Sunny

← All weeks

What I’ve been reading (W16, 2024)

Zuckerberg thinks · Here is more helpful/tactical advice from Kevin Kelly; it's a must-read

  1. Zuckerberg thinks – a. AI is a much lower level of innovation, just like computing (and not like platforms, i.e., computers, mobile or VR ) b. AI tool/model itself will never be a product – the core use case and infrastructure around it will be the product. c. Open source is less riskier than any powerful actor owning the most capable model. d. Reminds everyone to keep the main thing the main thing e. A lot of decisions should be based on conviction and values f. PS: They also launched a much bigger/better open-source AI model recently – it’s Llama3
  2. Here is more helpful/tactical advice from Kevin Kelly; it’s a must-read. I benefited greatly from reading this, some of his old advice and a recent book.
  3. This is the video editing setup of Joe Walker, the editor of Dune movies – a good reminder to obsess over outcomes/mastery, not tools.
  4. Linus Torvalds is convinced that AI can/will write better code, but it’s not as revolutionary as people make it sound. LLMs will be a big help in finding bugs (autocorrect the code)
  5. If you ever find yourself overwhelmed by everything you need to build to make something worth switching, you should just try narrowing your audience – use it as a wedge to make it incrementally more useful to a broader audience. (Tweet)
  6. It’s the best time to learn to code because you need less of it and produce increasingly more magic with AI. Eventually, “coding” will almost entirely be a natural language, but you will still be programming. You will be paid for your creativity and ability to get things done with computers. (Amjad on future of programming)
  7. There is a difference between being interested in something and being committed to it. When you are simply interested, you do what needs to be done when convenient. When you’re committed, you’ll do whatever it takes. (Some random tweet)
  8. A high degree of competence and basic financial security are prerequisites for a good life, but they are not necessarily the key factors in maintaining happiness. (Shreyas on Wisdom and Clarity)
  9. A good enough tool should be readily available on most readable platforms, be readable, be easy to navigate and play nicely with CLI and environment variables. Some gold standard examples: Redis Makefile, Llamafile Makefile, EdgeDB setup.py
  10. Boston Dynamics launched a new version of their Atlas bot – and it’s wild.
Substack edition

View the original post

  1. Zuckerberg thinks –

    1. AI is a much lower level of innovation, just like computing (and not like platforms, i.e., computers, mobile or VR )

    2. AI tool/model itself will never be a product – the core use case and infrastructure around the product.

    3. Open source is less riskier than any powerful actor owning the most capable model.

    4. Reminds everyone to keep the main thing the main thing

    5. A lot of decisions should be based on conviction and values

    6. PS: They also launched a much bigger/better open-source AI model recently – it’s Llama3

  2. Here is more helpful/tactical advice from Kevin Kelly; it's a must-read. I benefited greatly from reading this, some of his old advice and a recent book.

  3. This is the video editing setup of Joe Walker, the editor of Dune movies – a good reminder to obsess over outcomes/mastery, not tools.

  4. Linus Torvalds is convinced that AI can/will write better code, but it’s not as revolutionary as people make it sound. LLMs will be a big help in finding bugs (autocorrect the code)

  5. If you ever find yourself overwhelmed by everything you need to build to make something worth switching, you should just try narrowing your audience – use it as a wedge to make it incrementally more useful to a broader audience. (Tweet)

  6. It’s the best time to learn to code because you need less of it and produce increasingly more magic with AI. Eventually, “coding” will almost entirely be a natural language, but you will still be programming. You will be paid for your creativity and ability to get things done with computers. (Amjad on future of programming)

  7. There is a difference between being interested in something and being committed to it. When you are simply interested, you do what needs to be done when convenient. When you’re committed, you’ll do whatever it takes. (Some random tweet)

  8. A high degree of competence and basic financial security are prerequisites for a good life, but they are not necessarily the key factors in maintaining happiness. (Shreyas on Wisdom and Clarity)

  9. A good enough tool should be readily available on most readable platforms, be readable, be easy to navigate and play nicely with CLI and environment variables. Some gold standard examples: Redis Makefile, Llamafile Makefile, EdgeDB setup.py

  10. Boston Dynamics launched a new version of their Atlas bot – and it's wild.

Originally on Substack