It’s really starting to feel like an AI news summer. Updates, announcements, and even leaks are everywhere. More and more companies are adopting AI within their daily work. Mythic is no exception and our anticipation for OpenAI’s new ChatGPT-4 features has been, in a word, extreme.
Table of Contents
In today’s post, I’ll discuss some new ChatGPT-4 bells and whistles that:
- are currently available,
- may be around in the near future,
- I’d really love to use (are you listening, OpenAI?)
New Features
Web Browsing / Browse with Bing
Internet users are (mostly) human beings, and if left unchecked they will use new powers in creative ways. That’s exactly what happened last week, when OpenAI disabled its new web browsing feature to protect paywalled content of major media outlets. As usual, people figured out a work-around and were happily reading their WSJ without a subscription.
At least while we wait with bated breath for OpenAI to fix this, there’s plenty more to discuss.
Code Interpreter
Now the intelligence may be artificial, but my excitement here is definitely natural.
On July 6th Open AI announced this feature’s rollout in Beta over the coming weeks. Use it to handle a wide range of tasks. From data analysis to creating diagrams, using Python libraries and doing complex math - it streamlines operations and will boost working efficiency in ways that I don’t think we can fully comprehend yet.
Oh, and you can finally upload files too.
Take data visualization, for example. By uploading a CSV file containing marketing data (i.e. website traffic or campaign performance metrics) you can ask ChatGPT to interpret the data and generate insightful visualizations.
While AI can’t be relied on 100% over a good human here, it can simplify the process of data analysis and reporting, enabling us to quickly identify trends and make data-driven decisions.
Content creation and manipulation is also changing. The Code Interpreter can transform images, create GIFs, and even generate videos from simple prompts. This can be particularly useful for social media marketing, where visual content is crucial.
For instance, one can use this feature to quickly create variations of an image for A/B testing or generate engaging GIFs for social media posts.
Pretty neat, and that’s just a sampling of what Code Interpreter can do. Let’s look into one a bit deeper.
File Uploads
With Code Interpreter toggled on, you can upload various types of files such as images, texts, audios, videos, PDFs, spreadsheets, slideshows, etc.
Try uploading a photo of a landscape and ask ChatGPT to generate a poem inspired by it, or take a voice recording of a meeting and have ChatGPT transcribe it. You can also upload multiple files at once and instruct it to combine them or compare them.
This feature enables us all to carry out more intricate and imaginative tasks by combining or comparing multiple files, vastly expanding the range of possibilities for utilizing ChatGPT.
Leaked Features
Recently, some leaked screenshots displaying much-anticipated features have been spreading around the subreddits. We can’t truly know their veracity, but the leaker (who is supposedly an Open AI employee) has been accurate in the past.
If they are right, here’s a little of what we can expect.
Workspaces
Why would one want to create personalized virtual environments using ChatGPT?
Well, you could establish a workspace for mastering a musical instrument, where the AI would act as a music teacher. Another possible use includes a workspace for cooking, where the bot could be like your culinary guide. You can make multiple workspaces and switch between them easily. Invite others to join your workspaces and collaborate with both ChatGPT and other users.
Profiles
Finally we'll be able to craft personal profiles with preferences, interests, personality traits, etc. If a user tells ChatGPT that they are a fan of mystery novels, for example, it might recommend some books of that genre that they haven't read yet. Tell ChatGPT that you are an extrovert, and it might suggest social activities or events that you would enjoy.
This feature helps to make the interaction between users and ChatGPT more human-like, natural, engaging, and interesting by personalizing its responses based on the profile created.
Or, maybe we should be careful what we wish for ?
Features Wishlist
I changed my mind. Let's keep wishing.
Folders
As someone who uses the tool a lot, my chat history has become a bit messy. I can’t be the only one, either. How useful it would be to have the ability to create folders and subfolders to organize everything?
I know, there are some browser extensions (like this one for Chrome) to accomplish this but they can be clunky. A native feature within the tool itself would be preferable. Again, who’s with me here?
Prompt Suggestions
We’re all trying to improve our prompts. There are courses you can take, plugins to use and good old fashioned research to do, but with limitations on plugin usage (currently 3 at once) and time, I’d be thrilled to see this as a standard feature for both free and paid versions.
Imagine if ChatGPT tells would suggest new ways of asking questions or giving instructions. Something like, “here’s a list of ways you could input your prompt to get a different answer,” would be super helpful.
I could go on and on, but something tells me that Sam Altman and the OpenAI team just might not be reading this.