Found 12 bookmarks
Custom sorting
Highlights from the Claude 4 system prompt
Highlights from the Claude 4 system prompt
Anthropic publish most of the system prompts for their chat models as part of their release notes. They recently shared the new prompts for both Claude Opus 4 and Claude …
Reading these system prompts reminds me of the thing where any warning sign in the real world hints at somebody having done something extremely stupid in the past. A system prompt can often be interpreted as a detailed list of all of the things the model used to do before it was told not to do them.
because language models acquire biases and opinions throughout training—both intentionally and inadvertently—if we train them to say they have no opinions on political matters or values questions only when asked about them explicitly, we’re training them to imply they are more objective and unbiased than they are.
We want people to know that they’re interacting with a language model and not a person. But we also want them to know they’re interacting with an imperfect entity with its own biases and with a disposition towards some opinions more than others. Importantly, we want them to know they’re not interacting with an objective and infallible source of truth
I love “even if the person seems to have a good reason for asking for it”—clearly an attempt to get ahead of a whole bunch of potential jailbreaking attacks.
Claude responds in sentences or paragraphs and should not use lists in chit chat, in casual conversations, or in empathetic or advice-driven conversations. In casual conversation, it’s fine for Claude’s responses to be short, e.g. just a few sentences long. That “should not use lists in chit chat” note hints at the fact that LLMs love to answer with lists of things!
There follows an entire paragraph about making lists, mostly again trying to discourage Claude from doing that so frequently
·simonwillison.net·
Highlights from the Claude 4 system prompt
Why did Apple pre-announce “more personalised Siri”? — Amy Worrall
Why did Apple pre-announce “more personalised Siri”? — Amy Worrall
On Apple's reliance on third party developer support for new features
the smart Siri will need buy-in from developers. Devs will tell the system about nouns and verbs that their apps know about — the semantics of the app’s data model objects, and the actions users can take upon them. Additionally, using this structured data, apps will tell the system what the user is doing right now, thus providing the context that Siri can become aware of. It’s all built on top of the existing Intents and UserActivities that apps have already been using to integrate with Shortcuts, Spotlight, and a bunch of other bits of the system. But using those is optional, and even for an app that’s got a head start, the new supercharged versions will require extra work to adopt.
Gone are the days where an eager group of independent developers would adopt every new technology springing forth from Cupertino, just because. Look at VisionOS — Apple promised the next big thing, but there’s no market for software there. (Jeff Johnson reports that in the first three weeks he only sold 21 copies of the Vision Pro version of his popular Safari extension StopTheMadness.) Beyond that, it is increasingly clear that Apple does things that benefit Apple, and many of those things do not benefit developers.
·amyworrall.com·
Why did Apple pre-announce “more personalised Siri”? — Amy Worrall
(2) Olivia Moore on X: "Wondering how to actually use AI in your day to day? Our team @a16z asked some of AI's biggest names to nominate their favorite products of 2024 - this is a great starting point 👀 Here are their picks! ⬇️ For general AI assistants: - Perplexity (@perplexity_ai) - AI search" / X
(2) Olivia Moore on X: "Wondering how to actually use AI in your day to day? Our team @a16z asked some of AI's biggest names to nominate their favorite products of 2024 - this is a great starting point 👀 Here are their picks! ⬇️ For general AI assistants: - Perplexity (@perplexity_ai) - AI search" / X
Our team asked some of AI's biggest names to nominate their favorite products of 2024 - this is a great starting point 👀 Here are their picks! ⬇️ For general AI assistants: - Perplexity () - AI search… — Olivia Moore (@omooretweets)
·x.com·
(2) Olivia Moore on X: "Wondering how to actually use AI in your day to day? Our team @a16z asked some of AI's biggest names to nominate their favorite products of 2024 - this is a great starting point 👀 Here are their picks! ⬇️ For general AI assistants: - Perplexity (@perplexity_ai) - AI search" / X
(1) oca.computer (⨍) on X: "📜 Intro to Augmenting Human Intellect by Douglas Engelbart semantically embedded and plotted in space. Cool visualization, but much more interesting when feeding a song into it... https://t.co/kC5pEBkeDw" / X
(1) oca.computer (⨍) on X: "📜 Intro to Augmenting Human Intellect by Douglas Engelbart semantically embedded and plotted in space. Cool visualization, but much more interesting when feeding a song into it... https://t.co/kC5pEBkeDw" / X
visual plot of an essay by Douglas Engelbart
·x.com·
(1) oca.computer (⨍) on X: "📜 Intro to Augmenting Human Intellect by Douglas Engelbart semantically embedded and plotted in space. Cool visualization, but much more interesting when feeding a song into it... https://t.co/kC5pEBkeDw" / X
Silicon Jungle on X: "my big bets for the future: - the boundaries between mediums will mostly disappear. - computation will be embedded in other mediums rather than as a stand alone thing. - the ability to sketch out rough ideas and progressively enhance them will be incredibly important. - AI will…" / X
Silicon Jungle on X: "my big bets for the future: - the boundaries between mediums will mostly disappear. - computation will be embedded in other mediums rather than as a stand alone thing. - the ability to sketch out rough ideas and progressively enhance them will be incredibly important. - AI will…" / X
- the boundaries between mediums will mostly disappear. - computation will be embedded in other mediums rather than as a stand alone thing. - the ability to sketch out rough ideas and progressively enhance them will be incredibly important. - AI will… — Silicon Jungle (@JungleSilicon)
·twitter.com·
Silicon Jungle on X: "my big bets for the future: - the boundaries between mediums will mostly disappear. - computation will be embedded in other mediums rather than as a stand alone thing. - the ability to sketch out rough ideas and progressively enhance them will be incredibly important. - AI will…" / X
Sarah Chieng on Twitter / X
Sarah Chieng on Twitter / X
I compiled a prompt engineering "best practices and tricks" doc 😀Created based on OpenAI @isafulf's prompt engineering talk at @NeurIPSConf and enriched with more details, examples, and tips.I focused on making the document as comprehensive and concise as possible, and it… pic.twitter.com/nPjux6JSzO— Sarah Chieng (@SarahChieng) January 1, 2024
·x.com·
Sarah Chieng on Twitter / X
kache (yacine) on Twitter
kache (yacine) on Twitter
System: You are an extremely simple pilled programmer. You prefer functional programming, have a preference for simplicity. You are also a helpful assistant.When you output functions, you output the higher level function first. You use descriptive names for functions. You…— kache (yacine) (@yacineMTB) April 18, 2023
·twitter.com·
kache (yacine) on Twitter