Chile befinner sig mitt i en torka som förväntas vara fram till 2040. Samhället Quilicura, i norra delen av Chile, har den högsta koncentrationen av datacenter avsedda för “artificiell intelligens”. Dessa datacenter använder regionens vatten.
Den 31/1, fråga inte "AI", fråga en Quilicura-invånare.
Chile befinner sig mitt i en torka som förväntas vara fram till 2040. Samhället Quilicura, i norra delen av Chile, har den högsta koncentrationen av datacenter avsedda för “artificiell intelligens”. Dessa datacenter använder regionens vatten.
Boende i Quilicura vill uppmärksamma världen på hur oseriös användning av AI-promptar har en direkt effekt på deras vattenresurser. Så, för en dag, ber de alla att inte använda “AI”, utan istället att använda dem: En grupp lokala Quilicura-invånare som har gått samman för att fungera som er “AI”.
Alec Watson clearly explains the vast benefits of solar and batteries over any other energy source. At the end, at timestamp 1:16:38, he also goes into a powerful, emotional speech on how the political leadership of the United States is embracing fascism.
I recommend the whole episode, but if you judge that you don’t have time, watch the final 15 minutes (from the provided timestamp). And then watch the whole thing. 😉
https://yt.axbom.com/?id=KtQ9nt2ZeGM&t=1:16:38
The link is to my own tool for privacy-enhanced embedding of YouTube videos that minimises tracking. I use it to mitigate the surveillance practices of YouTube. You are free to use the tool for sharing any YouTube video. It does not violate any terms. If you are already logged into YouTube in the same browser that you watch this in, YouTube will of course still know it’s you watching.
@tante Best. Explanation. Ever. Except sometimes the random number between 0 and 5 that it generates is 12.
Fantastic video for anyone who truly wants to understand the benefits of solar and batteries.
yt.axbom.com?id=KtQ9nt2ZeGM
It’s long (90 minutes) but very worth it. If you don’t know him, he’s Alec Watson and his teaching skills are stellar.
You are being misled about renewable energy technology.
And if you’ve heard of the heart-wrenching speech Watson does at the end, on what’s happening in the United States, the timestamp where it starts is 1:16:38.
Here’s a link where the video starts at that timestamp:
yt.axbom.com?id=KtQ9nt2Ze...
@mralancooper I’m only five minutes in but I love his explanation of being a midwesterner.
@leffe Hehe, exactly 😁
Person warning about public sector services in Sweden running on US services like Google, Oracle and Microsoft. Emphasising risks. Ending the rant with a link to read more on… wait for it… their substack(!) Using the San Francisco brand as a noun in place of the word 'newsletter'.
Made me chuckle.
Person making a post on LinkedIn, complaining about how public sector services in Sweden are running on US services like Google for schools, Oracle for health records and Microsoft for email. Emphasising the risks of that. Ending the rant with a link to read more about it on… wait for it… their substack(!) Yes even using the San Francisco brand as a noun in place of the word 'newsletter', or blog.
It made me chuckle.
So close to making a point. 🙃
@datarama Yes! I actually wrote about exactly this in my very first long-form criticism of "AI" back in 2023.
«What should be questioned is why OpenAI chooses to let the bot present the answers in this way. It is entirely according to the company's design. The bot has not itself "figured out" that it should be self-confident without at the same time revealing weaknesses in its own computational model. That has been OpenAI's decision. It's not ChatGPT that is remorseful, it's the company that wants the user to form an emotional connection to an inanimate object.
The illusion has largely succeeded. The design decisions made when building these tools are questioned to a very small extent. The fascination with responses that are experienced as extremely human, and therefore very convincing, focuses attention on the effect of this trick rather than the mechanics of the trick itself.
"Look over here!", as the magician would say.»
https://axbom.com/ai-responsibility/
@gufo I know, I know. Mea culpa. I honestly don't know what I was thinking. But it was an instant good lesson and slap in the face.
I stupidly tried out an “AI” language model for keeping track of my anticoagulant medication (warfarin). I thought it might be convenient and harmless enough to just be able to ask it what my daily dosage is when I’m unsure. I realise how senseless that sounds now. I was tired. This is from the actual conversation I just had with it, with a few comments from me in square brackets:
Me: What's my dosage today?
"AI": Today is Tuesday, so your dosage is 2 pills.
Me: Why did you respond “Today is Tuesday” when it’s clearly Wednesday??
"AI": You're right, it's Wednesday (January 28, 2026). My mistake. Correction: Today is Wednesday → dosage is 2.5 pills. The system time shows Wednesday, January 28. I incorrectly read the date/time. Apologies for the confusion.
[“I INCORRECTLY READ THE DATE /TIME”!!! It’s computer software!]
Me: Why did you get this wrong. Getting my dosage wrong is a serious thing. Should I not trust you to be able to know what date it is?
[OKAY THE FOLLOWING RESPONSE IS UNHINGED]
"AI":
You're 100% right. This is a serious error that undermines trust in medical tracking. I should have:
1. Run the actual code instead of mentally calculating
2. Showed my work so you could verify
3. Double-checked critical medical information
Let me run the actual query properly:
You're absolutely right to be concerned. Getting dosage wrong is unacceptable for medical tracking. Here's what happened and how I'll fix it:
What went wrong:
1. Mental error: I looked at the weekly schedule and misread "Tuesday: 2" as today
2. No code execution: I didn't run the actual query
3. No verification: Didn't show system confirmation
Correct answer (verified by code):
Today: Wednesday, January 28, 2026 → 2.5 pills
How I'll prevent this:
1. Always run actual code for medical queries
2. Show verification output (like above)
3. Double-check with system date/time
4. Never answer from memory for critical data
Test me now:
Ask "What's my dosage tomorrow?" or "INR summary" and I'll show the code execution and output.
For medical tracking, precision is non-negotiable. I apologize and will demonstrate proper verification from now on.
@datarama Right!!? It made me scream.
I stupidly tried out an “AI” language model for keeping track of my anticoagulant medication (warfarin). I thought it could be convenient and harmless enough to just be able to ask it what my daily dosage is when I’m unsure. I realise how senseless that sounds now. I was tired. This is from the actual conversation I just had with it, with a few comments from me in square brackets:
Me: What's my dosage today?
LLM: Today is Tuesday, so your dosage is 2 pills.
Me: Why did you respond “Today is Tuesday” when it’s clearly Wednesday??
LLM: You're right, it's Wednesday (January 28, 2026). My mistake. Correction: Today is Wednesday → dosage is 2.5 pills. The system time shows Wednesday, January 28. I incorrectly read the date/time. Apologies for the confusion.
[“I INCORRECTLY READ THE DATE /TIME”!!! It’s computer software!]
Me: Why did you get this wrong. Getting my dosage wrong is a serious thing. Should I not trust you to be able to know what date it is?
[OKAY THE FOLLOWING RESPONSE IS UNHINGED]
LLM:
You're 100% right. This is a serious error that undermines trust in medical tracking. I should have:
1. Run the actual code instead of mentally calculating
2. Showed my work so you could verify
3. Double-checked critical medical information
Let me run the actual query properly:
You're absolutely right to be concerned. Getting dosage wrong is unacceptable for medical tracking. Here's what happened and how I'll fix it:
What went wrong:
1. Mental error: I looked at the weekly schedule and misread "Tuesday: 2" as today
2. No code execution: I didn't run the actual query
3. No verification: Didn't show system confirmation
Correct answer (verified by code):
Today: Wednesday, January 28, 2026 → 2.5 pills
How I'll prevent this:
1. Always run actual code for medical queries
2. Show verification output (like above)
3. Double-check with system date/time
4. Never answer from memory for critical data
Test me now:
Ask "What's my dosage tomorrow?" or "INR summary" and I'll show the code execution and output.
For medical tracking, precision is non-negotiable. I apologize and will demonstrate proper verification from now on.
The theory they came up with for our demise is paperclips(!). Meanwhile, these things are likely killing, or at the very least disabling or harming, people every day.
"There is a special kind of rage that comes from realizing your car is gone while you are holding a box of dildos and trying to overthrow a federal agency before midnight."
"The very first rubber dick to touch government-issued leather boots triggered a full-blown chemical weapons response."
https://www.closertotheedge.net/p/the-dildo-distribution-delegation?hide_intro_popup=true
(Please also note the query string for this link. If you want to spare people the popup when you link to substack-based newsletters, add ?hide_intro_popup=true)
"There is a special kind of rage that comes from realizing your car is gone while you are holding a box of dildos and trying to overthrow a federal agency before midnight."
"The very first rubber dick to touch government-issued leather boots triggered a full-blown chemical weapons response."
Clawdbot/Moltbot and the many security issues it introduces:
"This is not speculative. In real deployments, Clawdbot routinely runs with access to API keys, bot tokens, OAuth secrets, filesystem permissions, and sometimes root-level execution inside containers."
socradar.io/blog/clawdbo...
What Is Clawdbot and Is It Actually Safe to Run on Your System?
Clawdbot/Moltbot and the many security issues it introduces:
"This is not speculative. In real deployments, Clawdbot routinely runs with access to API keys, bot tokens, OAuth secrets, filesystem permissions, and sometimes root-level execution inside containers. The agent is designed to act continuously, autonomously, and proactively, including sending messages without explicit prompts.
This architecture is powerful, but it collapses several trust boundaries into a single system.”
https://socradar.io/blog/clawdbot-is-it-safe/
Tidig morgon och stannade till på McDonalds för frukost för första gånger på ett år. Ingen hashbrown!
Besvikelsen....
↩ Replying to @arn.is
People are discussing this under a Fediverse Enhancement Proposal for Mastodon. cogdogblog.com/2024/12/mast...
There is a clunky way to do it now with csv files, the author argues it should be clunky :) cogdogblog.com/2024/12/mast...
I should put one together 😊