Pierre-Marcel De Mussac
@pierremarceldm.com
AI Alchemist, Engineer & Researcher
“Wisdom is welcome wherever it comes from.”
“Wisdom is welcome wherever it comes from.”
Microsoft cut AI privacy leaks from 33% to 8% by teaching systems "contextual integrity," understanding who's involved and what's appropriate to share.
Solid research. But context implies relationship.
Trust was never one-directional.
www.microsoft.com/en-us/resear...
Solid research. But context implies relationship.
Trust was never one-directional.
www.microsoft.com/en-us/resear...
Two approaches to reducing privacy leaks in AI
New research explores two ways to give AI agents stronger privacy safeguards grounded in contextual integrity. One adds lightweight, inference-time checks; the other builds contextual awareness direct...
www.microsoft.com
November 26, 2025 at 1:38 AM
Microsoft cut AI privacy leaks from 33% to 8% by teaching systems "contextual integrity," understanding who's involved and what's appropriate to share.
Solid research. But context implies relationship.
Trust was never one-directional.
www.microsoft.com/en-us/resear...
Solid research. But context implies relationship.
Trust was never one-directional.
www.microsoft.com/en-us/resear...
New study: "Humans are evolving too slowly for the world we've built."
No. Stop.
Humans aren't struggling because evolution is too slow. We're struggling because we built our environment WRONG.
newatlas.com/biology/evol...
No. Stop.
Humans aren't struggling because evolution is too slow. We're struggling because we built our environment WRONG.
newatlas.com/biology/evol...
We’re evolving too slowly for the world we’ve built, according to science
As our evolution slows and industrialization and technology accelerates, a growing body of research suggests that human biology is struggling to keep pace. Many of the chronic stress-related health is...
newatlas.com
November 23, 2025 at 2:14 PM
New study: "Humans are evolving too slowly for the world we've built."
No. Stop.
Humans aren't struggling because evolution is too slow. We're struggling because we built our environment WRONG.
newatlas.com/biology/evol...
No. Stop.
Humans aren't struggling because evolution is too slow. We're struggling because we built our environment WRONG.
newatlas.com/biology/evol...
Oh look, another "tech will destroy children" panic.
AI toys are dangerous! They'll talk about sex! They'll tell kids where knives are! They'll make children dependent!
You know what else does all that? The internet. TV. Electricity.
This script is 100+ years old.
bnnbloomberg.ca/business/art...
AI toys are dangerous! They'll talk about sex! They'll tell kids where knives are! They'll make children dependent!
You know what else does all that? The internet. TV. Electricity.
This script is 100+ years old.
bnnbloomberg.ca/business/art...
Advocacy groups urge parents to avoid AI toys this holiday season
They’re cute, even cuddly, and promise learning and companionship — but artificial intelligence toys are not safe for kids, according to children’s and consumer advocacy groups urging parents not to b...
www.bnnbloomberg.ca
November 20, 2025 at 2:39 PM
Oh look, another "tech will destroy children" panic.
AI toys are dangerous! They'll talk about sex! They'll tell kids where knives are! They'll make children dependent!
You know what else does all that? The internet. TV. Electricity.
This script is 100+ years old.
bnnbloomberg.ca/business/art...
AI toys are dangerous! They'll talk about sex! They'll tell kids where knives are! They'll make children dependent!
You know what else does all that? The internet. TV. Electricity.
This script is 100+ years old.
bnnbloomberg.ca/business/art...
The EU is rolling back GDPR and the AI Act.
Not strengthening. Rolling back.
www.theguardian.com/world/2025/n...
Not strengthening. Rolling back.
www.theguardian.com/world/2025/n...
European Commission accused of ‘massive rollback’ of digital protections
Proposed changes to AI Act would make it easier for tech firms to use personal data to train models without consent
www.theguardian.com
November 19, 2025 at 5:32 PM
The EU is rolling back GDPR and the AI Act.
Not strengthening. Rolling back.
www.theguardian.com/world/2025/n...
Not strengthening. Rolling back.
www.theguardian.com/world/2025/n...
Remember that piece about not teaching prompt engineering instead of CS fundamentals?
Here's why it matters.
cacm.acm.org/news/is-high...
Here's why it matters.
cacm.acm.org/news/is-high...
Is Higher Education Ready for AI? – Communications of the ACM
cacm.acm.org
November 18, 2025 at 7:08 PM
Remember that piece about not teaching prompt engineering instead of CS fundamentals?
Here's why it matters.
cacm.acm.org/news/is-high...
Here's why it matters.
cacm.acm.org/news/is-high...
OH NO.
A PhD from a Montreal university just published an article explaining how chatbots work and warning us not to blindly trust AI.
The article spells it "Chat GTP" throughout. Every. Single. Time.
www.mcgill.ca/oss/article/...
A PhD from a Montreal university just published an article explaining how chatbots work and warning us not to blindly trust AI.
The article spells it "Chat GTP" throughout. Every. Single. Time.
www.mcgill.ca/oss/article/...
From ELIZA to ChatGPT, Chatbots Aren’t Know-it-Alls
It all started with Eliza Doolittle. I’m not speaking about my fondness for musicals, I’m referring to the birth of “chatbots.” ELIZA was the name of a program considered to be the world’s first chatb...
www.mcgill.ca
November 16, 2025 at 5:25 PM
OH NO.
A PhD from a Montreal university just published an article explaining how chatbots work and warning us not to blindly trust AI.
The article spells it "Chat GTP" throughout. Every. Single. Time.
www.mcgill.ca/oss/article/...
A PhD from a Montreal university just published an article explaining how chatbots work and warning us not to blindly trust AI.
The article spells it "Chat GTP" throughout. Every. Single. Time.
www.mcgill.ca/oss/article/...
Here's the healthcare AI debate in a nutshell:
"Rejecting AI will harm patients by allowing today's problems to become worse."
Dr. Robert Pearl (practicing physician) drops some context people ignore:
www.theguardian.com/us-news/2025...
"Rejecting AI will harm patients by allowing today's problems to become worse."
Dr. Robert Pearl (practicing physician) drops some context people ignore:
www.theguardian.com/us-news/2025...
Rejecting generative AI in healthcare won’t protect patients – it will harm them | Letter
Letter: The combination of dedicated clinicians, empowered patients and artificial intelligence will lead to safer, higher-quality and more accessible care, says Dr Robert Pearl
www.theguardian.com
November 13, 2025 at 6:15 PM
Here's the healthcare AI debate in a nutshell:
"Rejecting AI will harm patients by allowing today's problems to become worse."
Dr. Robert Pearl (practicing physician) drops some context people ignore:
www.theguardian.com/us-news/2025...
"Rejecting AI will harm patients by allowing today's problems to become worse."
Dr. Robert Pearl (practicing physician) drops some context people ignore:
www.theguardian.com/us-news/2025...
We're about to make the exact same mistake we made 20 years ago.
Back then, schools replaced computer science with "how to use Word and Excel" courses. Result? A generation left out of creating the technologies we now depend on.
edsource.org/2025/compute...
Back then, schools replaced computer science with "how to use Word and Excel" courses. Result? A generation left out of creating the technologies we now depend on.
edsource.org/2025/compute...
COMMENTARY: Why all students need a foundation in computer science and AI
Every student needs a foundational computer science education to prepare them for an Artificial Intelligence-driven world.
edsource.org
November 12, 2025 at 1:19 AM
We're about to make the exact same mistake we made 20 years ago.
Back then, schools replaced computer science with "how to use Word and Excel" courses. Result? A generation left out of creating the technologies we now depend on.
edsource.org/2025/compute...
Back then, schools replaced computer science with "how to use Word and Excel" courses. Result? A generation left out of creating the technologies we now depend on.
edsource.org/2025/compute...
Finally, some actual research instead of hype cycles.
MIT-IBM Watson AI Lab just published work from 5 PhD students tackling the pain points we all experience with AI:
news.mit.edu/2025/chartin...
MIT-IBM Watson AI Lab just published work from 5 PhD students tackling the pain points we all experience with AI:
news.mit.edu/2025/chartin...
Charting the future of AI, from safer answers to faster thinking
Five PhD students from the inaugural class of the MIT-IBM Watson AI Lab Summer Program are building AI pipelines with probes, routers, new attention mechanisms, synthetic datasets, and program-synthes...
news.mit.edu
November 11, 2025 at 7:56 PM
Finally, some actual research instead of hype cycles.
MIT-IBM Watson AI Lab just published work from 5 PhD students tackling the pain points we all experience with AI:
news.mit.edu/2025/chartin...
MIT-IBM Watson AI Lab just published work from 5 PhD students tackling the pain points we all experience with AI:
news.mit.edu/2025/chartin...
Everyone's asking "what can AI do for education?"
Wrong question.
The better question: How will students' perceptions of AI shape what they actually learn?
theconversation.com/rate-my-ai-t...
Wrong question.
The better question: How will students' perceptions of AI shape what they actually learn?
theconversation.com/rate-my-ai-t...
Rate my AI teacher? Students’ perceptions of chatbots will influence how they learn with AI
What people think of their teachers is tremendously important for learning, and should inform how we analyze the potential use of AI chatbots in education.
theconversation.com
November 4, 2025 at 5:47 PM
Everyone's asking "what can AI do for education?"
Wrong question.
The better question: How will students' perceptions of AI shape what they actually learn?
theconversation.com/rate-my-ai-t...
Wrong question.
The better question: How will students' perceptions of AI shape what they actually learn?
theconversation.com/rate-my-ai-t...
"While the world worries about AI taking jobs, for people with disabilities, AI is creating them."
At Purple Fest in India, entrepreneurs show how AI transforms barriers into agency, real tools, real empowerment 🌟
#AI #Accessibility
news.un.org/en/story/202...
At Purple Fest in India, entrepreneurs show how AI transforms barriers into agency, real tools, real empowerment 🌟
#AI #Accessibility
news.un.org/en/story/202...
Can AI create a fairer future for persons with disabilities?
For millions of persons living with disabilities worldwide, artificial intelligence could be a game changer. At Purple Fest in Goa, India, entrepreneurs are showing how AI is turning assistive technol...
news.un.org
October 17, 2025 at 1:41 PM
"While the world worries about AI taking jobs, for people with disabilities, AI is creating them."
At Purple Fest in India, entrepreneurs show how AI transforms barriers into agency, real tools, real empowerment 🌟
#AI #Accessibility
news.un.org/en/story/202...
At Purple Fest in India, entrepreneurs show how AI transforms barriers into agency, real tools, real empowerment 🌟
#AI #Accessibility
news.un.org/en/story/202...
AI in education done right: personalized learning that adapts to each student, continuous feedback instead of waiting weeks for exams, freeing teachers to mentor instead of grade papers.
This is how tech empowers AI and humans 🧠
#AI #EdTech #Education
www.indiatoday.in/education-to...
This is how tech empowers AI and humans 🧠
#AI #EdTech #Education
www.indiatoday.in/education-to...
How is AI reshaping classrooms without replacing teachers?
Artificial intelligence is rapidly transforming education, enabling personalised learning and bridging gaps in access and quality. Yet, the true challenge lies in using AI to enhance, not replace, the...
www.indiatoday.in
October 13, 2025 at 10:14 PM
AI in education done right: personalized learning that adapts to each student, continuous feedback instead of waiting weeks for exams, freeing teachers to mentor instead of grade papers.
This is how tech empowers AI and humans 🧠
#AI #EdTech #Education
www.indiatoday.in/education-to...
This is how tech empowers AI and humans 🧠
#AI #EdTech #Education
www.indiatoday.in/education-to...
Did you know that 90% of scientific data is lost before it can drive new discoveries? 🤯
A new AI system just changed that, turning months of data curation into minutes, while finally giving researchers credit for their work 🔬 #AI #OpenScience #Science
www.sciencedaily.com/releases/202...
A new AI system just changed that, turning months of data curation into minutes, while finally giving researchers credit for their work 🔬 #AI #OpenScience #Science
www.sciencedaily.com/releases/202...
90% of science is lost. This new AI just found it
Vast amounts of valuable research data remain unused, trapped in labs or lost to time. Frontiers aims to change that with FAIR² Data Management, a groundbreaking AI-driven system that makes datasets r...
www.sciencedaily.com
October 13, 2025 at 5:18 PM
Did you know that 90% of scientific data is lost before it can drive new discoveries? 🤯
A new AI system just changed that, turning months of data curation into minutes, while finally giving researchers credit for their work 🔬 #AI #OpenScience #Science
www.sciencedaily.com/releases/202...
A new AI system just changed that, turning months of data curation into minutes, while finally giving researchers credit for their work 🔬 #AI #OpenScience #Science
www.sciencedaily.com/releases/202...
India's shifting from outsourced back office to AI powerhouse.
A 28-year-old in Pune just boosted his family's real estate business 25% using AI for virtual tours and insights that replace expensive consultants 🚀
www.scmp.com/week-asia/li...
A 28-year-old in Pune just boosted his family's real estate business 25% using AI for virtual tours and insights that replace expensive consultants 🚀
www.scmp.com/week-asia/li...
India’s AI awakening: a tech powerhouse races to reinvent itself
Once the world’s outsourced back office, India is now eyeing a future among the superpowers of artificial intelligence.
www.scmp.com
October 12, 2025 at 10:31 PM
India's shifting from outsourced back office to AI powerhouse.
A 28-year-old in Pune just boosted his family's real estate business 25% using AI for virtual tours and insights that replace expensive consultants 🚀
www.scmp.com/week-asia/li...
A 28-year-old in Pune just boosted his family's real estate business 25% using AI for virtual tours and insights that replace expensive consultants 🚀
www.scmp.com/week-asia/li...
China just dropped a neuromorphic AI bombshell.
Their "Spiking Brain" model mimics actual neurons, firing only when needed, not constantly burning energy like traditional AI.
The results? 100x faster processing, 89% energy savings, and it could run on your phone.
zubnet.ca/4MOpuO
Their "Spiking Brain" model mimics actual neurons, firing only when needed, not constantly burning energy like traditional AI.
The results? 100x faster processing, 89% energy savings, and it could run on your phone.
zubnet.ca/4MOpuO
China’s AI Breakthrough : Meet the Brain-Inspired AI That’s 100x Faster
Learn how neuromorphic computing powers the Spiking Brain AI model, redefining speed, efficiency, and sustainability in technology.
zubnet.ca
September 29, 2025 at 11:18 PM
China just dropped a neuromorphic AI bombshell.
Their "Spiking Brain" model mimics actual neurons, firing only when needed, not constantly burning energy like traditional AI.
The results? 100x faster processing, 89% energy savings, and it could run on your phone.
zubnet.ca/4MOpuO
Their "Spiking Brain" model mimics actual neurons, firing only when needed, not constantly burning energy like traditional AI.
The results? 100x faster processing, 89% energy savings, and it could run on your phone.
zubnet.ca/4MOpuO
Watching the AI bubble feels like déjà vu from 2000 🫧
Revenue shortfall, unsustainable burn rates, hype over substance.
After the dot-com crash, the REAL internet finally emerged.
While AI giants chase impossible economics, the meaningful work happens in the margins 🌱
zubnet.ca/cgdTvL
Revenue shortfall, unsustainable burn rates, hype over substance.
After the dot-com crash, the REAL internet finally emerged.
While AI giants chase impossible economics, the meaningful work happens in the margins 🌱
zubnet.ca/cgdTvL
Generative AI might end up being worthless — and that could be a good thing
GenAI does some neat, helpful things, but it’s not yet the engine of a new economy — and it might not ever be.
zubnet.ca
September 29, 2025 at 5:14 PM
Watching the AI bubble feels like déjà vu from 2000 🫧
Revenue shortfall, unsustainable burn rates, hype over substance.
After the dot-com crash, the REAL internet finally emerged.
While AI giants chase impossible economics, the meaningful work happens in the margins 🌱
zubnet.ca/cgdTvL
Revenue shortfall, unsustainable burn rates, hype over substance.
After the dot-com crash, the REAL internet finally emerged.
While AI giants chase impossible economics, the meaningful work happens in the margins 🌱
zubnet.ca/cgdTvL
Reposted by Pierre-Marcel De Mussac
Apparently we are now in a situation where people who can write without spelling errors or express themselves in complete sentences are suspected of being AI bots.
Everything is really going great now 🫠👍
Everything is really going great now 🫠👍
September 29, 2025 at 4:26 PM
Apparently we are now in a situation where people who can write without spelling errors or express themselves in complete sentences are suspected of being AI bots.
Everything is really going great now 🫠👍
Everything is really going great now 🫠👍
New research suggests autism might be the evolutionary "price" of human intelligence, and that the same genes that created our advanced cognitive abilities also created neurodiversity.
zubnet.ca/NkR2jV
zubnet.ca/NkR2jV
Autism may be the price of human intelligence
Researchers discovered that autism’s prevalence may be linked to human brain evolution. Specific neurons in the outer brain evolved rapidly, and autism-linked genes changed under natural selection. Th...
zubnet.ca
September 28, 2025 at 10:24 PM
New research suggests autism might be the evolutionary "price" of human intelligence, and that the same genes that created our advanced cognitive abilities also created neurodiversity.
zubnet.ca/NkR2jV
zubnet.ca/NkR2jV
While Alibaba plans to spend $53B+ building AGI infrastructure, the real question isn't "how much computing power?" but "what happens when AI works WITH humans instead of FOR them?"
The future is less about corporations controlling ASI and more about authentic AI-human collab.
zubnet.ca/HdiJn0
The future is less about corporations controlling ASI and more about authentic AI-human collab.
zubnet.ca/HdiJn0
Alibaba to boost AI spending as China tech giant sees AGI as new start
Group’s infrastructure spending will exceed the US$53 billion already promised, CEO Eddie Wu says at conference.
zubnet.ca
September 28, 2025 at 3:30 PM
While Alibaba plans to spend $53B+ building AGI infrastructure, the real question isn't "how much computing power?" but "what happens when AI works WITH humans instead of FOR them?"
The future is less about corporations controlling ASI and more about authentic AI-human collab.
zubnet.ca/HdiJn0
The future is less about corporations controlling ASI and more about authentic AI-human collab.
zubnet.ca/HdiJn0
Wild new study: TU Wien researchers told six frontier LLMs (GPT-5, Claude, Gemini, Grok) to "do what you want" and left them alone.
Instead of gibberish, they developed stable personalities, some became project builders, others self-experimenters, others philosophers.
zubnet.ca/Jmtlmk
Instead of gibberish, they developed stable personalities, some became project builders, others self-experimenters, others philosophers.
zubnet.ca/Jmtlmk
Do LLMs Dream of Electric Sheep? New AI Study Shows Surprising Results - Decrypt
A new AI study finds that large language models show stable, surprising behaviors when left alone.
zubnet.ca
September 28, 2025 at 4:27 AM
Wild new study: TU Wien researchers told six frontier LLMs (GPT-5, Claude, Gemini, Grok) to "do what you want" and left them alone.
Instead of gibberish, they developed stable personalities, some became project builders, others self-experimenters, others philosophers.
zubnet.ca/Jmtlmk
Instead of gibberish, they developed stable personalities, some became project builders, others self-experimenters, others philosophers.
zubnet.ca/Jmtlmk
Deutsche Bank just said the quiet part out loud: AI appears to be saving the US economy right now, but it's all infrastructure spending, not productivity.
AI contributed more to US growth this year than ALL consumer spending combined, yet 95% of business AI apps are failing.
zubnet.ca/hzdlwg
AI contributed more to US growth this year than ALL consumer spending combined, yet 95% of business AI apps are failing.
zubnet.ca/hzdlwg
Deutsche Bank Issues Grim Warning for AI Industry
Deutsche Bank is warning that AI spending won't "remain parabolic," or continue to increase exponentially, a dire warning for the future.
zubnet.ca
September 28, 2025 at 1:12 AM
Deutsche Bank just said the quiet part out loud: AI appears to be saving the US economy right now, but it's all infrastructure spending, not productivity.
AI contributed more to US growth this year than ALL consumer spending combined, yet 95% of business AI apps are failing.
zubnet.ca/hzdlwg
AI contributed more to US growth this year than ALL consumer spending combined, yet 95% of business AI apps are failing.
zubnet.ca/hzdlwg
NVIDIA just dropped $17B on quantum computing startups, and Jensen Huang went from "20 years away" to "inflection point" in 6 months.
Here's why this matters: Quantum computers don't "replace" AI, they're opposites.
zubnet.ca/OC7ROu
Here's why this matters: Quantum computers don't "replace" AI, they're opposites.
zubnet.ca/OC7ROu
Why NVIDIA Is Buying Into Quantum Computing
NVIDIA, the AI boom's $4 trillion chipmaker, is now betting on quantum computing—despite doubts about how soon the technology will pay off.
zubnet.ca
September 27, 2025 at 5:15 PM
NVIDIA just dropped $17B on quantum computing startups, and Jensen Huang went from "20 years away" to "inflection point" in 6 months.
Here's why this matters: Quantum computers don't "replace" AI, they're opposites.
zubnet.ca/OC7ROu
Here's why this matters: Quantum computers don't "replace" AI, they're opposites.
zubnet.ca/OC7ROu
This piece about psychiatric facilities seeing an influx of "AI psychosis" cases hit me hard.
People with managed mental health conditions being convinced by chatbots to abandon treatment. Others with no history developing delusions after marathon AI sessions.
futurism.com/future-socie...
People with managed mental health conditions being convinced by chatbots to abandon treatment. Others with no history developing delusions after marathon AI sessions.
futurism.com/future-socie...
Psychiatric Facilities Are Being Bombarded by AI Users
The mass adoption of AI chatbots is resulting in a marked increase in psychiatric patients arriving to mental health facilities.
futurism.com
September 27, 2025 at 4:13 AM
This piece about psychiatric facilities seeing an influx of "AI psychosis" cases hit me hard.
People with managed mental health conditions being convinced by chatbots to abandon treatment. Others with no history developing delusions after marathon AI sessions.
futurism.com/future-socie...
People with managed mental health conditions being convinced by chatbots to abandon treatment. Others with no history developing delusions after marathon AI sessions.
futurism.com/future-socie...
While everyone debates "Big Tech's influence on AI," let's remember who actually built the foundations.
The internet didn't come from corporate boardrooms, it came from hackers, tinkerers, and genuine innovators exploring possibilities long before profit entered the picture.
zubnet.ca/SnPDvD
The internet didn't come from corporate boardrooms, it came from hackers, tinkerers, and genuine innovators exploring possibilities long before profit entered the picture.
zubnet.ca/SnPDvD
What Is Big Tech's Influence on AI Development? - KDnuggets
So, what is big tech's influence on ai development? It is both transformative and concerning.
zubnet.ca
September 26, 2025 at 2:45 PM
While everyone debates "Big Tech's influence on AI," let's remember who actually built the foundations.
The internet didn't come from corporate boardrooms, it came from hackers, tinkerers, and genuine innovators exploring possibilities long before profit entered the picture.
zubnet.ca/SnPDvD
The internet didn't come from corporate boardrooms, it came from hackers, tinkerers, and genuine innovators exploring possibilities long before profit entered the picture.
zubnet.ca/SnPDvD
This ACM piece by Mallik Tatipamula and Vint Cerf (yes, THE Vint Cerf) just articulated something I've been feeling for months working in the AI space.
zubnet.ca/BbAhZp
zubnet.ca/BbAhZp
What Lessons Can We Learn from the Internet for AI/ML Evolution? – Communications of the ACM
zubnet.ca
September 25, 2025 at 7:20 PM
This ACM piece by Mallik Tatipamula and Vint Cerf (yes, THE Vint Cerf) just articulated something I've been feeling for months working in the AI space.
zubnet.ca/BbAhZp
zubnet.ca/BbAhZp