Posts

Showing posts from July, 2025

What SQL Joins and Quilting Have in Common?

Image
  When I think of joins, I think of quilts. Let’s be real— JOINS are not always simple . Especially when your tables don’t have a primary key. Or when the data model is not documented anywhere and you’re just trying to make sense of it all. Picture this: You’re at your desk, staring at your screen (remember to blink). You’re writing a query to count the number of employees per department.  It should be simple Two tables: employees and departments . A quick JOIN, a simple GROUP BY , and done. Right? You hit "Execute" Boom— 5,732 rows . That can’t be right. Your company only has 200 employees. You frown. You try DISTINCT . Still too many rows. You tweak the join condition, change the table order, throw in a LIMIT just to see something normal—but nothing works. It’s still chaos. Welcome to the world of SQL joins—where your logic seems sound, but your results make no sense. Why This Happens Joins are powerful—but they’re picky. If you join on the wrong column, or if your...

Is it OK to kick a robot dog?

Image
Boston Dynamics created a robot dog named Spot that can dance Is it okay to kick a robot dog? It’s a funny question on the surface. It’s not alive . It doesn’t feel pain. So… who cares? Lately I’ve been reading a book on AI Ethics by philosopher Mark Coeckelberg, the book is filled with tricky questions that make you stop and rethink our relationship with machines—especially the smart ones. Let me walk you through a few of them, with a few everyday examples to help make sense of it. Q1: Should We Hold Robots Responsible for Their Actions? Imagine this: a self-driving car runs a red light and causes an accident. Who do we blame? The car? (It was the one driving!) The engineers who built it? The company that sold it? Mark brings up a similar question in the book: “We don’t hold very young children responsible for what they do because they don’t know better—should the same be done for AI?”  Q2: What are the things that should only be done by humans? Let’s say AI gets really, really...