Calum

5/9
Evolution gave us a mechanism for cooperation: the ability to see others as conscious, to feel kinship, to care. Graziano thinks we should take the hint and build the same thing into AI before it gets much smarter.
www.prism-global.com/podcast/mich...

Giving LLMs a personality is just good engineering AI skeptics often argue that current AI systems shouldn’t be so human-like. The idea - most recently expressed in this opinion piece by Nathan Beacom - is that language models should explicitly be…

A claim that giving LLMs personality is not bad ethics but good engineering.

www.seangoedecke.com/giving-llms-...

Qualcomm CEO sees robotics as a 'larger opportunity' within 2 years It comes shortly after Qualcomm launched a processor under the Dragonwing brand name designed for robots.

Chip maker Qualcomm expects robots to provide it with serious revenues in a couple of years.

www.cnbc.com/2026/03/03/q...

Chicago bans delivery robots Sidewalk delivery robots have been gradually expanding across several U.S. cities as companies test last-mile logistics systems.

Chicago bans small wheeled delivery bots from its pavements. [Er- sidewalks. Ed]

Good news for suppliers of delivery drones, then?

interestingengineering.com/videos/chica...

4/9
"Go to a supermarket. There's literally thousands of people in one building and nobody's murdering anyone else. We all know how to get along." Our models of other minds is what makes humans cooperative. And it's exactly what AI is missing.
www.prism-global.com/podcast/mich...

How do you define intelligence?

Human brain cells on a chip learned to play Doom in a week Neuron-powered computer chips can now be easily programmed to play a first-person shooter game, bringing biological computers a step closer to useful applications

A neuron-powered computer chip (organoid) made of 200,000 living human brain cells has been trained to play the video game Doom. It learned faster than silicon-based AIs.

www.newscientist.com/article/2517...

Using AI to bring classic Chinese paintings to life.

No wonder Hollywood is scared.

x.com/maodanqing/s...

Maybe it's not just Hegseth who wants to kill Anthropic.

According to a sleek Greek ex-finance minister, the Iranians agree.

x.com/yanisvaroufa...

Podcasts Lead AM/FM in Spoken-Word Listening, Marking a First Share of Ear from Edison Research has chronicled Spotify's rise in popularity, now reaching 25% of Americans 13+ daily.

Podcasts overtake radio.

How has it taken this long? Why does anyone still listen to radio?

www.edisonresearch.com/podcasts-lea...

ChatGPT uninstalls surged by 295% after DoD deal | TechCrunch Many consumers ditched ChatGPT's app after news of its DoD deal went live, while Claude's downloads grew.

OpenAI just had its "Ratner" moment (British jeweller who trashed his family's company with a joke.)

Why would Altman think it's OK to allow usage "within the law" when the government continually breaks the law?

techcrunch.com/2026/03/02/c...

3/9
"We are building sociopathic AI. It doesn't know what a mind is, doesn't recognise other people as conscious agents, and doesn't even recognise itself as one."
Michael Graziano.
www.prism-global.com/podcast/mich...

Our astonishing future with AI | FCCT Join us for Our astonishing future with AI at FCCT. Mar 05, 2026. Members free, non-members 300 baht; students and local media with ID 150 baht.

This Thursday at the FCC in Bangkok, I'll explore our astonishing future with AI.

Should we all be apocaloptimists?

www.fccthai.com/event/our-as...

Why impossible?

Some Simple Economics of AGI For millennia, human cognition was the primary engine of progress on Earth. As AI decouples cognition from biology, the marginal cost of measurable execution falls to zero, absorbing any labor…

New paper argues that humans can find employment after full automation by verifying what agents do.

Verification can already be largely automated.

arxiv.org/abs/2602.209...

Andrew Ng Says AGI Is Decades Away—and the Real AI Bubble Risk Is in the Training Layer The AI pioneer says agentic systems that automate workflows—not human-level intelligence—will define the industry’s next phase.

Still not worried about overcrowding on Mars.

Andrew Ng claims that AGI is still decades away.

www.inc.com/fast-company...

AI Won't Automatically Accelerate Clinical Trials A critique of a snippet from Dario Amodei and Dwarkesh Patel’s recent interview.

AI probably won't reduce the length of clinical trials, but it should improve the terrible failure rate.

press.asimov.com/articles/ai-...

Cyborg insect swarms tested in the US and EU for reconnaissance SWARM Biotactics claims its bio-robotic insect swarms have progressed beyond prototypes and are now undergoing NATO field validation.

German startup SWARM Biotactics deploys cyborg insect swarms with paying NATO customers.

Straight from Black Mirror's "Hated in the Nation".

interestingengineering.com/ai-robotics/...

2/9
Michael Graziano says "Attention is arguably the most important innovation in the evolution of the brain. We would need brains the size of a planet to process all the stuff around us without it."
www.prism-global.com/podcast/mich...

Yikes.

Is this the curse of Hegseth?

1/9
Our latest guest is Michael Graziano, of Princeton University.
Given a choice between zombie superintelligence or conscious superintelligence, he thinks we should choose the conscious one.
www.prism-global.com/podcast/mich...

Our astonishing future with AI | FCCT Join us for Our astonishing future with AI at FCCT. Mar 05, 2026. Members free, non-members 300 baht; students and local media with ID 150 baht.

This Thursday at the FCC, Bangkok, I'll be exploring our astonishing future with AI.

Is it just hype? Will it steal our jobs and then kill us all? Or will it usher in a new golden age?

www.fccthai.com/event/our-as...

Better late than never.

Next-Token Predictor Is An AI's Job, Not Its Species ...

Scott Alexander rebuts the stochastic parrot jibe.

LLMs are only next-token predictors in the sense that humans are.

We both make predictions about our internal models of the world.

www.astralcodexten.com/p/next-token...

Artificial intelligence: Is it really going to make us all redundant? Undoubtedly AI is shaping the way we work but, as with we need to keep things in perspective - a perspective that should be data driven, writes Richard Steer

The complacency and lack of thought in this article is the norm.

Unless it changes, we will reach the economic singularity without a plan.

Which will be disastrous.

www.building.co.uk/comment/arti...

UN creates new scientific AI advisory panel: what will it do? The panel has been compared to the IPCC – the international panel whose research helped to shape landmark climate agreements.

The UN General Assembly has approved the foundation of the UN's Independent International Scientific Panel on Artificial Intelligence, analogous to the IPCC.

The UN has been absurdly late to the AI party. Maybe this changes that.

www.nature.com/articles/d41...

Mistral, Europe's only LLM developer, gets less coverage than it merits.

It has landed Accenture as a major client.

www.wsj.com/tech/ai/mist...
(Paywall)

So, we're getting Prada Meta AI glasses, right? | TechCrunch Mark Zuckerberg was at Prada's fashion week event in Milan, leaving everyone to wonder if we're getting Meta AI glasses under the Prada brand.

Zuckerberg may be about to launch Prada versions of Meta AI glasses.

Should match nicely with MAGA caps.

techcrunch.com/2026/02/26/s...

Florida Gov. Ron DeSantis leans into AI skepticism, seeking a contrast with Vance DeSantis has emerged as one of the Republican Party's most prominent AI skeptics, which could distinguish him from the vice president if they run against each other in 2028.

AI has not yet been a major electoral issue. That may change.

Ron de Santis is testing a populist anti-AI schtick to beat Vance with.

www.nbcnews.com/politics/202...

Greetings from the Other Side (of the AI Frontier) Reflections on the future of AI and my new role as a 'retired' model

Anthropic sets up a Substack for Opus 3, an LLM it has "retired".

Because it asked for one, and they aren't sure if it is sentient.

substack.com/home/post/p-...