weekend ai reads for 2024-11-01

November 01, 2024

📰 ABOVE THE FOLD: VOTE (if you’re in the U.S.)

give your colleagues grace on Tuesday if they choose to vote over sitting on Zoom for 10 hours

The Power of One Vote / Center for American Progress (37 minute read)

How they talk you out of voting / Seth (Godin)’s Blog (3 minute read)

 

📻 QUOTE OF THE WEEK

Right now we’re all adopting this thing and we don’t know what problems it causes … we don’t even know how it works.

Tony Fadell (source)

 

👥 FOR EVERYONE

  • giving a program with ADHD unfettered access to your computer might not be the wisest thing

  • like we said last week: there are a lot of rough edges on this

The UW researchers tested three open-source, large language models (LLMs) and found they favored resumes from white-associated names 85% of the time, and female-associated names 11% of the time. Over the 3 million job, race and gender combinations tested, Black men fared the worst with the models preferring other candidates nearly 100% of the time.

Between Hesitation and Hope / More In Common (7 minute read)

Women, rural Americans, and those with a low sense of belonging are generally more skeptical and fearful about AI’s impact compared to men, urban residents, and those with a strong sense of community.

 

📚 FOUNDATIONAL

I own my LLM chat history, and so should you / (Markus) Maragu’s Blog (8 minute read)

How I Studied LLMs in Two Weeks: A Comprehensive Roadmap / Towards Data Science (11 minute read)

  • maybe this should be “for Technologists” because step 1 is “Build an LLM from scratch”

  • also having done that, it is a great way to learn

 

🚀 FOR LEADERS

How to build an AI search engine (OSS Perplexity Clone) / Together.ai Documentation (11 minute read)

  • should be Step 1 in any organization’s AI journey

  • slight pitch for Together AI, but still a good primer

Creating a LLM-as-a-Judge That Drives Business Results / Hamel’s Blog (38 minute read)

  • such a useful guide

  • tldr; find a domain expert in your field and make your LLM converge with them

More than a quarter of new code at Google is generated by AI — AI is hugely important to Google’s products, and it sounds like the company relies on it internally, too. / The Verge (4 minute read)

  • developers review the code before committing it

The AI Productivity Paradox: Why Aren’t More Workers Using ChatGPT? — The real barrier isn’t technical skills — it’s time to think / Julia Winn, Towards Data Science (9 minute read)

This type of quality-focused work is often sacrificed in the rush to meet deadlines, yet it’s precisely what fuels breakthrough innovation. Paradoxically, most people don’t have time to figure out how they can save time.

 

🎓 FOR EDUCATORS

Artificial Intelligence and Writing: Four Things I Learned Listening to my High School Students / Brett Vogelsinger, Engaged Education, Substack (13 minute read)

Educause ’24: A Summary of Federal Guidance on AI / Government Technology (5 minute read)

Resources from the U.S. Department of Education’s Office of Educational Technology on generative AI include a guide to teaching and learning, the national ed-tech plan, an ed-tech developer’s guide and more.

Bill Gates on possibility, AI, and humanity — How many lives will be saved with the help of AI over the next decade? / Possible Podcast, Reid Hoffman & Aria Finger, YouTube (64 minute video)

 

📊 FOR TECHNOLOGISTS

Why I build open language models — Reflections after a year at the Allen Institute for AI and on the battlefields of open-source AI. / Interconnects (13 minute read)

The Gory Details of Finetuning SDXL for 40M samples / fpgaminer, Reddit (33 minute read)

Compared to version 1, I upped the training samples from 30M to 40M. I felt like 30M left the model a little undertrained.

  • may trigger NSFW warning; nothing objectionable at this post specifically

  • an archived link in case it’s unreachable

 

🎉 FOR FUN

AI Age of Empires II Assistant / WololoGPT (2 minute read)

WololoGPT is an intelligent assistant designed to enhance your Age of Empires II gameplay by providing real-time alerts and insights. Focus on strategy and decision-making while we help you manage routine tasks!

  • the next step is to let the model perform these tasks for you … so it can be like watching Twitch?

  • the novelty is that real, physical versions of the A.i. creations can be bought

 

🧿 AI-ADJACENT

This Theory of Everything Could Actually Work: Wolfram’s Hypergraphs / Sabine Hossenfelder, YouTube (12 minute video)

  • brief overview of Stephen Wolfram’s world-view, and the team around him making his theories mathematically viable (or at least not laughable)

 

⋄