Savour Negative Emotions to Transform Them Into Energizing Wellbeing
The most powerful meditation technique I've come across recently: rather than just "observe" negative emotions, learn to actively savour them. With a bit of practice, it's often possible to find an enjoyable aspect of them: the energised quality of anger, or the tender feeling of sadness. Just imagine it feels really good. You can also notice the good intentions behind them, making it easier to feel warmly towards them: anxiety is trying to protect you from something, doubt makes you notice you don't understand. Savoured in this way, you may notice feeling more energised and alive. In formal practice, by focusing on the positive aspects and enjoyment, you can even ride a negative emotion into the jhanas. You can then start to relate to painful emotions not as things to be feared, or even passing "clouds" to be observed, but embraced as something that makes you feel a deeper sort of wellbeing. I'm certainly not there yet, but even moving 10% in this direction feels pretty helpful.
Offer UK Data Centers for Guaranteed Access to Frontier AI
More practical idea: make it easy for frontier AI companies to build datacentres in the UK, in return ask for guaranteed access to frontier models.
AI Leaps Five Years in Six Months
In 5 years, AI has gone from: Can barely talk → conversationally fluent but terrible at math and coding → solving 80-year-old math problems. In an algorithmic feedback loop, you could get another 5 years of progress in 6 months.
Irreversible Space Settlement: The Overlooked Existential Threat
What's the most underrated existential risk? Irreversible space settlement. 1. AI could make space settlement possible in our lifetimes: 1 minute of solar energy is enough to accelerate 10 billion 1kg self-replicating AI probes to 99% the speed of light.
Hardware Feedback Loops Drive Explosive AI Growth
New paper with the rigorous version of what I've been saying: * The feedback loops in AI are strong and multiple * The hardware ones are even more powerful than the algorithmic ones that get most attention * Most models that account for...
Real AI Risks Ignored for Imaginary Fears
When there's so many good reasons to be worried about AI, it's amazing people are focused on those that are totally fake.
Claude 4.6 and Myth
New post: Is AI already accelerating? A review of the evidence.🧵 Claude 4.6 and Mythos are actually on trend based on an index of 37 benchmarks post-2024: https://t.co/75D2i887OD
Evidence-Based Medicine Once Labeled Dangerous Innovation by Doctors
Obviously you want "evidence-based medicine", but the term was only coined in 1991. Doctors called it a "dangerous innovation", preferring to rely on their judgement over trials.
Question Single AI Skeptic Studies; Avoid Overreacting
This is crazy. Be extremely suspicious of single studies skeptical about AI. People will jump on a shred of made up evidence.
Do Waiter Services Add Unnecessary Luxury Costs?
If a 2* Michelin meal could magically appear on your dining table, would you prefer that to going somewhere to have a waiter hover at your elbow? How many "relational" jobs are actually a tax on luxury services many would prefer...
Recession Will Reveal Which Jobs AI Disrupted
We'll only find out for sure which jobs were AI-exposed in the next recession. https://t.co/3pc4idmThY
AI-Driven Productivity Could Lift 2030 GDP 3.3%
Economists think in a "rapid" scenario where in 2030 "AI outperforms all humans at many tasks, and run businesses at the level of a competent CEO" and robots can "perform nearly all industrial tasks faster than humans", GDP growth would...
Robots Likely in Homes by 2030, Scarce Industrially
Forecasters think there's a ~60% chance that by 2030 "robots exist that can navigate homes and perform basic tasks" or better, but only 17 million are used industrially by 2040. Seems insane. https://t.co/SgVhz1xW6C
AI Cuts Service Costs Unseen, Inflating Official Rates
AI is already making services far cheaper, but probably not in ways captured in the inflation deflator. So inflation is going to look higher than it really is, which could mean the Fed overtightens.
AI Potential Outpaces Adoption by Fivefold, Gap Likely Smaller
Based on an analysis of AI tools, the MIT Iceberg Index estimates ~11.7% of tasks could be done by AI vs actual adoption of ~2.2% (in terms of wage value). That suggests AI deployment is lagging what's possible by 5x. But...