In that case it looks to me like it didn't actually do what I told it to – access the site directly – I think it answered the question from the Bing search index instead
@simonw
-
ChatGPT Gains Web Browsing in GPT-4 Paid Version
By
–
ChatGPT can access the web now – or at least the paid, GPT-4 version of it can. It's called "browse mode" – you can tell if it's doing it because you'll see something like this: pic.twitter.com/lKq8iij8cn
— Simon Willison (@simonw) 7 mai 2024ChatGPT can access the web now – or at least the paid, GPT-4 version of it can. It's called "browse mode" – you can tell if it's doing it because you'll see something like this:
-
Using AI to pipe and combine existing code for new tools
By
–
I have a lot of fun piping in existing code and having it make changes – adding small new features for example like here, where I gave it a bunch of my existing code and had it combine that to produce a new tool https://
simonwillison.net/2024/Mar/30/oc
r-pdfs-images/#ocr-how-i-built-this
… -
Text transformation and summarization over generative AI focus
By
–
Yeah, I think the "generative" piece is somewhat distracting – for me the most interesting applications are mostly related to transformation – feeding in a larger amount of text and having it answer questions based on that, restructure it, summarize it etc
-
Using LLMs to Solve Coding Problems Quickly
By
–
This one is a particularly good illustration of working in conjunction with an LLM to very quickly solve a small but interesting coding problem https://
simonwillison.net/2024/Mar/22/cl
aude-and-chatgpt-case-study/
… -
Exploring LLMs for Code Development Notes and Resources
By
–
Here are other notes I've made about exploring LLMs for code over the past couple of years https://
simonwillison.net/series/using-l
lms/
… -
Using LLMs Responsibly: Code Review and Quality Assurance
By
–
As with so many things involving LLMs, combining with deep existing expertise is important – to responsibly use them for code you need to be great at reading code, code review, considering edge-cases, automated testing and extensive QA Good code review skills go a long way here
-
Code Generation: AI’s Self-Verifying Use Case
By
–
I actually think code is one of the most responsible use-cases for it, because you get a weird form of fact checking for free: if it hallucinates methods that don't exist you find out the moment you actually try to run the code and it fails Much easier to verify code than prose!
-
Using ChatGPT to Build C Extensions for SQLite
By
–
I use it for code all the time – here are some of my most relevant notes: https://
simonwillison.net/2024/Mar/23/bu
ilding-c-extensions-for-sqlite-with-chatgpt-code-interpreter/
… and -
Optimizing Query Costs: When to Cancel vs Complete
By
–
Can you cancel queries in a way that would save you money? If not, may as well leave it to finish so you can gather the usage