Max Little
@maxal.bsky.social
Academic mathematician/computer scientist, University of Birmingham, UK. AI and machine learning, causal inference, signal processing, applied mathematics, computational statistics. Ex Oxford PhD, MIT postdoc fellow.
Reposted by Max Little
Why we should rethink causal mediation, and what to do instead? Come to hear the answer from Vanessa Didelez at the next CIIG seminar!
The seminar will be hybrid. If you are in London, come join us in person at UCL! Otherwise, you can join on Zoom as usual. Registration links in comment below.
The seminar will be hybrid. If you are in London, come join us in person at UCL! Otherwise, you can join on Zoom as usual. Registration links in comment below.
May 13, 2026 at 3:05 PM
Why we should rethink causal mediation, and what to do instead? Come to hear the answer from Vanessa Didelez at the next CIIG seminar!
The seminar will be hybrid. If you are in London, come join us in person at UCL! Otherwise, you can join on Zoom as usual. Registration links in comment below.
The seminar will be hybrid. If you are in London, come join us in person at UCL! Otherwise, you can join on Zoom as usual. Registration links in comment below.
Reposted by Max Little
Does engaging with the arts help to slow the aging process? According to this new paper from Fancourt et al the answer is yes!
Alas, the paper is actually unsuitable for such claims.
Let’s examine by discussing how you'd establish whether arts engagement actually caused future health!
1/18 🧵
Alas, the paper is actually unsuitable for such claims.
Let’s examine by discussing how you'd establish whether arts engagement actually caused future health!
1/18 🧵
Does leisure activity matter for epigenetic ageing? Analyses of arts engagement and physical activity in the UK Household Longitudinal Study
AbstractBackground and Objectives. Over the past decade, ageing clocks have become widely adopted as important tools for understanding biological ageing an
academic.oup.com
May 12, 2026 at 8:32 PM
Does engaging with the arts help to slow the aging process? According to this new paper from Fancourt et al the answer is yes!
Alas, the paper is actually unsuitable for such claims.
Let’s examine by discussing how you'd establish whether arts engagement actually caused future health!
1/18 🧵
Alas, the paper is actually unsuitable for such claims.
Let’s examine by discussing how you'd establish whether arts engagement actually caused future health!
1/18 🧵
Reposted by Max Little
This recent RCT of an "AI stethoscope" claims the technology "shows promise" for diagnosing cardiovascular conditions.
It does not.
It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed.
🧵 1/
It does not.
It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed.
🧵 1/
AI-enabled stethoscopes show promise for improving diagnosis of cardiovascular conditions, UK trial finds.
Explore the research: spkl.io/63326AsjJ8
Explore the research: spkl.io/63326AsjJ8
February 25, 2026 at 4:44 PM
This recent RCT of an "AI stethoscope" claims the technology "shows promise" for diagnosing cardiovascular conditions.
It does not.
It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed.
🧵 1/
It does not.
It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed.
🧵 1/
Reposted by Max Little
Depends on how you spin it I guess (screenshots from 2 different articles).
People working at universities are pushed so incredibly hard to ensure that every study is a breakthrough that they just...lie. All the time. Probably without even realizing it.
www.standard.co.uk/news/tech/im...
People working at universities are pushed so incredibly hard to ensure that every study is a breakthrough that they just...lie. All the time. Probably without even realizing it.
www.standard.co.uk/news/tech/im...
September 2, 2025 at 4:15 AM
Depends on how you spin it I guess (screenshots from 2 different articles).
People working at universities are pushed so incredibly hard to ensure that every study is a breakthrough that they just...lie. All the time. Probably without even realizing it.
www.standard.co.uk/news/tech/im...
People working at universities are pushed so incredibly hard to ensure that every study is a breakthrough that they just...lie. All the time. Probably without even realizing it.
www.standard.co.uk/news/tech/im...
Reposted by Max Little
Every Jan 1 I post a scorecard on predictions I made, with dates, on Jan 1, 2018 on cars (self-driving), robots, AI, & ML, and on human spaceflight. Besides telling which turned out right and which wrong in the last year I also talk a lot of smack about these topics. rodneybrooks.com/predictions-...
Predictions Scorecard, 2025 January 01 – Rodney Brooks
rodneybrooks.com
January 1, 2025 at 7:37 AM
Every Jan 1 I post a scorecard on predictions I made, with dates, on Jan 1, 2018 on cars (self-driving), robots, AI, & ML, and on human spaceflight. Besides telling which turned out right and which wrong in the last year I also talk a lot of smack about these topics. rodneybrooks.com/predictions-...
Reposted by Max Little
LLMs can be also seen as big bags of heuristics: arxiv.org/abs/2410.21272
Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics
Do large language models (LLMs) solve reasoning tasks by learning robust generalizable algorithms, or do they memorize training data? To investigate this question, we use arithmetic reasoning as a rep...
arxiv.org
December 21, 2024 at 5:46 PM
LLMs can be also seen as big bags of heuristics: arxiv.org/abs/2410.21272
Reposted by Max Little
Reposted by Max Little
𝗢𝟯 𝘄𝗮𝘀 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝗻 𝟳𝟱% 𝗼𝗳 𝘁𝗵𝗲 𝗽𝘂𝗯𝗹𝗶𝗰 𝘀𝗲𝘁 𝗳𝗼𝗿 𝗔𝗥𝗖-𝗔𝗚𝗜.
OpenAI did not disclose this in the video. Sam said they didn’t target the test.
Never trust a staged demo.
Never trust a product you haven’t tried.
Never trust OpenAI.
OpenAI did not disclose this in the video. Sam said they didn’t target the test.
Never trust a staged demo.
Never trust a product you haven’t tried.
Never trust OpenAI.
December 21, 2024 at 10:06 PM
𝗢𝟯 𝘄𝗮𝘀 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝗻 𝟳𝟱% 𝗼𝗳 𝘁𝗵𝗲 𝗽𝘂𝗯𝗹𝗶𝗰 𝘀𝗲𝘁 𝗳𝗼𝗿 𝗔𝗥𝗖-𝗔𝗚𝗜.
OpenAI did not disclose this in the video. Sam said they didn’t target the test.
Never trust a staged demo.
Never trust a product you haven’t tried.
Never trust OpenAI.
OpenAI did not disclose this in the video. Sam said they didn’t target the test.
Never trust a staged demo.
Never trust a product you haven’t tried.
Never trust OpenAI.
Reposted by Max Little
o3, AGI, and the art of the demo. Long read on what OpenAI didn’t tell you yesterday. garymarcus.substack.com/p/o3-agi-the...
o3, AGI, the art of the demo, and what you can expect in 2025
OpenAI’s new model was revealed yesterday; its most fervent believers think AGI has already arrived. Here’s what you should pay attention to in the coming year.
garymarcus.substack.com
December 21, 2024 at 3:31 PM
o3, AGI, and the art of the demo. Long read on what OpenAI didn’t tell you yesterday. garymarcus.substack.com/p/o3-agi-the...
Likewise, a simple adversarial strategy beats "superhuman" Go-playing algorithms: goattack.far.ai It's wise to remember that there is no scientific consensus on what "intelligence", actually is.
December 21, 2024 at 6:31 PM
Likewise, a simple adversarial strategy beats "superhuman" Go-playing algorithms: goattack.far.ai It's wise to remember that there is no scientific consensus on what "intelligence", actually is.
Just for those who don't know: the vast majority of open problems in maths, are not numerical in nature.
The questions have numerical answers, so it is easy to check whether it gets them right.
December 21, 2024 at 11:55 AM
Just for those who don't know: the vast majority of open problems in maths, are not numerical in nature.
Reposted by Max Little
The questions have numerical answers, so it is easy to check whether it gets them right.
December 21, 2024 at 9:17 AM
The questions have numerical answers, so it is easy to check whether it gets them right.
Reposted by Max Little
How many times do we have to see this same movie, where an AI beats some benchmark and influencers gleefully shout “It’s So Over” without even trying out the AI and then on careful inspection the AI turns out to not be robust or reliable?
Thousands?
(It’s already been hundreds.)
Thousands?
(It’s already been hundreds.)
December 21, 2024 at 12:59 AM
How many times do we have to see this same movie, where an AI beats some benchmark and influencers gleefully shout “It’s So Over” without even trying out the AI and then on careful inspection the AI turns out to not be robust or reliable?
Thousands?
(It’s already been hundreds.)
Thousands?
(It’s already been hundreds.)
Reposted by Max Little
It seems that OpenAI's latest model, o3, can solve 25% of problems on a database called FrontierMath, created by EpochAI, where previous LLMs could only solve 2%. On Twitter I am quoted as saying, "Getting even one question right would be well beyond what we can do now, let alone saturating them."
December 20, 2024 at 11:15 PM
It seems that OpenAI's latest model, o3, can solve 25% of problems on a database called FrontierMath, created by EpochAI, where previous LLMs could only solve 2%. On Twitter I am quoted as saying, "Getting even one question right would be well beyond what we can do now, let alone saturating them."
Reposted by Max Little
It's widely agreed that scholars are supposed to say when they use ChatGPT. Yet phrases like "I am an AI language model"—with no disclosure—are popping up in papers.
I wrote about how journals seemingly aren't enforcing their AI policies, according to a new study: www.chronicle.com/article/scho...
I wrote about how journals seemingly aren't enforcing their AI policies, according to a new study: www.chronicle.com/article/scho...
Scholars Are Supposed to Say When They Use AI. Do They?
Journals have policies about disclosing ChatGPT writing, but enforcing them is another matter, according to a new study.
www.chronicle.com
December 18, 2024 at 9:02 PM
It's widely agreed that scholars are supposed to say when they use ChatGPT. Yet phrases like "I am an AI language model"—with no disclosure—are popping up in papers.
I wrote about how journals seemingly aren't enforcing their AI policies, according to a new study: www.chronicle.com/article/scho...
I wrote about how journals seemingly aren't enforcing their AI policies, according to a new study: www.chronicle.com/article/scho...
Reposted by Max Little
This seems like a pretty balanced commentary. They certainly get this right: "connection between capability improvements & AI’s social or economic impacts is extremely weak. The bottlenecks for impact are the pace of product development and the rate of adoption" www.aisnakeoil.com/p/is-ai-prog...
Is AI progress slowing down?
Making sense of recent technology trends and claims
www.aisnakeoil.com
December 18, 2024 at 8:46 PM
This seems like a pretty balanced commentary. They certainly get this right: "connection between capability improvements & AI’s social or economic impacts is extremely weak. The bottlenecks for impact are the pace of product development and the rate of adoption" www.aisnakeoil.com/p/is-ai-prog...
Good reporting here, but sadly, these tragedies were predictable. Those of us who actually work on machine learning know that deep-learning based computer vision simply isn't reliable enough for safety-critical applications such as self-driving cars. @garymarcus.bsky.social @filippie509.bsky.social
CW: car crashes with fatal outcomes.
The Hidden Autopilot Data That Reveals Why Teslas Crash | WSJ
”Computer vision is such a deeply flawed technology” – Missy Cummings, fighter pilot and director of George Mason University's Autonomy and Robotics Center
YouTube: per.ax/aptesla
The Hidden Autopilot Data That Reveals Why Teslas Crash | WSJ
”Computer vision is such a deeply flawed technology” – Missy Cummings, fighter pilot and director of George Mason University's Autonomy and Robotics Center
YouTube: per.ax/aptesla
December 17, 2024 at 4:09 PM
Good reporting here, but sadly, these tragedies were predictable. Those of us who actually work on machine learning know that deep-learning based computer vision simply isn't reliable enough for safety-critical applications such as self-driving cars. @garymarcus.bsky.social @filippie509.bsky.social
Reposted by Max Little
The late Suchir Balaji’s blog post on AI, copyright and fair use, reposted in his memory.
suchir.net/fair_use.html
suchir.net/fair_use.html
When does generative AI qualify for fair use?
suchir.net
December 14, 2024 at 6:07 AM
The late Suchir Balaji’s blog post on AI, copyright and fair use, reposted in his memory.
suchir.net/fair_use.html
suchir.net/fair_use.html
Reposted by Max Little
The bootstrap can be used to generate a new random sample from an existing random sample. Its validity can be guaranteed by the Glivenko-Cantelli theorem, which demonstrates how the empirical cumulative distribution (CDF, top panel), converges on the CDF of the sample (bottom panel).
December 14, 2024 at 12:08 PM
The bootstrap can be used to generate a new random sample from an existing random sample. Its validity can be guaranteed by the Glivenko-Cantelli theorem, which demonstrates how the empirical cumulative distribution (CDF, top panel), converges on the CDF of the sample (bottom panel).
Reposted by Max Little
For an increasing function 𝑓:ℝ→ℝ, max(𝑓(𝑎),𝑓(𝑏))=𝑓(max(𝑎,𝑏)). An important special case is 𝑓(𝑥)=𝑥+𝑐, for which we obtain max(𝑎+𝑐,𝑏+𝑐)=𝑐+max(𝑎,𝑏).
December 14, 2024 at 12:23 AM
For an increasing function 𝑓:ℝ→ℝ, max(𝑓(𝑎),𝑓(𝑏))=𝑓(max(𝑎,𝑏)). An important special case is 𝑓(𝑥)=𝑥+𝑐, for which we obtain max(𝑎+𝑐,𝑏+𝑐)=𝑐+max(𝑎,𝑏).
Reposted by Max Little
Since 2016 Waymo raised ~$25B, so they burn ~$3B/y or little over 8mln/day. With ~700 cars, assuming they operate each car every day, it costs them over 11k dollars to operate each of their cars per day. $11k PER DAY per CAR. If you don't find this ridiculous IDK what else to say.
December 14, 2024 at 12:04 AM
Since 2016 Waymo raised ~$25B, so they burn ~$3B/y or little over 8mln/day. With ~700 cars, assuming they operate each car every day, it costs them over 11k dollars to operate each of their cars per day. $11k PER DAY per CAR. If you don't find this ridiculous IDK what else to say.
Reposted by Max Little
Suchir Balaji was a good young man. I spoke to him six weeks ago. He had left OpenAI and wanted to make the world a better place. This is tragic.
December 14, 2024 at 12:19 AM
Suchir Balaji was a good young man. I spoke to him six weeks ago. He had left OpenAI and wanted to make the world a better place. This is tragic.
Very proud of the Birmingham HDRUK PhDs!
Health Data Research UK PhD meet! Work from Ant Lee and Jianqiao Mao (latter with @maxal.bsky.social)
December 13, 2024 at 11:16 PM
Very proud of the Birmingham HDRUK PhDs!
Reposted by Max Little
Health Data Research UK PhD meet! Work from Ant Lee and Jianqiao Mao (latter with @maxal.bsky.social)
December 13, 2024 at 11:26 AM
Health Data Research UK PhD meet! Work from Ant Lee and Jianqiao Mao (latter with @maxal.bsky.social)
Apple "Intelligence". @garymarcus.bsky.social
As I said -- only a matter of time before news orgs lashed out at Apple's lackluster (and evidently dangerously wrong) AI summaries of their work. This is shambolic and the BBC has every right to be furious about it.
BBC complains to Apple over misleading shooting headline
Apple's new artificial intelligence features falsely made it seem the BBC reported Luigi Mangione had shot himself.
www.bbc.com
December 13, 2024 at 10:54 PM
Apple "Intelligence". @garymarcus.bsky.social