Mihai Lazarescu
mtlbsk.bsky.social
Mihai Lazarescu
@mtlbsk.bsky.social
SUA di Trump si allontanano e l'UE deve armarsi per diventare un "boccone indigesto" per Putin, ma con coerenza morale invece dei doppi standard tra Gaza e Ucraina. Resistere all'occupazione è un diritto universale: l'UE deve difendere la libertà ovunque, senza ipocrisie.
youtu.be/DIIYHE60wLo?...
L’Europa deve attrezzarsi per essere un “boccone indigesto”
YouTube video by Lo Studio di Parsi
youtu.be
December 8, 2025 at 4:22 PM
A cohort study of 29M found that COVID-19 vaccinated individuals had a 74% lower risk of severe COVID-19 death and no increased all-cause mortality over 45 months. No increased risk of 4-year all-cause mortality was found in 18-59-year-old vaccinated individuals, supporting mRNA vaccine safety.
COVID-19 mRNA Vaccination and 4-Year All-Cause Mortality Among Adults in France
This cohort study uses the data from all adults aged 18 to 59 years living in France on November 1, 2021, to evaluate whether there is an association of receipt of an mRNA COVID-19 vaccine with long-t...
jamanetwork.com
December 6, 2025 at 10:16 AM
Many of the most influential personalities in the “Make America great again” (Maga) movement on X are based outside of the US, including Russia, Nigeria and India, a new transparency feature on the social media site has revealed.

www.theguardian.com/us-news/2025...
Many prominent Maga personalities on X are based outside US, new tool reveals
Users posing as rightwing Americans are operating internationally, per the platform’s transparency feature
www.theguardian.com
November 24, 2025 at 9:53 AM
AI-generated cover letters are longer and higher-quality, but lost their value for hiring. Employers stopped rewarding good ones ($26 premium) and wages fell 5 %. Worker losses outweighing employer gains.
www.economist.com/finance-and-...
How AI is breaking cover letters
And leading to lower pay
www.economist.com
November 16, 2025 at 1:49 PM
Grokipedia adds to Musk’s online media ecosystem cohering with his political views. On X, he has reinstated right-wing creators and allowed them to reach enormous audiences, and used X as a bully pulpit to drive government funding cuts. He leaned Grok further right.
www.nytimes.com/2025/10/27/t...
Elon Musk Challenges Wikipedia With His Own A.I. Encyclopedia
www.nytimes.com
November 8, 2025 at 3:17 AM
Polymorphic malware first emerged in the late 1980s and was about varying known techniques; LLM-based malware generates techniques that are unpredictable and fundamentally new with each execution.
www.perplexity.ai/page/google-...
Perplexity
Perplexity is a free AI-powered answer engine that provides accurate, trusted, and real-time answers to any question.
www.perplexity.ai
November 6, 2025 at 5:30 PM
Wealth acquisition, defense (minimize taxation: lawyers, accountants, managers), political capture (above $40 million), and hyper-extraction. Top 10% ≈67% of wealth, bottom 50% ≈2.5% (3.48% in 1990). Top 1% ≈30.82%, 12× of the bottom's 50%.
www.theguardian.com/news/2025/oc...
Boom time for US billionaires: why the system perpetuates wealth inequality
As the super rich grow even richer, inequality expert Chuck Collins says the system is broken – but it can be fixed
www.theguardian.com
October 12, 2025 at 11:05 AM
It's possible that rapid evolution of autism-linked genes advantageously slow postnatal brain development or increase language capacity; longer brain development in early childhood was beneficial to human evolution because it led to more complex thinking.
www.sciencedaily.com/releases/202...
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...
www.sciencedaily.com
September 28, 2025 at 2:11 PM
By charging for API access rather than monetizing advertising, Perplexity avoids some conflicts of interest that critics argue have degraded search quality. This aligns incentives with providing accurate, helpful information rather than driving commercial traffic.
venturebeat.com/ai/perplexit...
venturebeat.com
September 28, 2025 at 1:26 PM
Psychologists are often working at the limits of their capacity, and levels of burnout in the profession are high. That context makes the appeal of AI-powered tools obvious.
www.technologyreview.com/2025/09/02/1...
Therapists are secretly using ChatGPT. Clients are triggered.
Some therapists are using AI during therapy sessions. They’re risking their clients’ trust and privacy in the process.
www.technologyreview.com
September 2, 2025 at 7:04 PM
Survey of 813 scientists confirms Twitter's decline since Musk takeover. 75% report markedly reduced utility, 40% deleted accounts entirely. Bluesky emerged as preferred alternative with 38M users, higher engagement rates, no algorithm, superior moderation policies.
arstechnica.com/science/2025...
Bluesky now platform of choice for science community
It’s not just you. Survey says: “Twitter sucks now and all the cool kids are moving to Bluesky”…
arstechnica.com
August 29, 2025 at 2:38 AM
Trump is drawn to anyone exuding an aura of strength. Talk of morality, rules, duty or strategy bores him. He yearns to strike bold and quick deals with tough guys and win prizes and adoration as a result. That makes for an unstable world.
www.economist.com/united-state...
The real collusion between Donald Trump and Vladimir Putin
It may be scarier than their critics long suspected
www.economist.com
August 27, 2025 at 12:16 PM
Gemini AI median energy per prompt, from Google: 0.24 Wh (1 s of microwave), emits 0.03 g CO2, and uses 0.26 ml water (5 drops). AI chips consume only 58 %, the supporting infrastructure the rest. Energy use dropped 33x since May 2024 through optimizations.
www.technologyreview.com/2025/08/21/1...
In a first, Google has released data on how much energy an AI prompt uses
It’s the most transparent estimate yet from one of the big AI companies, and a long-awaited peek behind the curtain for researchers.
www.technologyreview.com
August 22, 2025 at 12:44 AM
Non è utile liquidare i punti di vista opposti come ignoranti o corrotti per una comunicazione inter-ideologica efficace.
mitpress.mit.edu/books-for-un...
August 21, 2025 at 12:52 AM
The chaos on timelines isn’t just a glitch. A plain social network with only AI “people” and chronological feed without ads or recommendation engine quickly sorts into partisan bubbles, attention to a noisy 1 %, and loudest and angriest shoot to the top.
arstechnica.com/science/2025...
Study: Social media probably can’t be fixed
“The [structural] mechanism producing these problematic outcomes is really robust and hard to resolve.”…
arstechnica.com
August 15, 2025 at 2:30 PM
Our neighborhood has become the only area in Gaza governed by a Palestinian administration not affiliated with Hamas since 2007. People have access to shelter, food, water, and medical supplies, without fear of Hamas stealing aid or crossfire risk with Israeli military.
www.wsj.com/opinion/gaza...
Opinion | Gazans Are Finished With Hamas
My Popular Forces control significant parts of eastern Rafah and we are ready to build a new future.
www.wsj.com
July 26, 2025 at 11:00 AM
True explosive growth requires AI to automate not just production, but research and innovation itself. Many constraints—energy, regulation, institutional sluggishness—could limit the pace of change.
www.economist.com/briefing/202...
What if AI made the world’s economic growth explode?
Markets for goods, services and financial assets, as well as labour, would be upended
www.economist.com
July 25, 2025 at 10:23 AM
Tiny AI models revealed that human and animal decision-making strategies are often suboptimal but systematic, unlike the usual optimality-based models.
neurosciencenews.com/ai-decision-...
Tiny AI Models Reveal How We Really Make Decisions - Neuroscience News
Decision-making often involves trial and error, but conventional models assume we always act optimally based on past experience.
neurosciencenews.com
July 23, 2025 at 10:18 AM
Once the training data reaches a critical threshold, the network abruptly switches to relying primarily on word meaning rather than position. This transition occurs suddenly, not gradually, resembling what physicists call a phase transition.
www.sciencedaily.com/releases/202...
July 11, 2025 at 11:33 AM
People often repeat familiar actions when making a series of decisions, even if better options are available. The strength of this repetition bias varied significantly among individuals, with some people being more inclined to stick to habitual patterns than others.
phys.org/news/2025-07...
Humans tend to repeat familiar actions when making sequential decisions, even when better options exist
Behavioral scientists have been trying to uncover the patterns that humans follow when making decisions for decades. The insights gathered as part of their studies can help shape public policies and i...
phys.org
July 5, 2025 at 8:35 AM
Artificial intelligence companies don't need permission from authors to train their large language models (LLMs) on legally acquired books, US District Judge William Alsup ruled Monday (June 23).
arstechnica.com/tech-policy/...
Key fair use ruling clarifies when books can be used for AI training
In landmark ruling, judge likens AI training to schoolchildren learning to write.
arstechnica.com
June 30, 2025 at 7:51 AM
Both gradualism and punctuated bursts likely operate: classic Darwinian processes explain day-to-day population change, while rare genomic explosions may drive the big evolutionary leaps.
www.earth.com/news/worm-dn...
Earthworm DNA is changing how we understand evolution, possibly proving Darwin wrong
Worms radically reorganized their genomes during their move from sea to land, supporting the theory of punctuated equilibrium in evolution.
www.earth.com
June 28, 2025 at 9:09 AM
Current AI systems can do sophisticated analysis, complex reasoning, and multi-step plans. But they can also develop persistent delusions, make economically destructive decisions that seem reasonable in isolation, and experience confusion about their own nature.
venturebeat.com/ai/can-ai-ru...
Can AI run a physical shop? Anthropic’s Claude tried and the results were gloriously, hilariously bad
Anthropic's AI assistant Claude ran a vending machine business for a month, selling tungsten cubes at a loss, giving endless discounts, and experiencing an identity crisis where it claimed to wear a b...
venturebeat.com
June 28, 2025 at 8:13 AM
The length of tasks AI agents can complete has been consistently doubling every 7 months over the past 6 years. Extrapolating, in under a decade AI agents would independently complete a large fraction of software tasks that currently take humans days or weeks.
metr.org/blog/2025-03...
Measuring AI Ability to Complete Long Tasks
We propose measuring AI performance in terms of the *length* of tasks AI agents can complete. We show that this metric has been consistently exponentially increasing over the past 6 years, with a doub...
metr.org
May 30, 2025 at 12:01 AM