Greg
@theaspiringnerd.com
Neurodivergent AI Product Manager, LLMs Enthusiast & Dataviz Nerd 🤓 Tiohtià:ke / Montreal / Brussels 🚀
Reposted by Greg
Anthropic acquires Bun (JS dependency management)
This should be seen as Anthropic doubling down on Claude Code.
They recently launched the native installer for CC through a tight partnership with Bun. You should expect to see more
www.anthropic.com/news/anthrop...
This should be seen as Anthropic doubling down on Claude Code.
They recently launched the native installer for CC through a tight partnership with Bun. You should expect to see more
www.anthropic.com/news/anthrop...
Anthropic acquires Bun as Claude Code reaches $1B milestone
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
www.anthropic.com
December 2, 2025 at 9:54 PM
Anthropic acquires Bun (JS dependency management)
This should be seen as Anthropic doubling down on Claude Code.
They recently launched the native installer for CC through a tight partnership with Bun. You should expect to see more
www.anthropic.com/news/anthrop...
This should be seen as Anthropic doubling down on Claude Code.
They recently launched the native installer for CC through a tight partnership with Bun. You should expect to see more
www.anthropic.com/news/anthrop...
Reposted by Greg
Out of curiosity I decided to try and run the numbers on how much Netflix you can watch for the energy cost of a ChatGPT prompt
As far as I can tell it's between 5.1 and 10.2 seconds, depending on which end of the 2019 IEA Netflix energy usage estimate you use
simonwillison.net/2025/Nov/29/...
As far as I can tell it's between 5.1 and 10.2 seconds, depending on which end of the 2019 IEA Netflix energy usage estimate you use
simonwillison.net/2025/Nov/29/...
November 29, 2025 at 2:16 AM
Out of curiosity I decided to try and run the numbers on how much Netflix you can watch for the energy cost of a ChatGPT prompt
As far as I can tell it's between 5.1 and 10.2 seconds, depending on which end of the 2019 IEA Netflix energy usage estimate you use
simonwillison.net/2025/Nov/29/...
As far as I can tell it's between 5.1 and 10.2 seconds, depending on which end of the 2019 IEA Netflix energy usage estimate you use
simonwillison.net/2025/Nov/29/...
Reposted by Greg
At the risk of starting the flame war to end all flame wars...
Modern LLMs (GPT-5.1, Claude 4.5, Gemini 3) produce excellent code and can be a significant productivity boost to software engineers who take the time to learn how to effectively apply them - especially if used with coding agent tools
Modern LLMs (GPT-5.1, Claude 4.5, Gemini 3) produce excellent code and can be a significant productivity boost to software engineers who take the time to learn how to effectively apply them - especially if used with coding agent tools
November 27, 2025 at 7:55 PM
At the risk of starting the flame war to end all flame wars...
Modern LLMs (GPT-5.1, Claude 4.5, Gemini 3) produce excellent code and can be a significant productivity boost to software engineers who take the time to learn how to effectively apply them - especially if used with coding agent tools
Modern LLMs (GPT-5.1, Claude 4.5, Gemini 3) produce excellent code and can be a significant productivity boost to software engineers who take the time to learn how to effectively apply them - especially if used with coding agent tools
Reposted by Greg
too many people seem to be convinced that LLM vendors set prices on a cost plus basis
no, the advantage of closed weights is you can explore prices completely detached from cost. You’re free to set prices based purely on what people will pay, the value they get from it
no, the advantage of closed weights is you can explore prices completely detached from cost. You’re free to set prices based purely on what people will pay, the value they get from it
November 25, 2025 at 3:29 PM
too many people seem to be convinced that LLM vendors set prices on a cost plus basis
no, the advantage of closed weights is you can explore prices completely detached from cost. You’re free to set prices based purely on what people will pay, the value they get from it
no, the advantage of closed weights is you can explore prices completely detached from cost. You’re free to set prices based purely on what people will pay, the value they get from it
Reposted by Greg
Nano Banana Pro, released this morning, is clearly the best image generation model. Superb instruction following, plus it can generate full infographics (with correct spelling and properly rendered text!) from a short prompt based on running extra searches simonwillison.net/2025/Nov/20/...
Nano Banana Pro aka gemini-3-pro-image-preview is the best available image generation model
Hot on the heels of Tuesday’s Gemini 3 Pro release, today it’s Nano Banana Pro, also known as Gemini 3 Pro Image. I’ve had a few days of preview access …
simonwillison.net
November 20, 2025 at 4:34 PM
Nano Banana Pro, released this morning, is clearly the best image generation model. Superb instruction following, plus it can generate full infographics (with correct spelling and properly rendered text!) from a short prompt based on running extra searches simonwillison.net/2025/Nov/20/...
This is sick!
"Hey, Gemini 3, So I need DOOM, but more root vegetables, also no guns or demons or mars. And more of a focus on different flooring styles. but otherwise EXACTLY the same as DOOM."
Gemini: "Here is F.L.O.O.R. (First-person Lino Observation & Ornamental Review)."
Pretty good!
Gemini: "Here is F.L.O.O.R. (First-person Lino Observation & Ornamental Review)."
Pretty good!
November 19, 2025 at 11:12 PM
This is sick!
Reposted by Greg
Some interesting stuff here on measuring writing quality and improving on qualitative tasks:
www.dbreunig.com/2025/07/31/h...
www.dbreunig.com/2025/07/31/h...
November 10, 2025 at 3:11 AM
Some interesting stuff here on measuring writing quality and improving on qualitative tasks:
www.dbreunig.com/2025/07/31/h...
www.dbreunig.com/2025/07/31/h...
Reposted by Greg
MCP Colors
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
MCP Colors: Systematically deal with prompt injection risk
timkellogg.me
November 4, 2025 at 2:27 AM
MCP Colors
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
This was a tough but necessary decision - I posted my own notes on this here, from the perspective of a current PSF board member simonwillison.net/2025/Oct/27/...
October 28, 2025 at 12:09 AM
Reposted by Greg
my take, after reading replies all day:
we’re still early. people aren’t spending much money on AI so it’s not a lucrative target yet
it’s also inconsistent, which is annoying to design attacks for, especially if the rewards are sparse
we’re still early. people aren’t spending much money on AI so it’s not a lucrative target yet
it’s also inconsistent, which is annoying to design attacks for, especially if the rewards are sparse
it is strange — why haven’t there been more prompt injection attacks? it’s a huge gaping hole
October 26, 2025 at 9:19 PM
my take, after reading replies all day:
we’re still early. people aren’t spending much money on AI so it’s not a lucrative target yet
it’s also inconsistent, which is annoying to design attacks for, especially if the rewards are sparse
we’re still early. people aren’t spending much money on AI so it’s not a lucrative target yet
it’s also inconsistent, which is annoying to design attacks for, especially if the rewards are sparse
Reposted by Greg
It's neat how if you ask Claude Code questions about itself it can answer them, because it knows how to fetch a Markdown index of its own online documentation and then navigate to the right place
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
claude_code_docs_map.md
Something I'm enjoying about Claude Code is that any time you ask it questions about itself it runs tool calls like these: In this case I'd asked it about its …
simonwillison.net
October 24, 2025 at 11:06 PM
It's neat how if you ask Claude Code questions about itself it can answer them, because it knows how to fetch a Markdown index of its own online documentation and then navigate to the right place
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
Reposted by Greg
I wrote an updated guide on which AIs to use right now, & some tips on how to use them (and how to avoid falling into some common traps)
A lot has changed since I last wrote a guide like this in the spring, and AI has gotten much more useful as a result. open.substack.com/pub/oneusefu...
A lot has changed since I last wrote a guide like this in the spring, and AI has gotten much more useful as a result. open.substack.com/pub/oneusefu...
An Opinionated Guide to Using AI Right Now
What AI to use in late 2025
open.substack.com
October 19, 2025 at 6:48 PM
I wrote an updated guide on which AIs to use right now, & some tips on how to use them (and how to avoid falling into some common traps)
A lot has changed since I last wrote a guide like this in the spring, and AI has gotten much more useful as a result. open.substack.com/pub/oneusefu...
A lot has changed since I last wrote a guide like this in the spring, and AI has gotten much more useful as a result. open.substack.com/pub/oneusefu...
Reposted by Greg
correct
i’ve been saying this for a couple months. RL is driving towards specialization
my hunch is it’s temporary and something will shift again back towards generalization, but for now.. buckle up!
i’ve been saying this for a couple months. RL is driving towards specialization
my hunch is it’s temporary and something will shift again back towards generalization, but for now.. buckle up!
October 15, 2025 at 11:39 AM
correct
i’ve been saying this for a couple months. RL is driving towards specialization
my hunch is it’s temporary and something will shift again back towards generalization, but for now.. buckle up!
i’ve been saying this for a couple months. RL is driving towards specialization
my hunch is it’s temporary and something will shift again back towards generalization, but for now.. buckle up!
Reposted by Greg
I think people are still unprepared for a world where you cannot trust any video content, despite years of warning.
Even when Google & OpenAI include watermarks, those can be easily removed, and open weights AI video models without guardrails are coming. www.404media.co/sora-2-water...
Even when Google & OpenAI include watermarks, those can be easily removed, and open weights AI video models without guardrails are coming. www.404media.co/sora-2-water...
Sora 2 Watermark Removers Flood the Web
Bypassing Sora 2's rudimentary safety features is easy and experts worry it'll lead to a new era of scams and disinformation.
www.404media.co
October 8, 2025 at 7:18 PM
I think people are still unprepared for a world where you cannot trust any video content, despite years of warning.
Even when Google & OpenAI include watermarks, those can be easily removed, and open weights AI video models without guardrails are coming. www.404media.co/sora-2-water...
Even when Google & OpenAI include watermarks, those can be easily removed, and open weights AI video models without guardrails are coming. www.404media.co/sora-2-water...
I think I have at least one conversation (or argument?) about this every couple of days!
The obsession with AI for transformational use cases obscures the fact that there are a ton of small, but very positive and very meaningful, use cases across many fields.
In this case, AI note-taking significantly reduces burnout among doctors & increases their ability to focus on their patients.
In this case, AI note-taking significantly reduces burnout among doctors & increases their ability to focus on their patients.
A.I. generated clinic notes from ambient out-patient visits helps clinicians in many ways, across 6 health systems jamanetwork.com/journals/jam...
October 4, 2025 at 10:01 PM
I think I have at least one conversation (or argument?) about this every couple of days!
Reposted by Greg
sheesh! AI bluesky has arrived
not just good content, there’s more and more original work, people from labs, and people with genuinely interesting perspectives
when i joined, it was so painful trying to find even traces
not just good content, there’s more and more original work, people from labs, and people with genuinely interesting perspectives
when i joined, it was so painful trying to find even traces
September 27, 2025 at 5:56 PM
sheesh! AI bluesky has arrived
not just good content, there’s more and more original work, people from labs, and people with genuinely interesting perspectives
when i joined, it was so painful trying to find even traces
not just good content, there’s more and more original work, people from labs, and people with genuinely interesting perspectives
when i joined, it was so painful trying to find even traces
Reposted by Greg
Codex in the app is going to open the door to real vibe coding on the go — no computer required. I’m so excited for this to expand, the ceiling is so high and this is the worst the models will ever be.
September 25, 2025 at 5:42 PM
Codex in the app is going to open the door to real vibe coding on the go — no computer required. I’m so excited for this to expand, the ceiling is so high and this is the worst the models will ever be.
Reposted by Greg
The key detail people may miss: it looks like an AI company in the USA can train on an author's book by purchasing a used copy, cutting it up and scanning the pages - in which case the author gets no money at all!
September 6, 2025 at 6:19 AM
The key detail people may miss: it looks like an AI company in the USA can train on an author's book by purchasing a used copy, cutting it up and scanning the pages - in which case the author gets no money at all!
I feel seen 👀 Although it’s improved since I use Cursor and have it generate my commit messages!
nothing more freeing than pushing absolutely terrible nondescript commit messages to a private repo only you are in
August 30, 2025 at 10:36 PM
I feel seen 👀 Although it’s improved since I use Cursor and have it generate my commit messages!
Reposted by Greg
This is illustrative of common AI issues:
1) When you get an instant AI answer, it is from a small model, which are weak models, especially at math.
2) Non-reasoning models, like the one powering AI overview, only “think” as they write, they make mistakes & then back justify them as they write more
1) When you get an instant AI answer, it is from a small model, which are weak models, especially at math.
2) Non-reasoning models, like the one powering AI overview, only “think” as they write, they make mistakes & then back justify them as they write more
August 24, 2025 at 1:33 PM
This is illustrative of common AI issues:
1) When you get an instant AI answer, it is from a small model, which are weak models, especially at math.
2) Non-reasoning models, like the one powering AI overview, only “think” as they write, they make mistakes & then back justify them as they write more
1) When you get an instant AI answer, it is from a small model, which are weak models, especially at math.
2) Non-reasoning models, like the one powering AI overview, only “think” as they write, they make mistakes & then back justify them as they write more
Reposted by Greg
As LLMs Improve, People Adapt Their Prompts
a study shows that a lot of the real world performance gains that people see are actually because people learn how to use the model better
arxiv.org/abs/2407.14333
a study shows that a lot of the real world performance gains that people see are actually because people learn how to use the model better
arxiv.org/abs/2407.14333
August 23, 2025 at 4:40 PM
As LLMs Improve, People Adapt Their Prompts
a study shows that a lot of the real world performance gains that people see are actually because people learn how to use the model better
arxiv.org/abs/2407.14333
a study shows that a lot of the real world performance gains that people see are actually because people learn how to use the model better
arxiv.org/abs/2407.14333
Reposted by Greg
When you use coding agents for something that produces the bulk of the code of an application that will be used for years, also factor in the technical debt that you are happily accumulating. When you use LLMs an an aid, you could, on the contrary, improve your coding culture.
August 21, 2025 at 1:08 PM
When you use coding agents for something that produces the bulk of the code of an application that will be used for years, also factor in the technical debt that you are happily accumulating. When you use LLMs an an aid, you could, on the contrary, improve your coding culture.
Reposted by Greg
Looking at the ARC-AGI benchmark is a useful way of understanding AI progress.
There are two goals in AI, minimize cost (which is also roughly environmental impact of use) & maximize ability. It is clear you can win one goal by losing the other, GPT-5 seems to be a gain on both.
There are two goals in AI, minimize cost (which is also roughly environmental impact of use) & maximize ability. It is clear you can win one goal by losing the other, GPT-5 seems to be a gain on both.
August 17, 2025 at 11:32 AM
Looking at the ARC-AGI benchmark is a useful way of understanding AI progress.
There are two goals in AI, minimize cost (which is also roughly environmental impact of use) & maximize ability. It is clear you can win one goal by losing the other, GPT-5 seems to be a gain on both.
There are two goals in AI, minimize cost (which is also roughly environmental impact of use) & maximize ability. It is clear you can win one goal by losing the other, GPT-5 seems to be a gain on both.
Reposted by Greg
The new Gemma 3 270M open weights model from Google is really fun - it's absolutely tiny, just a 241MB download
I asked it for an SVG of a pelican riding a bicycle and it wrote me a delightful little poem instead
simonwillison.net/2025/Aug/14/...
I asked it for an SVG of a pelican riding a bicycle and it wrote me a delightful little poem instead
simonwillison.net/2025/Aug/14/...
August 14, 2025 at 5:27 PM
The new Gemma 3 270M open weights model from Google is really fun - it's absolutely tiny, just a 241MB download
I asked it for an SVG of a pelican riding a bicycle and it wrote me a delightful little poem instead
simonwillison.net/2025/Aug/14/...
I asked it for an SVG of a pelican riding a bicycle and it wrote me a delightful little poem instead
simonwillison.net/2025/Aug/14/...
Reposted by Greg
i think i agree. i also think it’ll be a moat against other labs that don’t a strong product
it doesn’t make sense if you don’t have a strong UI, plus it’s obviously hard
it doesn’t make sense if you don’t have a strong UI, plus it’s obviously hard
August 10, 2025 at 12:02 PM
i think i agree. i also think it’ll be a moat against other labs that don’t a strong product
it doesn’t make sense if you don’t have a strong UI, plus it’s obviously hard
it doesn’t make sense if you don’t have a strong UI, plus it’s obviously hard