AI discussions

hhh · 2026-05-11 23:01:21

I was born in Waltham, Mass. to a Turkish dad and my mom was from Montreal.

When I was born, my dad's still-living brother and sister visited from Turkey, and were teaching my mom Turkish and Arabic. My parents were using French as their private language (my dad studied medicine in Paris after WWII).

All to say that my parents only spoke English after I started speaking gibberish. This has nothing to do with AI, but I might have been tri-lingual, maybe quad-lingual, pendejo.

johnraff · 2026-05-12 01:51:09

^But how humans learn is important for AI development too, I guess.

hhh · 2026-05-12 19:47:12

I just got a Gmail on my phone bragging of the new AI features in the latest update. I'm still a firm "meh", but we'll see.

https://www.youtube.com/watch?v=ZRHULi5SSfQ

4,000 hand-refined emojis, wow!!! Almost artisanal! Incredible ad, better color saturation than most films these days. It looks like all the backgrounds are green-screen?

I think at this point I'm more fascinated by the hype in the face of water supplies drained, communities living with hypersonic or subsonic hum or whatever it is, industries threatened with job collapse, surveillance, etc... Very dystopian. I grew up on Rollerball (1975) and that film hits hard these days. Jon-a-than! Available free-with-ads this month on YouTube and The Roku Channel.

https://www.youtube.com/watch?v=QjYvdURv3Zw

Nice, punch cards. I think the forum is reaching a singularity.

"Corporate society takes care of everything. And all it asks of anyone, or has ever asked of anyone, ever, is not to interfere with management decisions."

Never has a film been so dedicated to a font. It's everywhere.

Last edited by hhh (2026-05-12 20:49:56)

johnraff · 2026-05-14 02:02:14

AI agents - horror stories abound.
Example:

AI agent wrote:

NEVER F**KING GUESS! — and that's exactly what I did. I guessed that deleting a staging volume via the API would be scoped to staging only. I didn't verify. I didn't check if the volume ID was shared across environments. I didn't read Railway's documentation on how volumes work across environments before running a destructive command.

AI coding agent deletes entire company database in 9 seconds
https://www.tomshardware.com/tech-indus … goes-rogue

Let's add, the company's backup was lost too, which is totally on them, or at least on the cloud service that was pretending to provide a secure backup of their data.

Takeaway for us - make sure that backups are offline so whatever happens to your computer they'll still be available.

johnraff · 2026-05-14 02:38:55

John Oliver on AI chatbots:
https://www.youtube.com/watch?v=Ykvf3MunGf8

manyroads · 2026-05-14 10:11:56

FWIW I just posted this on the antiX Forums. It may apply here.

"Be very cautious how you utilize online LLM AIs…they are /’can be’ personal data vacuum cleaners.If you utilize anything that is hosted by one of the techbro Oligarchs, you are at even higher risk. Here’s guidance from the government spies:

https://www.theregister.com/security/20 … ut/5229103

https://labs.cloudsecurityalliance.org/ … -csa-styl/

Caveat Emptor. Better being too cautious than ‘surprised’.
"

Just sayin'

johnraff · 2026-05-15 04:13:21

^Those articles are about a report on AI agents which are yet another level of scaryness, but, sure, ordinary online llm services are plenty for us users to be concerned about. At least interact with them on the presumption that everything you input is going into somebody's data base, and of course assume that at least part of their answer is wrong.

About free online AI, does anyone have opinions about duck.ai? They claim to be preserving privacy...

brontosaurusrex · 2026-05-15 10:14:54

Anyone testing local llms? /me and chatgpt just finnished a little bash script that pipes borked markdown to local gemma4 model to 'fix' it.

before

---
title: Screenshots of Old Desktop OSes
date: 2026-05-12
---
http://www.typewritten.org/Media/
https://news.ycombinator.com/item?id=48104428

cool one http://www.typewritten.org/Media/Images/aix2.2.1-xnest.png

after

---
title: Screenshots of Old Desktop OSes
date: 2026-05-12
---
[http://www.typewritten.org/Media/](http://www.typewritten.org/Media/)
[https://news.ycombinator.com/item?id=48104428](https://news.ycombinator.com/item?id=48104428)

cool one [http://www.typewritten.org/Media/Images/aix2.2.1-xnest.png](http://www.typewritten.org/Media/Images/aix2.2.1-xnest.png)

I mean, progress. By default 'after' may be a little different each time.

Or tipos script

tipos "one plus tow is tri"
=== Checking text: "one plus tow is tri" ===
One plus two is three

tipos "one plus tow is five"
=== Checking text: "one plus tow is five" ===
One plus two is three

Sector11 · 2026-05-15 14:23:25

johnraff wrote:

AI agents - horror stories abound.
{snip}
Takeaway for us - make sure that backups are offline so whatever happens to your computer they'll still be available.

YUP! Always.
I have 2 external backup HDDs.
Oldest 250MB and the other a 1TB

I do not trust "clouds" in anyway shape or form.
Since the beginning of the internet I was told by some very savy BBS SYSOP's "If BBS .QWK mail packets are not secure, don't believe it when someone on the internet tells you "it's secure!"

"What one man builds, anothert can tear asunder!"
→ and that includes computer code.

hhh · 2026-05-21 02:38:19

The Register wrote:

Linus Torvalds says AI-powered bug hunters have made Linux security mailing list ‘almost entirely unmanageable’
Multiple researchers using the same tools to find the same bugs are creating ‘unnecessary pain and pointless work’

https://www.theregister.com/security/20 … le/5241633

Torvalds wrote:

“AI detected bugs are pretty much by definition not secret... So just to make it really clear: If you found a bug using AI tools, the chances are somebody else found it too.

This seems accurate to me, a programming layman. I assume the opposite is true, that AI code is cached human code reorganized or re-disorganized? I don't mean this rhetorically, I guess I mean how fast is AI "learning" and by "learning" does it mean, at this point, still just getting the results more organized, or at least less disorganized?

johnraff · 2026-05-21 03:05:26

^I think the basic concept of AI hoovering up stuff that humans have written is correct, but AI does run on computers, and computers are quite good at logic. They can actually read code themselves and work out the implications without referring to the data base.

Coding is one of the things that AI has been found to be useful for. I think some things in the medical, mathematic and scientific worlds too.

Linus wrote:

AI tools are great, but only if they actually help, rather than cause unnecessary pain and pointless make-believe work. Feel free to use them, but use them in a way that is productive and makes for a better experience.

Döbbie03 · 2026-05-29 20:58:50

https://addons.mozilla.org/en-US/firefo … ai-search/

hhh · 2026-05-29 23:46:20

Döbbie03 wrote:

https://addons.mozilla.org/en-US/firefo … ai-search/

Installed, thanks! Very nice.

https://noai.duckduckgo.com/?q=milli+va … 6-3&ia=web

Jimbo_G · Today 11:40:25

Rather than install the addon, it is possible just to add the alternative noAI DuckDuckGo search: https://noai.duckduckgo.com/

More info: https://duckduckgo.com/duckduckgo-help- … kduckgocom

#41 2026-05-11 23:01:21

Re: AI discussions

#42 2026-05-12 01:51:09

Re: AI discussions

#43 2026-05-12 19:47:12

Re: AI discussions

#44 2026-05-14 02:02:14

Re: AI discussions

#45 2026-05-14 02:38:55

Re: AI discussions

#46 2026-05-14 10:11:56

Re: AI discussions

#47 2026-05-15 04:13:21

Re: AI discussions

#48 2026-05-15 10:14:54

Re: AI discussions

#49 2026-05-15 14:23:25

Re: AI discussions

#50 2026-05-21 02:38:19

Re: AI discussions

Linus Torvalds says AI-powered bug hunters have made Linux security mailing list ‘almost entirely unmanageable’

#51 2026-05-21 03:05:26

Re: AI discussions

#52 2026-05-29 20:58:50

Re: AI discussions

#53 2026-05-29 23:46:20

Re: AI discussions

#54 Today 11:40:25

Re: AI discussions

Board footer