Did Code Interpreter Just Replace Your Data Team?

Everything you need to know about OpenAI's latest release; and another way to create your own custom AI chatbots

Author

The AI Exchange
July 11, 2023

Welcome to another edition of the best damn newsletter in AI. Here we’ll break down AI topics that matter, open your mind to use cases, and keep you ahead of the curve.

Our #1 goal is to be useful. So please shoot us an email 📩 if you have questions or feedback, and especially if you implement something we share!

Here's what we're covering today:

  • Everything you need to know about Code Interpreter

  • Another way to create your own custom AI chatbots, without coding!

  • Other big OpenAI announcements and some drama in the AI world

... and if someone forwarded this email to you, thank them 😉, and subscribe here!

Let’s get to it! 👇

TODAY'S PERSPECTIVE

Did Code Interpreter just replace your data team?

Everything you need to know about OpenAI’s latest beta release.

OpenAI rolled out access to their Code Interpreter model to all ChatGPT Plus members.

While it sounds technical and jargony - it’s actually really freakin’ cool and in this newsletter we’re going to break down how it works + how it can help you in your work.

What is Code Interpreter?

Just like ChatGPT is based on an AI model specifically designed to be good at chat. Code Interpreter is based on an AI model specifically designed to be good at solving problems by writing and running simple code.

Right now, it’s mostly good at coding tasks that data teams do - like data analysis and creating graphs. Which is why you might be seeing headlines and social media filled with the “did Code Interpreter just replace your data team”.

So, how do you get access?

  1. You need a ChatGPT Plus acount

  2. Go to Your Name in the bottom left corner > Settings > Beta Features

  3. Toggle on Code Interpreter

How do I use it?

  1. Create a new chat thread

  2. Select GPT-4, then Code Interpreter in the dropdown

What can you ask it to do?

You can upload PDFs, text files, google docs, excel spreadsheets and a lot more directly into Code Interpreter.

And then ask it to do things for you, with that information.

🤯 yeah.

For example:

  1. Upload a CSV of customer orders and ask it to graph monthly revenue over time

  2. Upload a PDF and ask it to make the PDF searchable using OCR (a way for computers to read documents)

  3. You can have it convert files between pretty much any different format

  4. Create visual flowcharts for any business process based on a written explanation

  5. Have it create a word cloud of the most frequent words used in an article, or customer reviews

  6. … and a lot more

Is this really replacing data teams?

Ahem, no. But it is incredibly powerful and should be looked at seriously by anyone who uses excel, or often asks someone to use excel for them.

We think Code Interpreter is a perfect glimpse of how easy certain tasks are going to become as AI gets integrated deeper into our everyday lives.

Many tasks are about to become as easy as simply describing what you want done, and it’s done.

But before you dig in - a word about data privacy

DO NOT upload anything sensitive to Code Interpreter. Again, we like the Reddit Rule - if you wouldn’t post it anonymously on Reddit, don’t put it in ChatGPT.

You have to keep your thread history & data sharing on in order to use Code Interpreter at the moment. So please be mindful. And train yo teams.

FROM OUR PARTNERS

Build a custom chatbot over your docs - with absolutely zero code

Context chatbots, or “ChatGPT with my docs”, have gained a ton of traction as a way to bring AI efficiently and effectively into your internal ops.

And with a no-code tool like Bubble, you can get a custom one up and running + take it further with custom features the more that you learn how to use Bubble.

Here’s how you can do it!

  1. Sign up for Bubble account and create your first Bubble app

  2. Create a simple chat interface using Bubble’s editor (or use a template like this one)

  3. Connect your chat interface to one of the many ChatGPT plugins

  4. Connect to a vector database, like Pinecone, using Bubble’s API Connector

  5. Add your docs to your vector database

  6. Configure a workflow that takes a user’s question, queries your vector database, and sends the question + returned docs to ChatGPT

  7. Return the response to the user in your chat interface

  8. Deploy your custom context chatbot!

Premium Members: access the Bubble workshop recording here!

MEMBER RESOURCES 🔥 

🔒️ Premium Only: What’s Dropping This Week

  1. Learn best practices for Building Prompt Libraries

  2. Understand the risks of Prompt Drift and how to guard against it

  3. Read our full playbook chapter on Code Interpreter here

  4. Check out the new tutorial on how to use Code Interpreter to analyze public financial reports to make better business decisions 🔥

Have an AI question? Chat with AIxChat, your friendly AI consultant trained on all the knowledge in The AI Exchange 🤖 

LINKS

For your reading list 📚

A few official updates from OpenAI…

Meanwhile controversy strikes in the content world…

And if you're really nerdy...  

  • If you remember the Google “we have no moat” memo. Turns out, it was real. But the head of Google’s AI division, DeepMind, disagrees

That's all!

We'll see you again on Thursday. Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected].

... and if someone forwarded this email to you, thank them 😉, and subscribe here!

Cheers,

🪄 The AI Exchange Team