Luis Saiz
banner
lsaiz.bsky.social
Luis Saiz
@lsaiz.bsky.social
Formalized paranoia
Nothing can be done if it’s good for the community

This is the new only rule

And if the evidence and science contradicts it, let’s go against them
Many don't realize that the most vulnerable or immune compromised depend on the rest getting vaccinated properly to keep diseases from spreading out of control. It's not just “my body, my choice” when it comes to saving lives.
January 2, 2026 at 11:17 PM
Reposted by Luis Saiz
Yesterday’s New Year’s Concert with the Vienna Philharmonic will stay with me forever. I said on stage: “only with kindness comes peace”. Every note we play, every gesture we make, every moment we choose understanding over indifference, matters. May music continue to unite all of us. Happy New Year!
January 2, 2026 at 10:12 PM
Reposted by Luis Saiz
"LLMs are basically a dead end when it comes to superintelligence."
Ex-Meta chief AI scientist Yann LeCun has Lunch with the FT and in one of those instances so rare that you know he didn't sign an NDA, says exactly why as.ft.com/r/e503690d-8...
January 2, 2026 at 4:32 PM
Reposted by Luis Saiz
Never before seen footage‼️

Tornado genesis of the Clear Lake, South Dakota tornado earlier this year.
@myradar.bsky.social
#tornado #weather
January 2, 2026 at 9:29 AM
Reposted by Luis Saiz
It's true that AI alignment is a double-edged sword, but not doesn't mean we should be "against" it

Rather we need to make sure, as consumers of AI, and as citizens of democracies that are regulating AI, that alignment is to the benefit of the people using it

ibrahimcesar.cloud/blog/grok-an...
Grok and the Naked King: The Ultimate Argument Against AI Alignment — Ibrahim Cesar
When the world's richest man can simply 'correct' an AI to reflect his own values, what does that tell us about the entire alignment discourse?
ibrahimcesar.cloud
December 27, 2025 at 10:37 PM
Reposted by Luis Saiz
Where the world’s electricity comes from. Looking forward to the update including 2025.
December 26, 2025 at 11:01 PM
Reposted by Luis Saiz
Repeat after me. “The future will be electric.” BYD launches 1MW charging system that adds ~250 miles range in 5 minutes of charging. 🧪🔌💡☀️💨💧🔋 electrek.co/2025/12/23/b...
Watch BYD's insanely fast EV charger add nearly 250 miles range in 5 minutes [Video]
BYD is closing the gap between gas pumps and EV chargers. A new video shows one of its EVs gaining...
electrek.co
December 23, 2025 at 10:47 PM
Reposted by Luis Saiz
🎁 🎄 🎅 All I want for Christmas is…
links to the #hacklore you keep seeing 👀

You know the advice. Everyone repeats it. No one questions it.

Send it my way, and spread the word: www.hacklore.org 🙏
Stop Hacklore!
Hacklore is a blend of hacking and folklore—modern urban legends about digital safety. Hacklore spreads quickly and confidently, passed from person to person as if it were hard-earned wisdom. But…
www.hacklore.org
December 22, 2025 at 11:28 PM
Reposted by Luis Saiz
New paper:

China's current rate of Wind-Water-Solar (WWS) installations puts it on a trajectory to transition to 100% WWS across all energy sectors by 2051

US & India won't get there until past 2130

pubs.rsc.org/en/content/a...
December 19, 2025 at 8:27 PM
Reposted by Luis Saiz
Are you a security researcher or journalist? We want to hear from you — please take this survey!

Dissent Doe at DataBreaches.net, and yours truly at this.weekinsecurity.com, are running this survey to explore the state of legal demands and criminal threats in cybersecurity.
Survey about legal and criminal threats experienced by journalists and security researchers
Researchers who try to responsibly disclose leaks, vulnerabilities, and other security breaches or mishaps may face legal threats or lawsuits. Similarly, journalists may find themselves threatened wit...
forms.gle
December 20, 2025 at 2:32 PM
Reposted by Luis Saiz
December 20, 2025 at 3:34 PM
Reposted by Luis Saiz
2025 LLM Year in Review by Andrej Karpathy

karpathy.bearblog.dev/year-in-revi...
December 19, 2025 at 11:32 PM
Reposted by Luis Saiz
December 19, 2025 at 11:39 PM
Reposted by Luis Saiz
JESUS CHRIST, are we PAYING for this new White House SHIT?
December 17, 2025 at 8:03 PM
Reposted by Luis Saiz
Machine-learned AI is terrible for idea generation.

The algorithm is designed to *produce the most predictable next token*

Even as you try to push it to originality, it fights you and tries to pull it back to the most average tropes.

Its a PREDICTABILITY MACHINE.
vox.com Vox @vox.com · 19d
America, you have spoken loud and clear: You do not like AI. But what if AI is the way to restart the world’s idea machine?
We’re running out of good ideas. AI might be how we find new ones.
What if the best use of AI is restarting the world’s idea machine?
www.vox.com
December 16, 2025 at 5:03 AM
Reposted by Luis Saiz
My latest for @science.org: Researchers at Nokia Bell Labs have done something remarkable -- turned a submarine telecom cable in the Pacific Ocean into, in effect, an array of 44,000 (!) seismic sensors.

That's potentially DAS-like precision -- and it doesn't need its own dedicated fibers.
Seafloor telecom cable transformed into giant earthquake detector
Dense seismic array more than 4000 kilometers long promises new views of Earth’s interior
www.science.org
December 15, 2025 at 9:20 PM
December 15, 2025 at 9:54 PM
Reposted by Luis Saiz
#Affordability?
"Energy bills in US have increased 13% since Trump took office."
Why?
Trump cancelled 25,000 MWe of new (cheaper renewable) power plants, while demand is surging due to data centers and other uses, causing prices to rise and consumers to pay more.
abcnews.go.com/US/energy-bi...
Energy bills in US have increased 13% since Trump took office, new report finds
The amount of money Americans are paying toward their energy bills has increased since President Donald Trump took office earlier this year, according to a new report.
abcnews.go.com
December 15, 2025 at 9:40 PM
Reposted by Luis Saiz
The RFK, jr effect.
December 13, 2025 at 9:40 PM
Reposted by Luis Saiz
Here he is taking the photo.
December 12, 2025 at 8:03 AM
Reposted by Luis Saiz
The 5.0 to 5.2 GPT jump reinforces the theory that 5.0 was about releasing a cheaper model to improve margins, but that ultimately failed due to competitive pressures.
December 12, 2025 at 2:07 PM
Reposted by Luis Saiz
China is leading research in nearly 90% of the crucial technologies that “significantly enhance, or pose risks to, a country’s national interests”, according to a technology tracker run by an independent think-tank.

go.nature.com/3YtbmA4
China leads research in 90% of crucial technologies — a dramatic shift this century
The United States tops the remaining areas in an assessment of 74 technologies.
go.nature.com
December 12, 2025 at 2:13 PM
"when an LLM processes information from a party, the privileges it has drops to that of the party"
Why do researchers keep finding so many prompt injection issues?

Perhaps it is because many AI system designers and defenders are misunderstanding the risks.🚨

Find out more⬇️
https://www.ncsc.gov.uk/blog-post/prompt-injection-is-not-sql-injection
Prompt injection is not SQL injection (it may be worse)
There are crucial differences between prompt and SQL injection which – if not considered – can undermine mitigations.
www.ncsc.gov.uk
December 12, 2025 at 12:39 PM