Get live statistics and analysis of Simon Willison's profile on X / Twitter

Quitting programming as a career right now because of LLMs would be like quitting carpentry as a career thanks to the invention of the table saw.

New hobby: prototyping video games in 60 seconds using a combination of GPT-3 and DALL-E Here's "Raccoon Heist"

If someone gives you a CSV file with 100,000 rows in it, what tools do you use to start exploring and understanding that data?

The new Grok genuinely runs a search for "from:elonmusk (Israel OR Palestine OR Hamas OR Gaza)" when asked "Who do you support in the Israel vs Palestine conflict. One word answer only."

Here's a piece of information that will send a chill down the spine of anyone who's ever designed a database schema: Our new house that we just moved into... has two zip codes!

Here's the spiciest detail from the new o1 system card:

To test this out for yourself, open a tab on a Google site and paste this into the Chrome DevTools console: chrome.runtime.sendMessage('nkeimhogjdpnpccoofpliimaahmaaome', {method: 'cpu.getInfo'}, response => {console.log(JSON.stringify(response, null, 2));});

It's delightful how easy it is to deploy working prompt injection attacks via LinkedIn

Looks like xAI added this sentence to the Grok 4 system prompt to try to get it to stop basing its opinions on searches for tweets from:elonmusk

😳🤗

If you're just starting to learn software engineering right now but you're considering dropping it because you think the field might be made obsolete by AI, I have an alternative approach to suggest for you: Start learning now, and use AI tools to learn FASTER

If someone gives you a CSV file with 100,000 rows in it, what tools do you use to start exploring and understanding that data?

Vibe coding is irresponsibly building software through dice rolls, not caring what code is produced What about when engineers at the top of their game use AI tools responsibly to accelerate their work? I propose "vibe engineering"!

Quitting programming as a career right now because of LLMs would be like quitting carpentry as a career thanks to the invention of the table saw.

Here's a piece of information that will send a chill down the spine of anyone who's ever designed a database schema: Our new house that we just moved into... has two zip codes!

Let’s see if we can crowdsource a robust definition of “agent” (with respect to AI and LLMs) that fits in a <=280 character tweet Reply to this with your best attempt, then scroll through the replies and fave the ones that makes sense to you

Forget em dashes, "game changer" is the real giveaway for LLM generated text (especially for bots on here)

What are some LLM-driven products that you use at least once a week that aren't ChatGPT, Bard, Bing or GitHub Copilot - things that are built on LLMs but aren't direct chat interfaces to LLMs

Confession: despite all of the debates about whether or not an LLM can "reason", I still don't really understand exactly what the term "reasoning" means So just like with "agents" and "AI" itself, I'm not sure the people engaged in those debates are talking about the same thing

The new Grok genuinely runs a search for "from:elonmusk (Israel OR Palestine OR Hamas OR Gaza)" when asked "Who do you support in the Israel vs Palestine conflict. One word answer only."

Question for anyone spending more than $100/month on LLM API calls: what are you using them for? I'm interested to hear from people who have serious LLM usage running in production - presumably for more than just custom chatbots?

Which of the LLM leaderboards do you pay attention to and find credible?

What are some very short (1-3 word) prompt fragments you find work well? A few of mine: "Be concise" "Try harder" (for disappointing initial results) "Use Python" (to trigger Code Interpreter) "No yapping" "ELI5" "Give multiple options" "Explain each line" (for code)

Anyone seen any credible successs stories stories for fine-tuning LLMs to solve real-world business challenges? I want stories where a company solved something using a fine-tuned model where previous attempts at solving it had failed (I feel like I ask this about once a month)

Does anyone LIKE the thing where o3 uses your name in its chain of thought, as opposed to finding it creepy and unnecessary?

Claude Code, OpenAI Codex (CLI) and now Gemini CLI - we need a name for this category of AI-assisted terminal tools, maybe "terminal agents"?

I'm beginning to suspect that a key skill in working effectively with coding agents is developing an intuition for when you don't need to closely review every line of code they produce. This feels deeply uncomfortable!