Login
You're viewing the dementedandsadbut.social public feed.
  • Oct 24, 2026, 1:49 PM

    I was really into this “New Skill November” idea — I started teaching myself Rust on October 16 — until I read the right-hand side and realized these people would not be at all happy about my project.

    I ask Claude lots of questions about Rust. When I say “This trait thing looks a lot like typeclasses” it has a coherent reply. It doesn't say “well, I don't know Haskell, so I can't really say.”

    Probably its answers are not always correct. If someone thinks that is a serious problem, they should sit down and take a deep breath before I reveal some news about human experts that they will find horrifying.

    And when this graphic says “ask your fellow human beings for help”, it brings up memories of being used as a beach ball by insecure geeks on IRC.

    Nevertheless, I am “doing the hard part myself”, and when I'm done it will be me who can program in Rust, because I am a serious and diligent scholar, regardless of what tools I use along the way.

    (Original post: infosec.exchange/@0xabad1dea/1 )

    A graphic promoting "New Skill November":

Center:

New Skill November

art by Charles Louis Hinton for a November issue of Harper’s Bazaar, 1896
banner by 0xabad1dea@infosec.exchange

Left column:

LEARN A NEW SKILL OR REBUILD AN OLD ONE

Sketching, knitting, whittling, a musical instrument, a foreign language, a tricky video game, cooking, weaving, changing oil, riding a bike, painting, pixel art, website design, digital image editing, bookbinding, calligraphy, identifying mushrooms, embroidery, soldering, that one thing you’ve been putting off

Right column:

THE ONLY RULE: DO IT YOURSELF, DO NOT USE AI

Do the hard part yourself, no matter how imperfect it feels. Learn and internalize how it really works. Consult guides written by human experts, and ask your fellow human beings for help. Take pride in knowledge and abilities that are an intrinsic part of your existence and not dependent on renting from corporations.
    💬 4🔄 2⭐ 1

Replies

  • Oct 24, 2026, 2:01 PM

    People whose first language was not English used to come into EFNet's #perl channel and ask something like “how do u remove first char from a string in perl?” and a certain one of the regulars there, indisputably a human expert on Perl, would regularly reply:

    “No u”.

    The visitor was expected to understand from this that this human expert would answer only if they spelled “you” as y-o-u.

    Let me know when Claude starts doing that, I'll reconsider my position on this.

    💬 0🔄 1⭐ 2
  • Oct 24, 2026, 1:59 PM

    @mjd , doing the hard part yourself is one factor of responsible use, the one most positivists are ignoring. You'll take professional responsibility and never blame your tools because you're a pro and a craftsman both, and that's not new to us; if my tool isn't sharp, that's on me too. So you'll avoid the self-hollowing downside.

    Your limited use mostly avoids my original reason for preferring engineered expert systems over trained networks which mostly still applies

    💬 1🔄 0⭐ 0
  • Oct 24, 2026, 2:09 PM

    @mjd
    ... that trained NNet Models can't provide explanations or audits (subtly different but overlapping the more usual 'non deterministic' complaint), which is still an indictment for decision support, finance, HR applications of this 4th generation of perceptrons.

    The several new ethical issues with the economics of omnipresent Large deep general purpose core models ( incl. externality costs) remain.

    (small local models don't pose fresh externalities but may be derived...)

    💬 0🔄 0⭐ 0
  • Oct 24, 2026, 7:50 PM

    @mjd I had the exact same feeling. Love the left half and what it is suggesting. Absolutely think the right half is missing that AI can be used as a tool for learning rather than copying (or whatever)

    💬 0🔄 0⭐ 0
  • Oct 25, 2026, 12:51 AM

    @mjd 'guides written by human experts'? The same people who sneer and tell you that you're doing it wrong and use circular answers to never really answer questions? Who, instead of answering a question tell you that the 'right' way is to do something else entirely?

    Yeahhh, I can see why you might balk at that ;)

    💬 0🔄 0⭐ 0