Urusan / is on Nostr

Java developer by day, Julia developer by night.

Amateur philosopher

Sometimes funny...

Working Dad

Controversial things about me:
Everyone: transhumanist, into AI (art)
Right-wing: polyamorous (married), agnostic atheist, leftist, working class consciousness
Leftist: corporate drone by day, loyal citizen of the US (but a serious reformer), former libertarian

I hope you can look past all that though, we people need to stick together

Lives with: Wife, T (son), and A (daughter).

Pronouns: he/him/his
I am a: cis het white male

Last Notes

2024-07-08 00:26:39

The Steam Deck is...

2024-06-23 21:36:47

Let's say you find yourself frequently using rsync to synchronize files between two computers. You're repeatedly entering in the same commands to either sync to or from each machine at the same locations.
How would you make this process more convenient for yourself?

2024-06-14 00:36:44

Someone says "Wow, you're such a nerd!"
What is your reaction?

2024-06-01 01:16:50

Which of these best describes Batman?

2024-05-12 13:03:17

This is an interesting video:
https://youtu.be/dDUC-LqVrPU
TL;DW We're starting to see early evidence of diminishing returns with our current AI architectures. If this is true, then eventually they start to improve in a logarithmic manner, making superintelligence (at least using our current architectures) impossible to achieve from a practical standpoint. The issue is that we need too much data on specific things for it to perform well on them all.

2024-05-12 02:58:35

- reply

There are also more options than just simple allow/deny filtering, especially with the details provided by the AI system.
For instance, it could auto-CW posts (including those from foreign instances without the same CW rules) rather than doing simple allow/deny filtering.
This would be especially useful at the individual level since it would make it easier to set up consistent filtering based on CWs.

2024-05-12 02:23:11

- reply

This idea really isn't that dissimilar to a spam filter (you even train spam filters on spam examples).
The main difference would be in detail and flexibility. Spam filters are great at detecting immediate, obvious abuse. That's great, and that's still necessary to help deal with large scale attacks.
What about more subtle patterns of abuse though? Or rules specific to an individual/instance/community? If you put this stuff in a spam filter you would filter too much.

2024-05-12 01:50:24

Over on my AI-focused account, there was some discussion about AI moderation on Fedi.
The concept is to use the AI system to categorize posts according to some classification system that instance admins can easily tailor, which makes moderation much easier.
This might be actually beneficial for a relatively decentralized network like Fedi. By reducing the cost of moderation, it would mean more small and single user instances, which might be worth the downsides.

2024-05-10 00:03:32

Is exclusivity valuable?

2024-05-09 03:33:21

I'm really annoyed by the recent rhetoric that AI training is the worst intellectual (property) theft in history when the enclosure of the 20th century's public domain under effectively perpetual copyright is in fact the worst such theft in history.
I'm against copyright entirely, but if it must exist then it should last at most 14 years, not a century-ish.
If we had reasonable copyright terms (even 30 years!) there would be no lack of public domain training data for AI.

2024-04-01 00:05:18

I'm sorry to inform you that society has cancer.
It's pretty widespread, it's hard to provide a diagnosis for something like this but I'm pretty sure it's analogous to stage 3 or 4.

2024-03-20 23:47:31

Is your brain included in your body?