What I Read This Week w/c 17th Feb 2025

?? Weekly Reading Recap

Can you believe it? - We are almost about 15% through 2025 already. Time has flown by! It slips through your finger like sand, even if you are being intentional about it.

So here’s a rundown of what I’ve been exploring this week:

?? Articles That Stood Out:

• Everyone’s lonely, but no one can hang out: Amanda Litman’s article was beautiful. It discusses the paradox of widespread loneliness despite people being more connected than ever through technology. She explores how modern life, technology, busy schedules, social media, and the pressure to appear “busy” have made it harder for people to genuinely connect and spend quality time together. Highly recommend!
• To Whom Does the World Belong?: Who owns the copyright of what the LLM writes? This article deep-dives into different possibilities. Hilarious and poignant analogy to begin.
• The Anthropic Economic Index: Based on the usage data of Claude.ai, parent company, Anthorphic has launched an economic index to assess AI impact on jobs over time. One key highlight stood out:

“AI use is more prevalent for tasks associated with mid-to-high wage occupations like computer programmers and data scientists, but is lower for both the lowest- and highest-paid roles. This likely reflects both the limits of current AI capabilities, as well as practical barriers to using the technology.”

That tells you that either Claude is not yet popular among the low-paid roles or else outside the high-paid roles, the use cases of AI are still not as explicit as we think they are.

• Idiosyncrasies in Large Language Models: This was such an interesting paper. It studies idiosyncrasies in the outputs of LLMs; unique patterns which distinguish different models. The authors of the paper attempt to classify the source LLM based on generated text and is able to achieve 97.1% accuracy(!) in distinguishing five models: ChatGPT, Claude, Grok, Gemini, and DeepSeek. For example: ChatGPT often starts a response with “Certainly”, “Here” or “Such as” while Claude likes to start the response with “According to”, “Based on”, or “This text”.
• Attention to Details: Mike Fisher gives some great anecdotes in here to drive home the point on attention to detail in both personal and professional lives.
• Humane’s AI Pin is dead, as HP buys startup’s assets for $116M: One of the first hardware built on the hype of AI is dead long before the hype is dead. Also, the hardware engineers associated with Humane were let go while HP retained AI software engineers (a rare commodity in this market)
• The geography of generative AI’s workforce impacts will likely differ from those of previous technologies: Pretty interesting one. Perhaps, first of a kind I have read. It is a study done by Brookings, that articulates that unlike prior technology shifts, AI shift will impact the workforce a lot more in cities with tech hubs, financial services, and creative industries. At the same time, rural areas could be less affected. It emphasises the need for targeted policies to address these disparities and prepare workers for the evolving job market.
• AI cracks superbug problem in two days that took scientists years: Mind-blowing one! The way Prof Penadés dealt with the outcome blew my mind more than the outcome. It was stoic and so logical. It dawned on me while reading that this is what it means to parse a piece of bad news rationally!
• Choose to lose: health plan choices from a menu with dominant options: I found the mention of this paper in Nudge. Since I am trying to do my financial planning, this paper resonated with me. It proves that how lack of understanding, rather than rational preferences, drives many suboptimal health insurance choices. The findings challenge the assumption that people can effectively navigate complex financial decisions, particularly in health care. As a result, policymakers should rethink how health insurance plans are structured and presented to consumers. Something that this paper points out (looks obvious in hindsight, but striking nevertheless) is that there is a cohort of employees who are likely to choose poorly and it is something you can pre-empt: low-income employees, older employees, female workers, and those with chronic health conditions.
• Andrew Garfield Wants to Crack Open Your Heart: This was such a heart-wrenching rendering of a NYT article by Andrew Garfield. Shweta & I listened to it during Valentine’s Day weekend.

?? Stuff I’m Watching:

• Silo: Just finished watching 2nd season of Silo which is streaming on Apple TV. It is based on a book called Wool, which Balaji Nageswaran recommended to me in 2019. Hell of a book. Hell of a series!
• Can’t get out of my head: It is a 2021 BBC documentary. I discovered it as a quote in MANIAC. Extremely hard-hitting and eye-opening especially for the environment we are in.
• Psych: It is our family’s guilty pleasure. We all see it together after our dinner. It is hilarious and meme-worthy.

?? Books I’m Reading:

• Finished Maniac by Benjamín Labatut: What a crazy disturbingly beautiful book! I would highly recommend everyone to read it. Very topical yet evergreen.
• Started reading and am already 35% through The Nvidia Way: Jensen Huang and the Making of a Tech Giant: No Biography Fluff. Hardcore. Just like Jensen Huang.

For a glimpse at what else I’m reading, check out my Goodreads profile.

?? Audiobook I’m Listening To:

• Finished Nudge by Richard Thaler & Cass Sunstein: I listened to the latest edition which has been upgraded with the relevant sections. This book introduced me to a couple of brilliant papers and the great concept of “choice architecture”.
• Started listening to a rendition of 1984 by George Orwell that is narrated by Andrew Garfield & Tom Hardy. I now get the hype!

I hope you enjoy this newsletter. Happy Reading!

Originally published here