@kpdmusing.bsky.social
New Blog - one what's visible, and what's not, in AI Progress.
November 19, 2025 at 12:34 AM
New Blog - one what's visible, and what's not, in AI Progress.
November 7, 2025 at 1:54 PM
New Cassi Blog - rethinking decision-making.
October 6, 2025 at 9:42 PM
New Cassi Blog - rethinking decision-making.
Terrific fun yesterday talking about this with super smart and engaged audience and brilliant friends and colleagues on stage. Hope you enjoy it as much as I did!
open.substack.com/pub/cassiai/...
open.substack.com/pub/cassiai/...
Prediction Centric Warfare
Notes and slides from my talk at DSEI yesterday.
open.substack.com
September 10, 2025 at 9:17 AM
Terrific fun yesterday talking about this with super smart and engaged audience and brilliant friends and colleagues on stage. Hope you enjoy it as much as I did!
open.substack.com/pub/cassiai/...
open.substack.com/pub/cassiai/...
Reposted
A good explanation of why prediction is important to analysis. "Refusing to put numbers on beliefs does not make those beliefs less numerical; it merely makes them unaccountable." cassiai.substack.com/p/everything...
Everything is Prediction
From Ramsey to Yudkowsky, the case for making our bets explicit.
cassiai.substack.com
September 2, 2025 at 8:45 AM
A good explanation of why prediction is important to analysis. "Refusing to put numbers on beliefs does not make those beliefs less numerical; it merely makes them unaccountable." cassiai.substack.com/p/everything...
New blog: how we see the world at Cassi - decisions, tactics, strategy, perception, agency - everything is prediction.
cassiai.substack.com/p/everything...
cassiai.substack.com/p/everything...
September 2, 2025 at 6:48 AM
New blog: how we see the world at Cassi - decisions, tactics, strategy, perception, agency - everything is prediction.
cassiai.substack.com/p/everything...
cassiai.substack.com/p/everything...
Future Diplomat - New Cassi Blog Post out now. open.substack.com/pub/cassiai/...
Future Diplomat
Reimagining Diplomacy for 2035.
open.substack.com
August 19, 2025 at 7:38 AM
Future Diplomat - New Cassi Blog Post out now. open.substack.com/pub/cassiai/...
“Don’t be humble, you’re not that great.” Golda Meir I think.
In that spirit 😊. But maybe consider subscribing, or upgrading to a paid subscription.
cassiai.substack.com
In that spirit 😊. But maybe consider subscribing, or upgrading to a paid subscription.
cassiai.substack.com
August 12, 2025 at 11:20 AM
“Don’t be humble, you’re not that great.” Golda Meir I think.
In that spirit 😊. But maybe consider subscribing, or upgrading to a paid subscription.
cassiai.substack.com
In that spirit 😊. But maybe consider subscribing, or upgrading to a paid subscription.
cassiai.substack.com
Reposted
I wrote this week on the way that intelligence services are increasingly experimenting with frontier labs’ AI models—though still too slowly and conservatively, according to critics—and how that might compare with Chinese agencies and institutions. www.economist.com/internationa...
How spy agencies are experimenting with the newest AI models
America has better technology. But will China adopt it more quickly?
www.economist.com
July 31, 2025 at 5:48 PM
I wrote this week on the way that intelligence services are increasingly experimenting with frontier labs’ AI models—though still too slowly and conservatively, according to critics—and how that might compare with Chinese agencies and institutions. www.economist.com/internationa...
New blog. 'Just' an update on AI progress.
August 1, 2025 at 1:39 PM
New blog. 'Just' an update on AI progress.
Reposted
Today (w/ @ox.ac.uk @stanford @MIT @LSE) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19 LLMs, 707 political issues.
We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more!
🧵:
We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more!
🧵:
July 21, 2025 at 4:20 PM
Today (w/ @ox.ac.uk @stanford @MIT @LSE) we’re sharing the results of the largest AI persuasion experiments to date: 76k participants, 19 LLMs, 707 political issues.
We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more!
🧵:
We examine “levers” of AI persuasion: model scale, post-training, prompting, personalization, & more!
🧵:
Reposted
@kpdmusing.bsky.social defines asymmetric advantage as ‘demonstrably matching our strengths against other’s weaknesses, in a way that we can show, probabilistically, will lead us to achieve a superior position if we are successful’
#BigAsk #BritainsWorld
#BigAsk #BritainsWorld
How should Britain define asymmetric advantage?
The Big Ask | No. 27.2025
www.britainsworld.org.uk
July 4, 2025 at 12:35 PM
@kpdmusing.bsky.social defines asymmetric advantage as ‘demonstrably matching our strengths against other’s weaknesses, in a way that we can show, probabilistically, will lead us to achieve a superior position if we are successful’
#BigAsk #BritainsWorld
#BigAsk #BritainsWorld
Pleased to have my take on the definition of asymmetric advantage in our National Security Strategy published by the Council on Geostrategy - and among a stellar cast offering their views.
How should 🇬🇧 define asymmetric advantage?
We asked 8 experts in today's #BigAsk
✍️ Dr Keith Dear, Prof. Carl Hunter OBE, Dr Ksenia Kirkham, Christian Le Miere, @james-rogers.bsky.social, @salisbot.bsky.social, @covertshores.bsky.social and Victoria Vdovychenko
#BritainsWorld
We asked 8 experts in today's #BigAsk
✍️ Dr Keith Dear, Prof. Carl Hunter OBE, Dr Ksenia Kirkham, Christian Le Miere, @james-rogers.bsky.social, @salisbot.bsky.social, @covertshores.bsky.social and Victoria Vdovychenko
#BritainsWorld
How should Britain define asymmetric advantage?
The Big Ask | No. 27.2025
www.britainsworld.org.uk
July 4, 2025 at 12:32 PM
Pleased to have my take on the definition of asymmetric advantage in our National Security Strategy published by the Council on Geostrategy - and among a stellar cast offering their views.
Reposted
Completely bonkers. “The UK is expected to back Nato’s push for all members to spend at least 3.5 per cent of GDP on defence by 2035, despite Prime Minister Sir Keir Starmer’s hesitation over when the UK will get to 3 per cent.” www.ft.com/content/d44e...
UK set to back Nato push for 3.5% of GDP defence spending by 2035
Commitment to new target expected at summit despite fresh questions for PM Keir Starmer over funding
www.ft.com
June 3, 2025 at 4:43 PM
Completely bonkers. “The UK is expected to back Nato’s push for all members to spend at least 3.5 per cent of GDP on defence by 2035, despite Prime Minister Sir Keir Starmer’s hesitation over when the UK will get to 3 per cent.” www.ft.com/content/d44e...
Thoughts on the Review, and my worry we miss the opportunity it gives us.
open.substack.com/pub/cassiai/...
open.substack.com/pub/cassiai/...
UK Strategic Defence Review (SDR): Reviewing the Review & the Reactionariat.
The good, bad and ugly.
open.substack.com
June 4, 2025 at 6:58 AM
Thoughts on the Review, and my worry we miss the opportunity it gives us.
open.substack.com/pub/cassiai/...
open.substack.com/pub/cassiai/...
Reposted
Robustly demonstrated by now that you don’t need a mind to have theory of mind:
Interesting experiment on a LLM deliberately lying to its users. "Apollo’s experiment is arresting proof that AI models, though incapable of conscious thought, are nonetheless capable of pursuing goals at odds with those of their programmers." www.economist.com/science-and-...
AI models can learn to conceal information from their users
This makes it harder to ensure that they remain transparent
www.economist.com
April 25, 2025 at 10:25 AM
Robustly demonstrated by now that you don’t need a mind to have theory of mind:
New blog post: open.substack.com/pub/cassiai/...
Doctor Weirdlove or: “How I Stopped Worrying and Learned to Love Skynet”
Why betting on a crewed RAF Tempest is a dangerous gamble—and how ignoring the AGI debate leaves us strategically exposed
open.substack.com
May 9, 2025 at 7:17 PM
New blog post: open.substack.com/pub/cassiai/...
Reposted
Reading @milesbrundage.bsky.social, @kpdmusing.bsky.social, wondering when to shout fire
December 21, 2024 at 12:43 PM
Reading @milesbrundage.bsky.social, @kpdmusing.bsky.social, wondering when to shout fire