My mastodon account: @vardi@fediscience.org
Professor, Rice U., Houston
Joined April 2008
- Tweets72.7K
- Following136
- Followers13.2K
- Likes12.1K
Moshe Vardi retweeted
Has the generative AI unraveling begun?
Sure looks that way.
A recap of why everyone should have long seen it coming:
garymarcus.substack.com/p/th…
Moshe Vardi retweeted
"AI Agents Can’t Talk," by Abhilash Pakalapati, says teams will drown in #token costs and unpredictable runtime behavior until #developers stop framing multi- #agenticAI design around the metaphor of human #conversation. bit.ly/4exhf80
Moshe Vardi retweeted
Wow. Some things are definitely going to change at Nvidia.
And I think Jensen is going to look worse and worse over time.
NEW: In a tense late-summer meeting, Jensen Huang called his Chief Financial Officer Colette Kress a “deer in the headlights.”
His concern: Nvidia’s plans to fund the AI boom with its balance sheet were running into problems. He needed to figure it out.
He called her a “reporting CFO,” not someone who could help him define a strategy or make decisions.
With @RWhelanWSJ for @WSJ
wsj.com/tech/jensen-huang-co…
Moshe Vardi retweeted
the MSFT Outlook outage is very 2026, and a sign of even worse things that are likely to come.
Moshe Vardi retweeted
For the eighth year in a row, University of Houston and Rice University are No. 1 on The Princeton Review’s annual ranking of top undergraduate and graduate schools for entrepreneurship studies, respectively. bizjournals.com/houston/news…
Moshe Vardi retweeted
Last month, Winchester College, one of the most exclusive schools in the country, banished laptops from its classrooms and went back to pen and paper. Its headmistress, Elizabeth Stone, called them “distracting”, and said results had already improved. Eton has restricted iPads after more than 100 parents signed a petition.
Meanwhile, the Department for Education is pressing ahead with plans to give AI-powered personal tutors to 450,000 disadvantaged pupils a year by the end of 2027. In other words, the children of the elite are getting books, teachers and silence. The children of the poor are getting chatbots.
That, more than anything, is what has stayed with me from my conversation on The Brink with Sophie Winkleman, 46, the actress best known as Big Suze in Peep Show, and now one of Britain’s most formidable campaigners against the takeover of childhood by screens.
She made her case with passion born of a great deal of research. But perhaps the most powerful argument she offered came from the last place you would expect: a room full of AI evangelists...
New post. Link in first reply
Moshe Vardi retweeted
i am calling it.
in light of the below and all that has happened in the last 10 days I think we can safely say now that November has been the beginning of the end of the 🫧.
JUST IN: *Nvidia ($NVDA) is expected to slow its aggressive AI investment strategy following Jensen Huang's resignation, putting billions in startup funding and credit guarantees at risk, The Information reports.
Moshe Vardi retweeted
Everyone defending Jensen is trying to make it about the sex.
It’s not about the sex.
It’s about how he enabled and even lobbied for Chinese AI, about how he subverted US policy, and how he carelessly left himself vulnerable to both blackmail and espionage.
And it’s about how he utterly blew off the problems inherent appearance of conflict of interest, to the peril of his company, his country, and his industry.
He also left Nvidia open to endless litigation and investigation (which is why they may be forced to fire him despite his immense talents).
The fact that he also slept with an employee is just icing on the cake, in a way that further emphasizes his poor judgment, and that would only make the shareholder lawsuits worse.
Firing him will be a huge blow, but keeping him on is no longer tenable, either.
Moshe Vardi retweeted
Led by Philip Bedient, Rice University’s SSPEED Center recently brought together leaders from across Texas to advance development of a Flood Early Warning System (FEWS) for the Texas Hill Country. ow.ly/Bbzb5106jto
Moshe Vardi retweeted
Muhammad Ali came to Danny's memorial in Los Angeles and I praised him as "a champion of humanity". When he squeezed my hand I almost fainted -- glad I didn't have to fight him fist to fist.
@yudapearl when I read that Muhammad Ali, tried to rescue Danny and then attended his funeral I cried. He had to do neither. And I want to believe that it led to his embrace of his son in law and grandson who’s bar mitzvah he attended. I guess the point is that Ali’s attachments to Islam and corresponding actions and statements on Israel didn’t just evolve over the years but there was something about the openness of his heart that led to those changes. I’m also of the belief that your principled dedication to peace helped pave the way for such changes.
Moshe Vardi retweeted
Breaking News: Judge Richard A. Posner, the most-cited legal scholar of the latter half of the 20th century, has died. nyti.ms/4Aizsyo
Moshe Vardi retweeted
The 100 Year Study of AI
A remarkable project in which AI researchers report every five years, for a century, on AI progress. The latest report covers 2021–26, and it’s out NOW:
ai100.stanford.edu
Moshe Vardi retweeted
There’s a specific mantra drilled into government workers, military officers, and intelligence personnel during counterintelligence briefings: “If she’s a 10 and you’re a 5, she’s a spy!” So how did the world’s most powerful AI executive fall for it? wired.com/story/chinas-nvidi…
Moshe Vardi retweeted
German Chancellor @_FriedrichMerz is absolutely right. The fact that Jewish kindergartens and synagogues need police protection and bulletproof glass is horrifying. And sadly, this isn't just a reality in Germany. It's one Jewish communities face around the world.
He didn't mince words when describing antisemitism in Germany as "omnipresent." His warning that Germany faces a "moment of truth" for its commitment to #NeverAgain should resonate far beyond its borders. reuters.com/world/merz-says-…
Moshe Vardi retweeted
In just a few hours, Kinder Institute researchers will present findings from our first economic mobility report, an analysis examining 15 years of changes in wages, job availability and the region’s leading industries.
Discover what the research reveals about Houston's changing economy and how these insights could inform conversations about pathways to greater prosperity across the region.
This free webinar begins TODAY AT 10 A.M. and includes a live Q&A with the research team.
REGISTER: riceuniversity.zoom.us/webin…
Moshe Vardi retweeted
The Huang Affair is the new Watergate.
It no longer just about Jensen or Nvidia. It’s about what the US government knew and when did they know it.
🚨MAJOR BREAKING NEWS: President Trump lied on Monday about when he had been told that Jensen Huang was in a relationship with a Chinese spy to whom he was selling chips. In reality he has known since May, and chose to do nothing, despite the risks to national security, and despite warnings from Rubio.
nytimes.com/2026/11/10/us/po…
Moshe Vardi retweeted
It’s a rare and special opportunity when you get to share the stage with an older brother – especially one whom you admire greatly and have learned so much from. Thank you to Federal Reserve Bank of Dallas President Lorie Logan for bringing us together.
Moshe Vardi retweeted
Researchers found that letting ChatGPT do ALL the creative thinking for you can actually make your writing WORSE.
And the reason is fascinating.
In a study published in PNAS Nexus, researchers recruited 285 students for a creative-writing experiment.
Participants were asked to write two 1,000-word stories.
First, without AI.
Then, using ChatGPT 3.5 in one of three ways:
1. AI did everything: Generated ideas, created outlines, drafted, and edited. Humans mainly approved the results.
2. Humans created the ideas and outlines. AI handled drafting and editing.
3. Humans and AI worked together throughout the creative process.
Researchers collected 549 stories, each evaluated by four human raters.
The results were surprising.
The AI-led group received significantly worse overall quality results than the group that continuously collaborated with AI.
Their stories also received lower ratings for interestingness and coherence.
And the writers reported lower satisfaction with the process.
But here's where things get interesting.
Researchers measured how similar the stories were to one another.
The AI-led group produced the most similar stories.
Different writers. Different stories.
But their work kept following similar patterns.
Meanwhile, the group that created its own ideas and outlines produced the least similar stories among the AI-assisted groups.
And here's another surprising finding.
Letting AI handle everything didn't produce a statistically significant time advantage over the other AI-assisted approaches.
All three groups had similar completion times.
So keeping humans involved in creative decisions avoided some of the quality and diversity problems without a clear time penalty.
One important limitation: The experiment used ChatGPT 3.5 and a tightly controlled writing task. Newer models and real-world workflows may produce different results.
But the takeaway is powerful.
AI can help you write, organize, and edit.
Yet letting it make every creative decision can leave you with more generic, less engaging work.
Let AI help with the writing.
But keep your hands on the creative decisions.
