Rami's Readings #115 - Apple Fall Out
The latest on AI, LLMs, Apple's Illusion of Thinking, Andrej Karpathy, Leaky Thoughts, The Future of Work, Frontier AI Labs, and more.
Welcome to Rami’s Readings #115 - a weekly digest of interesting articles, papers, videos, and X threads from my various sources across the Internet. Expect a list of reads covering AI, technology, business, culture, fashion, travel, and more. Learn about what I do at ramisayar.com/about.
👋🏼 Welcome New Subscribers
Hello! A hearty thank you for subscribing to Rami's Readings! There are quite a few new subscribers this week, thanks to recommendations from The AI Ethics Brief, Global Fintech Insider, and The VC Corner. I am thrilled to have you on board! In this newsletter, I curate the best papers, tweets, and articles I have read during the week focusing on LLMs, AI, economics, business, and technology news. You can learn more about me on my website.
📈 Top Recent Editions According to Substack
🤖 AI Reads
I shared Apple’s Illusion of Thinking paper in #113 with the following TLDR: “Models aren’t really developing generalizable reasoning and training data is contaminated.” Unsurprisingly, the paper generated strong reactions across the AGI-optimism spectrum. Frankly, I’m not sure why the paper’s conclusions weren’t already obvious to most GenAI practitioners. I guess the dream of AGI is blinding. Now none of this invalidates the value of inference-time reasoning; it is useful, or at the very least entertaining to watch, and it should be exposed to users. Exposing the model’s reasoning chain to users would surface errors that wouldn’t be obvious when users can only see the final result. Two of the most interesting reads about Apple’s paper (both from
).The follow-up: Seven replies to the viral Apple reasoning paper – and why they fall short. I particularly like what Marcus wrote in response to one common dismissal, that “the paper was written by an intern.” There is a weird obsession with PhDs in AI circles… it’s hard for me not think about Claude Shannon whose master’s thesis laid the foundation for modern computing. I wonder what he would say.
MBZUAI Releases Nile-Chat: Egyptian Arabic LLM
Notes: Egypt has a population over 100 million. This model joins Mistral and Falcon 3 with improved dialect-specific generation.
From Bytes to Ideas: Language Modeling with Autoregressive U-Nets
Notes: Highly technical paper from FAIR at Meta and a few other universities.
Leaky Thoughts: Large Reasoning Models Are Not Private Thinkers
Notes: Yes.
How We Built Our Multi-Agent Research System
Notes: Great article from Anthropic.
💼 Business Reads
Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce
Notes: A great paper from Stanford about the future of work with AI Agents. It’s a very easy and short read. Hints at bad news for my friends building legal AI Copilots.
Frontier AI Labs: the Call Option to AGI
Notes: Great read from . Relevant for founders.
Silicon Valley’s ‘Tiny Team’ Era is Here
Notes: Solo founders are also being taken more seriously now.
Signing off from Redmond, WA.