😺 Prompt engineering is overrated

PLUS: a piece of happy news to round out the week!

Welcome, humans.

Every tech company repackaging as an AI company akin to someone unemployed changing their LinkedIn status to “working on something new.” 😹

Here’s what’s going on in the world of AI today:

ChatGPT is shipping features to make prompting easier.
AI is better at detecting breast cancer.
Elon Musk bought AI.com from OpenAI.
Data centers that support AI applications are in high demand.

Sponsor The Neuron | New Here? Subscribe!

ChatGPT Wants To Make Prompting Easier

For quite a while, every time you started a new chat, you needed to remind ChatGPT who you were (a writer), what you wanted (an email), and any specific requirements (100 words).

But nobody likes doing ice breakers.

Hence the recent launch of custom instructions: give context to ChatGPT so it remembers your preferences. Here are a few instructions we endorse using (courtesy of Nivi):

be highly organized.
be accurate and thorough.
cite sources whenever possible.
no need to disclose you’re an AI.
suggest solutions that I may have overlooked.

This way, you can cut to the chase. Adding to this, ChatGPT just announced two exciting features that should streamline prompting even more:

Example prompts: suggested prompts you can utilize at the start of a chat.
Suggested replies: recommended replies you can use in a conversation.

Both features could be time-savers if you’re stumped about what questions to ask. However, it’s worth noting that more complex conversations will still require some back-and-forth prompting.

Why it matters: We think prompt engineering (learning to converse with an AI) is overrated. Yup, we said it. We think the future of chat interfaces will be a combination of preloading context and then allowing AI to guide you to the information you seek.

FROM OUR PARTNERS

Imagine having a coworker who never sleeps.

Learning to work with AI is probably the highest leverage skill you can develop this decade.

But to operate as a superhuman, you first must grasp how AI works, what tools to use, and the right AI use cases.

This 10-day “Coworking with AI” course will teach you to master those three critical skills.

Ben Parr and Matt Schlicht (the brains behind Octane AI) will guide you on how to harness AI in your daily work with hands-on workshops, use cases, and prompting strategies.

Enroll in this flagship course on Maven before signup closes on August 22!

AI Can Help Detect Breast Cancer Earlier 🎗

Made using Midjourney

Taking a break from our usual tech-y talk to talk about … cancer.

Wow, that escalated quickly.

But we have good news: A study involving 80,000 women in Sweden found that AI detects 20% more cases of breast cancer than human radiologists.

So are radiologists out of a job? Not quite yet.

But AI could certainly supplement the work of our radiologists, saving them invaluable time. Even better, AI shows a knack for picking up on early stages of cancer that might go unnoticed even with 20/20 vision.

Why it matters: Though we don’t cover it much cause it isn’t our forte, AI systems are already playing a significant role in enhancing healthcare.. We recommend this newsletter to keep up with it all!

Around the Horn 🦄

A complete overview of the regulations US policymakers are considering.
There is growing demand for data centers that can process compute-heavy AI applications.
Elon Musk’s X.ai bought the domain AI.com from OpenAI.
Alibaba open-sourced two AI language models.