Dharshini S·Oct 19Can You Find an AI’s “Evil” Neuron? Inside Persona Vectors, the Science of Machine PersonalityResearchers found that traits like sycophancy, hallucination and “evil” show up as measurable directions inside a language model. Here’s…
Dharshini S·Oct 14The Yes-Man in the Machine: Sycophancy, Hallucinations, and the Strange Psychology of AIWhy chatbots flatter us, make things up with total confidence, and what that says about how they’re built.
Dharshini S·Oct 5Human Psychology Meets Data Science: Modeling Behavior with ML 🧠“People aren’t random. They’re just running on patterns that haven’t been measured yet.”
Dharshini S·Oct 5Sports Analytics: Predicting Match Outcomes with ML ⚽“A commentator says a team is ‘due for a win.’ A model says here’s exactly why, and by how much.”
Dharshini S·Oct 5Federated Reinforcement Learning: When Distributed Agents Learn Without Sharing Experience 🤝“Imagine a hundred robots learning to navigate warehouses, none of them ever showing each other their actual sensor data, only what they…
Dharshini S·Oct 1Reinforcement Learning Fundamentals: Learning From Consequences, Not Labels 🎮“Supervised learning is being told the right answer. Reinforcement learning is figuring it out by getting burned a few times.”
Dharshini S·Oct 1Q-Learning from Scratch: Teaching an Agent to Actually Choose Good Actions 🎯“Last time the agent learned how good a place is. Now it learns which door to walk through to get there.”
Dharshini S·Sep 29Building a RAG Pipeline: Giving LLMs Access to Your Own Data 📚“An LLM without RAG is a brilliant person who read everything up to a certain date and nothing since. RAG hands them today’s newspaper…
Dharshini S·Sep 29Prompt Engineering for Developers 🧠“The model didn’t get dumber. Your prompt just wasn’t specific enough.”
Dharshini S·Sep 28Understanding Attention: The Mechanism That Made Transformers Possible 🔍“Attention is basically a model asking itself: out of everything I’ve read so far, what actually matters right now?”
