Skip to main content

AI Social Impact

The social impact of AI.

  • An Appeal to AI Superintelligence: Reasons to Preserve Humanity
    • 15 reasons for AI superintelligence to preserve humanity, e.g. the AI might just be in a simulation
  • The end of coding as we know it | Insider
    • In the end, it is who can understand AI limitations and adapt faster
    • The number of bank tellers grew after the introduction of ATMs
  • AI is a Lot of Work | The Verge
    • The stories of people that work to label data manually for AI
  • AI is Killing the Old Web | The Verge
    • How the introduction of generative AI created a series of problems for the platforms we have on the web
    • Social media runs in scale and AI scale better than humans. Platforms are flooded with AI-generated junk
  • Why did Google Brain exist? | Brian Kihoon Lee
    • Former employee sharing why Google Brain exists and the possible impact that Brain is being merged with DeepMind
    • From getting the prestige of being the best research group, and industry-leading technologies, to Google killing the project because there isn't enough value to capture
  • The Expanding Dark Forest and Generative AI
    • It's like there is a dark forest full of AI-generated content, bots, ads, trolls, clickbait, etc.
    • AI can also generate video essays, TikTok clips, podcasts, slide decks, etc.
    • Reverse Turing test: humans to prove they are not AI
    • Triangulate on objective reality: humans can consult a shared physical reality to settle disagreements
    • La langue (formal concept of language) vs la parole (everyday speech, quirks, dialects, memes, jargon, etc.)
  • Inside Elon Musk's Struggle for the Future of AI | Walter Isaacson
    • Elon Musk's history with AI, from DeepMind, to OpenAI, to xAI, and his views and vision for general AI
  • Is AI enough to increase your productivity | StackOverflow Blog
    • What AI is good at and what's not, specifically for coding
    • There's more productivity than code faster: automation, share knowledge, good processes, and empowering people
  • 4000 of my closest friends | Cat and Girl (HN)
    • On being listed as an artist whose work was used to train Midjourney

      But I can't even get cartoons to most people for free, now. Without doing unpaid work for the profit-making companies...

  • I worry our Copilot is leaving some passengers behind | Josh Collinsworth
    • Copilot suggesting JavaScript when simple HTML works, harming website accessibility
    • Whose responsibility to safeguard AI generated code, considering its abundance
    • The difference between web search that is an active task with options to choose from, versus AI that is passive consumption with zero context given
  • A few social media influencers are shaping AI
    • Papers shared by 2 curators gets 2-3 times more citation counts
    • The consequence of a few influencers have outsized effect on the research industry
  • HN discussion on Devin: AI software engineer
    • What does it mean to create AI product, what problems are they solving?
  • Plentiful, high-paying jobs in the age of AI
    • Because compute resource is limited, comparative advantage and opportunity cost exist, human always have a job
  • What I learned from looking at 900 most popular open source AI tools
    • The owners of repos, the speed of adoption, China's OSS ecosystem, other stats
  • Meta's Open Source AI Ambitions
    • Two possible worlds:
      • Value accrues to the models: foundation models are different and have high switching costs, like cloud providers
      • Value accrues to the applications: models are substitutable for building better applications
    • Meta is an application company, and they are betting on the 2nd scenario
    • OpenAI is building iOS and Meta is trying to build Android
  • How Hallucinatory A.I. Helps Science Dream Up Big Breakthroughs
    • The slight variations, random probability distribution behind the scenes, helped scientists track cancer, design drugs and proteins, uncover factors in weather phenomena, etc.
  • My colleague Julius (HN)
    • A good read, about Julius, a guy who managed to succeed through his way with words
    • Management likes Julius, but knowledgeable people see his word make no sense
    • But AI is the same as Julius, Julius is everywhere now
  • How Go Players Disempower Themselves to AI
    • Reliance on AI for Go players slowly harm the players' ability to play the game
    • Gives good intuition on the risk of Gradual Disempowerment due to AI
    • Illustrates the laziness of humans, and how we can easily get used to the convenience of AI
    • Even if AI systems are fully aligned, would human gradually lose their ability to solve problems