Garth Luke retweeted
Now is the time for pacing the frontier in AI, and this is the best proposal I've seen for how to do it.
On the default path, superintelligence by the end of 2027 or 2028 seems fairly likely (35%+): companies are on track for fully-automated AI R&D soon, and my best guess is that that would drive ~6 years of AI progress in one.
That’s like going from GPT-3 to AI that solves Navier-Stokes in <1 year; surpassing the range of human abilities in most cognitive tasks. Even though it's not a leap to godlike superintelligence, it is enough to pose a serious risk of AI takeover, and probably to confer decisive power on whichever country gets there first.
Yet the leading AI companies are lacking in basic infrastructure; they can get hacked with a couple of days of work. This needs time to solve.
Verifiably reallocating compute from capability development to inference and safety gives that time, and means, in the near term, cheaper AI for users.
AI progress can still continue. AI development can progress at a reasonable pace (maybe around the pace we have now, slowing through more dangerous periods, with a dial that can be turned up or down), while avoiding the huge societal disruption and risk of catastrophe from rapid recursive self-improvement. In the meantime, we can enjoy a golden age of human-AI collaboration as AI helps make meaningful breakthroughs in science and technology.
We’re releasing a concrete proposal for how a near term AI slowdown could work. It involves:
1. AI companies declaring how they are using their compute and letting third party auditors verify those disclosures. We call this Verified Compute Allocation Disclosures (VCAD).
2. Compute Reallocation towards serving customers and safety, away from capabilities R&D.
(2) slows AI capabilities , while (1) makes the slowdown credible.
AI companies currently seem to be going for ‘safety case pacing’. But: (i) there’s no consensus on what a valid safety case looks like so there’s no way to consistently audit for compliance, and (ii) if the R&D doesn’t get paused, temporarily delaying a training run for safety doesn’t meaningfully slow the overall trajectory of capabilities progress. It would be great if we had a viable safety case regime, but we need much more time to get it working.
Compute reallocation is a good pacing proposal because it is relatively specific, can be verified, and would lead to a more durable slowdown.
Garth Luke retweeted
List of prominent people who have recently come out in favor of ~stopping further AI development. 🧵 (1/n)
Garth Luke retweeted
It's amazing that this is literally the first Democratic politician I've ever seen do this.
I walked the Trump rally line to listen, not make assumptions. I talked with voters, asked questions and heard what matters to them.
When I'm elected, I'll represent all 82 counties and every Mississippian, no matter who they voted for. That's what being your senator means. #MississippiMatters
Garth Luke retweeted
Former Anthropic researcher Jacob Coxon (@hilbertspaess) and Jon Stewart discuss AI, jobs disappearing, universal income, and whether tech CEOs are creating a dystopian future
Australia could promote the development of an international safety framework for AI by offering to play a strong testing and verification role. lowyinstitute.org/the-interp…
1
19
‘Anyone who tells me “God gave me this land and told me to do this violent act”, I can’t have a conversation with.’
@AndrewMarr9 'has no time for Hamas or Islamism', but says the idea of ‘allowing the Palestinians to be wiped out is intolerable'
Garth Luke retweeted
Solar will become the dominate source of energy this century - as fossils were in the last.
One third of the world’s total solar capacity was installed in the last 19 months.
The first terawatt took decades to build – the latest one took just over a year and a half.
China is the global leader, but India has moved to the second spot.
The number of countries with at least 1 gigawatt of solar has risen from 42 in 2020 to 74 now.
The underlying direction remains clear: Annual solar additions are moving towards 1 terrawatt while deployment continues to spread across all continents.
Garth Luke retweeted
Two observations on this story about Indonesia:
1. Latvia, near the Arctic, population 2m, has more installed solar capacity than Indonesia, a country of 300m people and 30 times Latvia's size, plastered on the Equator
2. Jakarta is that rare city anywhere in the world today: when you approach by plane, you see almost zero solar panels on its roofs
ft.com/content/ff321dc1-41d6…
Garth Luke retweeted
Aid budgets are shrinking, countries are calling for greater health sovereignty, and the global health architecture is under pressure.
@peterbaker17 explores whether an IDA Health Window could be the future of global health financing.
cgdev.org/publication/ida-he…
Garth Luke retweeted
The world has warmed by around 1.4C since 1850.
It took 148 years for the first half of that warming to occur, and just 27 years for the second half!
Garth Luke retweeted
The poorest economies in the world are not achieving economic growth.
The consequence is that progress against the very worst poverty is coming to a halt.
This is one of the worst problems in the world today, but hardly gets any attention.
Garth Luke retweeted
ODA threshold should reflect purchasing power (and be lower).
Blog:
cgdev.org/publication/any-ch…
Garth Luke retweeted
Today, India’s solar and wind deployment per capita is 5.5x higher than China’s was at the same stage of development. India is peaking its coal use a full decade earlier in its development curve than China did
>In 2025, India reached a GDP per capita of ~$10,000 (PPP). At that exact same wealth level (around the year 2010), China’s wind and solar generation was only 37 kWh per person. India is already at 205 kWh per person, literally 5.5 times higher
>India reached a 5% solar share in its power mix at a GDP per capita level 3 times lower than China’s when it hit the same milestone
>India already generates 33% more economic output per unit of energy than China did at the same stage of development
Garth Luke retweeted
AI detects tuberculosis from smartphone photos of x-ray films. Tested in Ethiopia and Guinea-Bissau.
Not digital x-rays—photos of physical films taken with a smartphone.
Accuracy: 92% digital, 91% smartphone photos.
TB diagnosis migrating from wealthy-country radiology departments to $200 smartphones in Ethiopia. No radiologist or expensive equipment required.
#HealthTech #GlobalHealth #DrYou #Ethiopia
buff.ly/FwBju2J
I’m devastated. A good friend and dormmate of mine, Amin Pourfarhang, has been sentenced to death after protesting in Iran 😭 He’s one of the kindest people I’ve ever known.
Please share. Visibility raises the cost for the Islamic Republic and may stop this execution!
Garth Luke retweeted
There's a scary possibility that in 50 or 75 years, much of the world's population will live in countries that basically can't take care of themselves.
Here's what we can do to avert that future.
noahpinion.blog/p/the-giant-…
Garth Luke retweeted
A single solar panel and a few kWh of storage can transform a person's life.
Full size refrigerator - 200 kWh/year
Induction cooktop - 200 kWh/year
Electric bike - 150 kWh/year
All-in-one washer/dryer - 100 kWh/year
Reverse osmosis water filter - 40 kWh/year
5 LED lights - 40 kWh/year
Laptop - 20 kWh/year
Cell phone - 5 kWh/year
Total - 755 kWh/year
Garth Luke retweeted
Could citizens’ assemblies do what parliament won’t on tax reform? tinyurl.com/mvr6t3tf
Garth Luke retweeted
If the Government is interested in tax reform, we have some ideas.
Between taxing the gas industry, fixing tax loopholes and ending fossil fuel subsidies, these changes could raise up to $63 billion, supporting better services and reducing inequality.
▶️ australiainstitute.org.au/po…






























