Open Challenges for AI Engineering: Simon Willison

[@afish5581]

Awesome presentation. That was so well done 👍🏻

[@aameenmd6659]

Hey, I require some help. I'm working in an organisation who has a client, they want to create a content creator agent using openai api, which will generate content in a specific format, it should have the context of the organization's personality and their vocab, and it should generate success stories, press release etc of their clients in a specific format. What approach should I use to solve this usecase. It would really mean alot if you could reply.

[@automatalearninglab]

Simon is the man! Love that dude! B)

[@embracethered]

Excellent content!

[@XEQUTE]

got damn AI engineer ! you post quality stuff at a really good pace this was posted 45 minutes ago ! ( is this event anyway affiliated to you ?)

[@aiDotEngineer]

Glad you enjoyed it! Yes, we organize this event.

[@rotteneggconcept]

A signed agreement openAi own train on data it had available to it, right sora was trained on?

[@swillison]

If you are a customer of OpenAI and you send private data directly to their API, they will not train on that data. That's a different issue from what data they are scraping from the web and using to train their core models.

[@IdPreferNot1]

This graph is way more helpful than llmsys

[@micbab-vg2mu]

Great talk- )

[@ATH42069]

what is the meaning of life? love that part

[@pythonlibrarian224]

Prompt injection not inection. As a meat based llm, I'm really good at predicting the next letter.