- That AI Thing
- Posts
- weekend ai reads for 2023-06-09
weekend ai reads for 2023-06-09
📰 ABOVE THE FOLD: DOOMSDAY (AGAIN)
via marc, Bill Gates Says "You'll Never Go to Amazon Again" After the Next Big Artificial Intelligence (AI) Breakthrough Motley Fool
point: Why AI Will Save the World Andreessen Horowitz
counterpoint: Marc Andreessen Is (Mostly) Wrong This Time Silicon Valley’s preeminent venture capitalist tries to craft the ur-narrative for generative AI, and in doing so lays bare its contradictions. Wired
Sci-fi writer Ted Chiang: ‘The machines we have now are not conscious’ Financial Times
I’m a rational optimist. Here’s why I don’t believe in an AI doomsday. Big Think
Top AI researcher dismisses AI ‘extinction’ fears, challenges ‘hero scientist’ narrative Venture Beat
📻 QUOTE OF THE WEEK
Shareholders are spending these extravagant sums for tasks that can now be done just as well by an AI program trained on 100 years of corporate reports and Wall Street Journal texts. What should we sell? What should we spend? What should we cut? Clearly, a sophisticated algorithm can now accomplish these tasks with a higher success rate than flawed humans who, all too often, do cocaine.
Hamilton Nolan, Automate the CEOs
🏗️ FOUNDATIONS
AI Compute 101: The Geopolitics of GPUs China Talk
Thinking Critically about AI in Healthcare [PDF, Google Drive] Jessica Morley, Oxford Internet Institute, University of Oxford
GPT best practices OpenAI
China Cracks Down on Surge in AI-Driven Fraud Authorities warn of hyperrealistic content generated by artificial intelligence Wall Street Journal
How I finally learned to love ChatGPT Fast Company
🎓 EDUCATION and AI
via patrick, A Map of AI for Education Laurence Holt, Medium
via rahim, Generative AI in Higher Education: From Fear to Experimentation, Embracing AI’s Potential Tyton Partners
Edtech giant Byju’s launches transformer models in AI push Tech Crunch
How AI Tools Both Help and Hinder Equity Inside Higher Ed
📊 DATA & TECHNOLOGY
point: What will GPT-2030 look like? Bounded Regret
counterpoint: Diminishing Returns in Machine Learning Part 1 From the New World
The Unintended Consequences of Censoring Digital Technology – Evidence from Italy’s ChatGPT Ban arXiv
spoiler: programmers know how to use VPNs
QLoRA: Efficient Finetuning of Quantized LLMs arXiv
Finetuning is feasible on a single GPU, aka: your laptop
Apple avoids “AI” hype at WWDC keynote by baking ML into products Reddit
AI chatbots lose money every time you use them. That is a problem. The cost of operating the systems is so high that companies aren’t deploying their best versions to the public Washington Post
🎉 FUN and/or PRACTICAL THINGS
Mini Studio The First Safe Chat-GPT for Kids
Redditor creates working anime QR codes using Stable Diffusion Ars Technica
Click for impressive examples.
Building a Tour de France Beer Map With GPT-4: A Non-Developer’s Perspective Nightingale Magazine
‘Flowers Blooming Backwards Into Noise’ (19:24) Eryk Salvaggio, Vimeo
The first four minutes will tell you if you’ll appreciate the whole video.
Painfinder Ask a question about user pain points in any software category:
Pretty good, but domain experts’ jobs are safe for now.
🧿 AI-ADJACENT
The Secret History And Strange Future Of Charisma Noema Magazine