
Writing for @TheEconomist | More at https://nitter.cf/t.co/qPYdNie0jd
London, DC and sometimes NY
Joined February 2011
- Tweets3.9K
- Following3.5K
- Followers10.9K
- Likes7.3K
Pinned Tweet
My latest for @TheEconomist: AI writing is taking over politics.
10% of what was is said in Parliament is now AI-drafted, per Pangram. Yet more in the House of Representatives, or in Australia and Canada.
Link: economist.com/britain/2026/0…
And a 🧵 below, w/ some main offenders
The other British productivity perspective I've found helpful. Relative to pre-GFC, the revisions keep Britain middle-of-the-pack for G7 ex-US on GDP per capita, but around the top for GDP per worker or hour worked.
America, across every measure, is decisively ahead.
Here's the chart I'd been wanting to see on the British productivity fracas.
The hours-worked series behind the shift went from saying British working hours fell the least in the rich world to, just about, the most. If we believe that, would love hypotheses on what's driving it
1
180
Here's the chart I'd been wanting to see on the British productivity fracas.
The hours-worked series behind the shift went from saying British working hours fell the least in the rich world to, just about, the most. If we believe that, would love hypotheses on what's driving it
I keep seeing these takes today - that the ONS's data update wipes away the UK's weak post GFC productivity record.
But that isn't quite right.
The new hours measure improves the UK productivity growth story, but the UK's productivity level was still weak.
A🧵
Take this chart which compares the level of UK productivity with peers - the US, Germany, France, Italy.
Using the old ONS hours, the UK fell further behind Germany in the 2010s. The new hours improves that record - but the UK was still losing ground relative to Germany.
Archie Hall retweeted
Congress acted within weeks when the financial crisis and Covid happened. It is a sleeping giant
Replying to @KelseyTuoc
I also think everyone who believes this is just wrong about how fast Congress can move if they conclude that there's a real emergency.
So, so good. Not many essayists who get AI, and get literature, quite like this
because you have not heeded my warnings and learned Spanish yet, The Point has republished in English my essay on Pangram, AI writing, the third world novel, and why so many humanists have chosen ignorance instead of grappling with what AI means for literature thepointmag.com/examined-lif…
Aaaargh
20% of the Milburn Review is AI. What are the norms on this? Is it basically now just OK?
Instinctively I really dislike it but I have been trying to examine that instinct and see if it is just snobbery.
I have an article coming out with Google's AI Policy Perspectives Substack soon about the issue.
Part of me thinks that if you are going to use AI this heavily maybe there should be some more disclosure about the prompts. What was prompted for this section? Here are some bullet points about some kids, turn them into an emotive couple of paras?
Archie Hall retweeted
Personal news: I just joined @NYTOpinion! I’m starting a new AI newsletter, writing on tech for the editorial board, and appearing on podcast feeds near you.
I’ve had the best two years of independence & it was not my plan to give that up. But I believe the NYT is the best place to shape the AI conversation as it stretches beyond Silicon Valley and into homes, workplaces, laws, and lives everywhere.
There’s so, so much to write about, and my first story will be out this week.
(As I onboard, doing a bunch of background learning convos from folks around the ecosystem; also always open to tips, feedback, etc - say hi via DM or on Signal at jws.27!)
Two new hires coming to @NYTOpinion:
Jasmine Sun has become one of the most-read voices on A.I., and now she’s our newest writer.
Alexa Sonnenfeld will join as Opinion’s chief of staff as we build out the next chapter of Opinion, from video to talent.
nytco.com/press/new-hires-in…
Archie Hall retweeted
Today the Milburn NEETs report is published.
It's a long report (278 pages!), but I've looked into it and here is my quick take:
The report treats youth inactivity largely as a problem with the benefits system, rather than with the employment costs placed on employers, which are stopping them from hiring young people.
This is a huge gap, and it means that recommendations fail to address one of the main reasons that young people are struggling to get into work.
Benefits reform is needed, but the impact of increased costs on hiring, particularly following increases in interest rates in the last few years, is making it much harder for younger workers to enter the labour market.
Employers are hiring fewer people, but young people are hit particularly hard because they are the least likely to have jobs already.
The share of young people neither in work nor in education rose in every age band, including 25–29.
The result is that we end up subsidising demand to offset employment costs. This is unlikely to be effective, and fails to deal with the underlying issue.
Cutting health benefits will force more young people into job search, but they will only find work if employers can recover the cost of employment through higher revenue.
The report acknowledges this to some extent, but leans towards using subsidies and targeted support to address it. This is not only fiscally expensive, but also unlikely to meet the magnitude of the challenge.
Instead, we need a stronger stance on the cost of hiring to minimise the potential for planned increases over the coming years, to ensure that young people are not left stranded in a labour market that will struggle to find them jobs.
britishprogress.substack.com…
gov.uk/government/publicatio…
We risk a lost generation if young people aren't prioritised, author of youth unemployment report Alan Milburn tells the BBC
Follow live: bbc.in/4fAR1BT
Archie Hall retweeted
Some personal news:
I could not be more excited to share that my debut book LEAPFROG: WHAT IT WILL TAKE TO WIN THE GLOBAL INNOVATION WAR will be out Feb. 2, 2027 and is available for preorder now!
LEAPFROG is about the topics I've devoted my professional life to over the past six years: global technology competition, reindustrialization, and America's economic future. It's about why the U.S. has fallen behind China in so many critical technologies from EVs and batteries to synthetic biology and robotics--because we've been blinded by envy and hamstrung by risk aversion, chasing the last generation of technology instead of leapfrogging to the next.
It draws on my time at the White House, where I helped make policy for the most consequential tech competition in a generation...and had a front row seat to some of our blind spots and missteps. And it takes inspiration from my current role as a tech investor where I've gotten to know and work with many of the companies building the technologies that will secure America's future and seen firsthand what they need to succeed.
We're at a delicate moment for America's technological future. The economic and national security stakes of getting our policy right and ensuring we lead the world in the technologies that matter most have never been higher. At the same time, the midterms were driven in part by a wave of anti-tech backlash (in many ways deserved).
Getting this balance right has never mattered more. LEAPFROG is a blueprint for how we do it.
Nowcast of British GDP growth projects ~2% annualised growth in Q4 (macromonitor.org/).
That would put overall 2026 growth at a remarkably healthy 2.5%, on a Q4/Q4 basis.
Archie Hall retweeted
Is polling broken? No, not really. Should it be? Probably! Despite the methodological challenges, polls are more accurate than ever. My deep dive into polling misses in Brazil and America, and what they (don't) tell you about future polling errors: economist.com/international/…
Archie Hall retweeted
Great look at the absolutely crunch questions
"Can the AI boom make enough money to sustain itself?" — I work through that question in a long-read for this week's @TheEconomist.
A few of the big ideas there:
-- Model-maker run-rate revenue today is in the low hundreds of billions. What's priced in to the data centre buildout is a rise to trillions, over the next few years.
-- Which is why, despite the continued advances in capabilities, data suggesting revenue growth may be coming down a little, from providers like Ramp, is so worrying.
-- The other danger is pattern-matching too much on extraordinary economics of SaaS, which had high switching costs and a low marginal cost of additional users: which meant wide moats and big returns to scale. AI is different: models are easy to switch between, and every new user costs money to serve.
-- So far, frontier models have been far enough ahead of the pack that AI companies have been able to charge healthy margins on them. That may continue, and their lead could even deepen as training costs get high enough to become their own sort of moat, and crackdowns on distillation hurt Chinese rivals—but could go the other way too.
-- The prototypical big-spender on AI is still a venture-backed company using agentic tools for coding. To get revenue in the trillions, the rest of the economy will need to be brought along.
Lots more in the piece, do take a look:
economist.com/briefing/2026/…
A really interesting research project I had the pleasure of working on with PF Global colleagues and @ArchieHall
We find ups and downs: data center developments tied to corruption and shady deals, but overall positivity in principle to the build out.
pf.global/interactives/light…
"Can the AI boom make enough money to sustain itself?" — I work through that question in a long-read for this week's @TheEconomist.
A few of the big ideas there:
-- Model-maker run-rate revenue today is in the low hundreds of billions. What's priced in to the data centre buildout is a rise to trillions, over the next few years.
-- Which is why, despite the continued advances in capabilities, data suggesting revenue growth may be coming down a little, from providers like Ramp, is so worrying.
-- The other danger is pattern-matching too much on extraordinary economics of SaaS, which had high switching costs and a low marginal cost of additional users: which meant wide moats and big returns to scale. AI is different: models are easy to switch between, and every new user costs money to serve.
-- So far, frontier models have been far enough ahead of the pack that AI companies have been able to charge healthy margins on them. That may continue, and their lead could even deepen as training costs get high enough to become their own sort of moat, and crackdowns on distillation hurt Chinese rivals—but could go the other way too.
-- The prototypical big-spender on AI is still a venture-backed company using agentic tools for coding. To get revenue in the trillions, the rest of the economy will need to be brought along.
Lots more in the piece, do take a look:
economist.com/briefing/2026/…

































