On the false marketing of LLMs
Really enjoyed this blog post and felt it raised some excellent points about the use of LLMs.
One of the things I had to go do further reading on was Kranzberg's laws:
- Technology is neither good nor bad; nor is it neutral.
- Invention is the mother of necessity.
- Technology comes in packages, big and small.
- Although technology might be a prime element in many public issues, nontechnical factors take precedence in technology-policy decisions.
- All history is relevant, but the history of technology is the most relevant.
- Technology is a very human activity – and so is the history of technology.
The blog post references the first one in this list as a note that it isn't true, especially as it relates to LLMs.
Leaving aside Kranzberg's first law, large language models are the very antithesis of a neutral technology. They're imbued with bias and political decisions at every level.
There's the obvious problem of where the training data comes from. It's stolen. Everyone knows this, but some people would rather pretend they don't know how the sausage is made.
But if you set aside how the tool is made, it's still just a tool, right? A building is still a building even if it's built on stolen land.
Except with large language models, the training data is just the first step. After that you need to traumatise an underpaid workforce to remove the most horrifying content. Then you build an opaque black box that end-users have no control over.
Take temperature, for example. That's the degree of probability a large language model uses for choosing the next token. Dial the temperature too low and the tool will parrot its training data too closely, making it a plagiarism machine. Dial the temperature too high and the tool generates what we kindly call "hallucinations".
Either way, you have no control over that dial. Someone else is making that decision for you.
A large language model is as neutral as an AK-47.
Slightly later in the post he also says:
You could even convince yourself that a large language model is like a bicycle for the mind. In truth, a large language model is more like one of those hover chairs on the spaceship in WALL·E.
I love this comparison because this is exactly it. Also, it reminded me of this chart which is wholly unrelated -- in showing how efficient bicycling is as a form of motion - not just for humans, but for all animals:
|
After spending 30 minutes trying to troubleshoot adding a new extension to my selfhosted FreshRSS I finally abandoned it and just did a quick 30 second Userscript that does it client side with Javascript.
I don't know what is going on with the FreshRSS extension, but sometimes it isn't worth doing it right - I just want to fix my problem.
|
Fountain of Youth (2025)
The movie is not amazing. It's the latest in the lineage of Indiana Jones, the Librarian, and National Treasure. But, it is not an amazing movie. I hope it does well enough they get to do another one and perhaps find their groove.
Also, I legit can't remember much of the soundtrack at all, but there is a Thai version of 'Bang Bang' at the start of the movie which is a banger:
|
Automated Archives for May, 23rd 2025
This post was automatically generated.
Articles To Read
The following are articles that I saved today. Substance and quality will vary drastically.
- The “Invasion” Invention: The Far Right’s Long Legal Battle to Make Immigrants the Enemy
- How to Disappear
Chess For the Day
Record: 4-0-10
Net Elo Change: -21
Games Played
- basuki_ckd - WIN
- IL1keC8ke - LOSS
- islam_ramzy - LOSS
- C3pooo - WIN
- yosef_alwese - LOSS
- NewToChessNob - LOSS
- henry18hd - WIN
- primodelpatodonald - WIN
- NeronKaligulic - LOSS
- oktabadi - LOSS
- sochoavi - LOSS
- Rustam_007 - LOSS
- ogbchris - LOSS
- SonerGokdag - LOSS
Blog Posts On This Day
- May 23, 2024 (2 posts)
- May 23, 2023 (4 posts)
- May 23, 2022 (13 posts)
- May 23, 2021 (1 post)
- May 23, 2002 (1 post)
|
