High intelligence working on Super Intelligence. Independent applied research in intelligent systems in Melbourne.
Melbourne, Australia
Joined March 2009
- Tweets5.6K
- Following1.4K
- Followers2.7K
- Likes79.4K
This has to be the worst call/post by an MP on AI I have seen. Disingenuous use of our of date trial as Anthropic sees 2x improvment in development due to AI
In a randomised trial, software developers using AI took 19% longer to complete tasks. Yet they believed AI had made them 20% faster.
At the ‘Evidence in Action’ event, I argued for a government that tests what works, and saves money by learning from the results. ministers.treasury.gov.au/mi… #randomistas
Readers added context they thought people might want to know
The 19%/20% figures are from a July 2025 METR study of experienced open-source developers using early-2025 AI tools, which the authors describe as an outdated snapshot; a 2026 follow-up found some evidence for speedup with newer tools.
metr.org/blog/2025-07-1…
metr.org/blog/2026-02-2…
arxiv.org/abs/2507.09089
Flink Labs retweeted
It was the best of times, it was the worst of times, it was the age of recursive self-improvement, it was the age of data center bans, it was the epoch of intelligence explosions, it was the epoch of local backlash, it was the season of acceleration, it was the season of red tape, it was the spring of superintelligence, it was the winter of compute.
Flink Labs retweeted
It’s pretty clear we are headed here to catastrophe.
Massive job losses coming to corporates.
Everyone will need to have agency. Become an entrepreneur.
Great afternoon adventure for young me. My second fave super computer, first computer I programmed on and war games vibes.
Flink Labs retweeted
Atlas shrugged, the Elves sailed into the West, and the Orbital Data Centers launched to Sun-synchronous orbit.
Flink Labs retweeted
Agent-Based Modeling: Worlds Within Worlds, a new interactive story from the Santa Fe Institute, chronicles the rise of ABMs in the social sciences, with a focus on SFI contributions.
Over the next few weeks we’ll highlight pieces of this story. First, how it begins:
PART 1: In a former convent in Santa Fe in the late 1980s, economists, physicists, and computer scientists scribble equations on a blackboard where an altar used to be. They are discussing a promising new approach to scientific modeling, one that represents individual actors in, say, an economy or a pandemic, and the rules those “agents” follow.
Today, the method — known as agent-based modeling — offers powerful insights for predicting financial crises, managing pandemics, and even uncovering truths about the past.
Read the full story and run the models yourself: santafe.edu/abm
Flink Labs retweeted
🤖 Made with AI
At its meeting today, the Board decided to increase the cash rate target by 25 basis points to 4.85 per cent.
Read the full statement here: ow.ly/8zaP5103coK
Flink Labs retweeted
Normalise saying “this is really fucking weird, but I don’t think it should be banned”. True liberalism needs to make a comeback.
Flink Labs retweeted
Ready the weapons. Man your battle stations. Get birds in the air. And then? Stand ready and watch the enemy turn around.
Flink Labs retweeted
The lack of empathy in the public service class for business owners is truly sad.
What they fail to understand is that not only are these business owners their customers AND employers, but that if they cease to exist, they will eventually cease to exist.
Social cohesion is a vague term, but one example of what social cohesion should be is every Australian seeing every other Australian as a team member.
We’re all on the same team and are trying to achieve greatness.
Most people wouldn’t let down their friends, families or co-workers.
Perhaps we should apply that same mindset to the customers we serve.
Public service exists to serve the people, that’s literally all it needs to do.
Flink Labs retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
Flink Labs retweeted
NASA가 공개한 화성의 밤하늘(4K 해상도)
Readers added context they thought people might want to know
이 영상은 NASA가 공개한 화성 밤하늘이 아닙니다. NASA 공식 계정과 사실 확인 결과 유사 영상은 디지털 합성·편집된 것입니다.
x.com/NASAMars/statu…
petapixel.com/2025/08/22/no-…
newtral.es/video-cielo-ma…
science.nasa.gov/solar-system/p…
Harsh but true! Offices have sooo many benefits
I think people underestimate how much personal freedom is available to a man who is physically at the office
You can get coffee, take a walk, have lunch with somebody, spend 40 minutes discussing college football, or just sit quietly and think
All of this is understood to be part of working
Try doing the same things at home and suddenly there are questions about what exactly you've been doing all day