Hello Fellow AI Enthusiasts,

As ever, it’s been a busy week in the AI space. Today, learn about introduce a brand new beta feature in ChatGPT+, and check out the top three interesting news items for the week, from ChatGPT going to college, to new voluntary safeguards for major players including Meta, OpenAI, and Microsoft, among others.

OpenAI just launched a super cool beta feature in the paid version of ChatGPT called “Custom Instructions.” If you have GPT+, you’ll need to turn it on in your settings. If you’re still working in the free version, it should be coming your way soon.

Custom Instructions empowers users to have more control and influence over the behavior of the language model during interactions. With this feature, users can specify detailed instructions across all conversations, guiding the AI's responses to better align with their specific needs.

By providing custom instructions, users can shape the tone, style, and content of the AI's responses, making them more relevant and tailored to their intended use case. Whether it's adopting a formal or informal tone, adhering to specific guidelines, or focusing on certain topics, custom instructions enable users to steer the conversation in the direction that suits their preferences.

This feature is particularly beneficial for entrepreneurs, content creators, and businesses, as it allows them to extract more value from ChatGPT by obtaining responses that align precisely with their target audience and business goals. By setting the right context and instructions, users can create more engaging, insightful, and creative content, enhancing their overall experience with ChatGPT and maximizing its potential for various applications.

ChatGPT Goes to Harvard - and gets a better GPA than I did, #humblebrag

Recursive Data Issues In AI - As AI increasingly gets trained on synthetic data produced by other AIs, things get… weird. Cool research by Rice and Stanford scientists explained here.

Major AI Players Agree to Voluntary Restrictions - As someone with low confidence that our US Congress is up to the task of developing useful regulation in a timely manner for the AI space, it’s nice to see this unexpected news about the Biden White House working to get major US AI companies to at least agree to some voluntary safeguards. Toothless? Maybe, but it’s a start.

That’s all for this week - as always, feel free to reach out with any feedback by replying to this email!

Happy Weekend,

Caitlin