its “Offers” division after a staggering $880 million loss. It wasn’t a market crash or a lack of data that did them in—it was the math itself. Explore how one of the world’s most sophisticated regression models became “confidently wrong” at scale, and why the most expensive mistake a data-driven company can make is confusing statistical confidence with situational awareness.
We are inviting educators and researchers to participate in a study at SUTD evaluating the pedagogical quality of AI-generated tutoring responses. Participants will review 24 pairs of anonymised dialogues for middle-school mathematics to help us understand how AI demonstrates qualities like clarity, scaffolding, and learner support. The online session takes approximately 45–60 minutes.
Math isn’t a gatekeeper—it’s a lens. For too many aspiring data scientists, mathematics is a wall of dry formulas to be memorized for interviews and then forgotten. But true mastery lies in moving “Beyond the Code.” In this post, we strip away the intimidating notation to reveal the intuitive elegance behind five core concepts. From the “magic” convergence of the Central Limit Theorem to the geometric “shadows” of the Dot Product, discover why understanding the mechanics of uncertainty is the precise boundary between a coder who merely executes scripts and a data scientist who truly understands meaning.
Is this the end of human coding? Anthropic’s Boris Cherny admits he hasn’t typed code in a month while building the new Claude Code tool. We analyze how agentic AI is shifting the developer’s role from writing syntax to orchestrating products.