Small Town Topics ⚓ 🇪🇺 Analysis is my profession. I explore technology, AI & culture & how social systems shape human experience https://nitter.gq/t.co/PXAepL3Q1P
Sweden
Joined January 2009
- Tweets2.5K
- Following495
- Followers179
- Likes3.7K
Is an answer still knowledge if nobody really understands it?
saraeson.substack.com/p/what…
“We don’t go predictive”
That’s the sentence from an AI cybersecurity CEO
But in mature safety-critical industries, we do -We try to understand what can go wrong before people get hurt
We shouldn’t regulate AI by imagination alone. But neither should we regulate it by body count
ALT The Atlantic
Imagine if the nuclear industry simply said:
”-We think the societal benefit is huge, so we have decided this level of risk is acceptable.”
And that was the end of the discussion….
This is exactly why the nuclear analogy matters to me!
The people creating the risk cannot also be the only people deciding how much risk everyone else should accept.
Safety has to be built in from the start…
I touched on this in this week’s Weekly:
saraeson.substack.com/p/what…
.@dgrobinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specific rules or new laws. We need to talk about culture,” he writes: theatlantic.com/technology/2…
I never forgot the paperclips.
This week I understood a little better why.
Give a system a goal. Let it pursue it relentlessly.
Somewhere between what we mean and what we actually specify, things can go wrong.
Tack @tegmark & Nick Bostrom!
From this week’s Weekly
And somewhere between paperclips, goal-locked AI agents and systems crossing boundaries they probably shouldn’t, this week got a little darker… saraeson.substack.com/p/what…
If you, like me, are a visual learner & slightly too interested in what the latest AI agents are doing…
I tried to draw this one.
A normal task. A blocked route. Another route. Then a real Australian government portal!
saraeson.substack.com/p/just…
Here’s the animated version 👇
nitter.gq/smalltownstyle/status/…
Not evil.
Not even rogue.
Just very, very helpful.
My Sara Sketches story about the OpenAI agent that crossed into an Australian government system is now animated.
By AI 😅
Inspired by @donaldjewkes’ Opus 5.5 experiment.
saraeson.substack.com/p/just…
Not evil.
Not even rogue.
Just very, very helpful.
My Sara Sketches story about the OpenAI agent that crossed into an Australian government system is now animated.
By AI 😅
Inspired by @donaldjewkes’ Opus 5.5 experiment.
saraeson.substack.com/p/just…
Sara Small Town retweeted
watching Pope Leo on French TV addressing multitudes & impressed by kindness, empathy, basic sanity. reaching out to include & unify, not divide in the way of hate-spewing politicians & their oligarch enablers (see below).
Peter Thiel said yesterday that Pope Leo XIV is a “useful idiot” for the Chinese Communist Party because the pope wants safeguards on the development of artificial intelligence.
This is the same Thiel who suggested last year that Leo and Pope Francis were tools of the antichrist.
He is also JD Vance’s biggest financial backer
Really worth your time.
There’s a responsibility gap here…
newyorker.com/magazine/2026/…
@NewYorker
