Taking a look how our most Advanced AI Language Model to-date can shape our perception of NLG technology — As of writing this, it’s hard to go surfing anywhere online and not see articles about ChatGPT. 🏄 What has interested me the most in this current news cycle about ChatGPT is: ChatGPT, and the discussion around the model’s capacity to have an “intelligent” conversation, leads to some interesting questions…

Core concepts of Intelligence that ChatGPT lacks, and tech available today that can make it smarter — ChatGPT is everywhere right now. And for good reason. It is really impressive. Some might call it intelligent. Some might call it conscious (it’s not.) But the interesting question to ask is: How intelligent is ChatGPT? How does it get more intelligent? According to Gary Marcus, there are a some…

Last Summer, the stars aligned for me to attempt my first multi-day fast. Fasting often has a religious connotation, but I was more driven by data. I was feeling stuck in my diet and exercise goals, and looking for something like a reset to “get back on the horse” as…

Strategies of quantifying bias in language models for responsible AI — In the past few years, the topic of building FAT (Fair, Auditable, and Transparent) Machine Learning has blossomed from a fringe topic to one dominating the conference scene. This is particularly the case for Natural Language Processing, where the impact of biased ML is quite obvious and pronounced. One such…

A simple-to-follow example about QA-ing your ML Models — I’ve been thinking a lot recently about how to better systematically understand and explore complex, “black box” machine learning models “Black Box” ML models referring to statistical models that it is unclear what exactly the learned activation function is. There are, of course, a host of incredibly important reasons as…

A Friendly how-to guide on extracting Garmin Venu 2 data for people in a hurry — About 6 months ago, I wrote on how to convert .fit files from Garmin into CSVs. Since then, I got a new Venu 2 Garmin watch, and with it there is more data to process!! Specifically around cleaning sleep and activity data from my Garmin wearables Just as before, I’m…

My experience/experiment quantifying the impact of breathing on stress — Just Breathe Man. Mindfulness and breathing exercises have always been around, but have become in-vogue in the past few years. Do they really have any benefit? 2 things piqued my interest in the subject: 1. New Research: I came across a new 2021 study from CU Boulder showing that a single 5 minutes…

The world is growing more distrustful of AI, and it’s AI developers fault — This is part 2 of our series on ML Fairness (link). Previously, we discussed how ML teams rarely attempt to address societal bias in models due to a myriad of (weak) reasons. In this article, we will explore the consequences of this neglect. Thesis: Due to high-profile ML failures and…

A brief look at Societal Fairness in Machine Learning models — ML has a PR problem. Even as our capacity for smarter, more nuanced, more powerful ML accelerates by leaps and bounds, producing hugely positive societal value, consumer apprehension around ML has not subsided. In fact, it seems to be growing. This trend is alarming for any company, team, or professional…