• 6 Posts
  • 639 Comments
Joined 10 months ago
cake
Cake day: January 16th, 2024

help-circle




  • I worked at MSFT between 2021 and 2023.

    The Growth Mindset is very much just a gaslighting tool. To be honest I didn’t get the culty vibe from it while being on the inside, on the other hand no one ever tried to make me to read Satya’s stupid book (thankfully).

    One important thing I just have to talk about is The Layoffs. If you ask me about “Growth Mindset”, or indeed if I ask around my former MSFT colleagues about the first thing that comes to their mind when they hear it, it will be that time when, not even a month after the massive 2023 layoffs where MSFT fired 11,000 people, we were told by management at a Townhall that it is time for us to “apply Growth Mindset and move forward”. I remember very clearly that they tried to spin it as if the layoffs were something that just “happened to us” and we had to move on, as if it was a hurricane that hit the office and not a deliberate act of management to cut costs. It was fucking amazing to hear that from them after I had a literal panic attack due to the uncertainty after the first wave of firings.

    I made the decision to quit not long after. When I was leaving the genAI brain rot was already in full swing. The stuff about autoplaging Connects is just a great affirmation of my decision, that company is fucked.








  • People who train neural networks do not write a bunch of tokens and weights.

    Reading this made me think of an analogy of generated code. This is basically exactly the same thing as distributing the code of your program but not in the source language, rather the assembly listing of the final binary, and calling it open source. You can turn any defense of the AI model of “open-source” into a defense of that model of distributing code. You can run my AI/code (if you have a powerful/similar enough machine), you can inspect it (it’s just not going to tell you anything), you can modify it (lol), so it’s open source!

    Edit: The more I think about it the more I come to the realisation that the assembly listing is actually still vastly more useful than the AI models. Like at least a very dedicated and insane enough programmer could technically track down a bug in the assembly and correct it if given enough coffee.