Alex Panetta

2/

-Making certain practices like automated weapons and mass-surveillance a crime against humanity

-More taxes for the rich. Warns his billionaire buddies that if they don't get on board with the 'good version' of progressive taxation they're gonna get the bad kind someday, imposed by a mob

A striking essay from the CEO of Anthropic warning that AI without safeguards could lead to the end of democracy and a return to feudalism. He proposes certain internal safeguards for AI companies but also societal ones like:

1/

www.darioamodei.com/essay/the-ad...

Spending more time thinking about Washington than I’d like. Back to my regularly scheduled current programming:

Studying the social implications of A.I. Including what happens to training and tutoring in a world where anyone can program video games

open.substack.com/pub/alexpane...

1. Trump says he’s fine with China-Canada deal. Been working on his own

2. Gets upstaged at Davos

3. Goes nuclear

'The old order is not coming back,' Carney says in provocative speech at Davos | CBC News Prime Minister Mark Carney delivered a frank assessment of how he views the world in a provocative speech in Davos, Switzerland, on Tuesday, where he said the longstanding U.S.-led, rules-based intern...

A eulogy for the postwar order: "A country that cannot feed itself, fuel itself or defend itself has few options. When the rules no longer protect you, you must protect yourself," Carney said. www.cbc.ca/news/politic...

Think your neighborhood gossip is awkward? Here’s mine in Ankara: U.S. and Danish ambassadors live next door to each other. Must be one helluva conversation across the backyard fence these days.

Craziest thing about living overseas is waking up Monday morning to posts from people I follow for politics like, “I CAN’T BELIEVE THIS HAPPENED,” and “Wow!” And, “HOLY CRAP,” and I’m making mental preparations to live in a bunker before realizing it’s about NFL football.

An OpenAI economics researcher quits, saying the company behind ChatGPT is burying research on painful economic consequences of A.I. and only interested in publishing propaganda. Piece here contrasts this with Anthropic publicly making dire claims about mass job losses. medium.com/predict/open...

! “It's OK. That's what he should be doing. It's a good thing for him to sign a trade deal. If you can get a deal with China, you should do that," Trump said, referring to Carney.

/ … by the way, the last time this dynamic surfaced, with Canada turning towards China amid tariff problems with Trump, the United States laid charges against Meng Wanzhou as she was flying through Canada… and that was that

2/ Yet some in DC suspect Canadian politics has been corroded by PRC money. Where’s the foreign agent registry? Where are prosecutions for TNOC money-laundering? Hence some skepticism Canada has been as steadfast an ally as it imagined.

One does not exclude the other.

Two contradictory-sounding takes that can both be true:

1/ In a geopolitical dog race, Canada bet everything on the aging, weakening dog — and for good reasons: Similar values to 🇺🇸, similar culture, shared geography. And then got bitten by the dog. So is now understandably hedging its bets. 🇨🇳

The reaction in Washington is going to vary on each side of the aisle. But I suspect Carney actually welcomes a degree of this chatter in the U.S. Comes with risks tho!

Today’s news is making me think of a conversation I had with a prominent Canadian official, fuming that the U.S. was negotiating trade liberalization with China while Canada was frozen out, having poisoned its relationship with the fastest-growing economy on Earth to please DC to little gain

Exactly. It’s a tell

Personal confession: I'm instantly suspicious of anyone who utilizes the word 'utilize'

Observation: Claude Opus 4.5 is a far, far superior writer to ChatGPT. I sometimes submit a draft to both for an edit, then make my two 'editors' argue with each other. And Claude is consistently better. Like here where it mocks ChatGPT for proposing to butcher my copy:

/ Why collect data? It's the first question any individual or institution should ask. I write about 3 specific use cases, from a current course reading.

I focus on one, using the scary stock-market data as an example.

And I swear this has nothing to do with Jerome Powell. The post was pre-written

I did not build this app to predict a stock market crash but… What I did want to show you: How easily you can now build data dashboards

You can visualize your finances. Diet. Travel plans. Sales. Customer-service complaints. Learning program. Exercise. Football pool strategy. Music and movies to check out. Media diet. It’s so easy now. Cost of building a data dashboard: $0

Here's how: alexpanetta.substack.com/p/i-did-not-...

I did not build this app to predict a stock market crash but...

What I wanted to show was how you or your institution can create a data dashboard, in a few easy steps. A project that would have cost tens of thousands is now virtually free for anyone. alexpanetta.substack.com/p/i-did-not-...

Developing…. Salman Rushdie vs the Iranian regime

After a few years stumbling through the desert, I could drink these words: "There were moments for the Canadiens when the hockey was so extraordinary it looked like the Central Red Army at the height of its powers."

I'm very sorry to hear that. In humans, dementia tends to mean forgetting recent details. In AI, it's forgetting older context. The emerging techie term for it is "context rot."

/ Editor's role: Generate assignment ideas. When the material comes in, test the accuracy. Check facts; especially implausible-sounding findings. Demand source info. Question whether the right baselines were used for comparisons. Basically...just like dealing with a flaky star reporter in a newsroom

... It just apologized to me for-- and I quote -- "making up plausible-sounding numbers and attributing them to a credible source."

Gonna write a blog post soon about how the skills you're taught as a news editor are going to be essential in dealing with AI.

/

Opus 4.5 is a beast. I'm using it semi-constantly. But anyone here pumping it as AGI or the start of the singularity or whatever is selling snake oil, if my exchange with it a few minutes ago is any indication.

It just apologized to me for ....

Happy New Year everyone, from 8 hours ahead of where I usually celebrate it.

Yesterday published a substack on how to control your own movie-recommendation algorithm 🎥

…right after enjoying a relic of pre-pre-algorithmic entertainment

I do miss home, so this unexpected appearance of 1976 Olympic paraphernalia during a Mediterranean boat cruise is as close as I’ll come these holidays. You take traces of home where you can get them.

Nothing to watch this holiday? Seize control of your movie algorithm A simple recommendation algorithm I built. Adjust it. Find movies with it

We are immersed in algorithms. Yet we rarely see these formulas with such influence over our lives. My latest blog post on practical, ethical A.I. lets you try something rare: Control your own algorithm. For movie recommendations. Adjust it.
Maybe even discover a fun film to watch these holidays