• /IMAGINE
  • Posts
  • Supercharge Your ChatGPT with Mentions

Supercharge Your ChatGPT with Mentions

And → AI Audio Deepfakes, Mixtral Model Developments, and Google Crushing It

Welcome back, my AI-obsessed friends.

Good morning, and Happy Monday! Hope you all had an excellent weekend. This is the latest edition of /Imagine, your go-to source for the freshest news and developments in AI. Let’s dive in…

High Level: Supercharge ChatGPT Workflow With New Mention Feature


The Bytes:

✣ Custom ChatGPT: We talked about this several newsletters ago, but OpenAI has introduced a new beta feature that allows users to create custom versions of ChatGPT, known as GPTs.

✣ @ Mention Functionality: The new @ mention feature enables users to integrate various custom GPTs into their ongoing ChatGPT conversations, similar to how mentions work in platforms like Slack and Discord.

✣ Enhanced User Experience: This update enhances the user experience by allowing a more personalized and universal assistant, combining the strengths of different GPTs. Here is a great example of a user deploying an app from scratch using Grimoire GPT and mentioning DesignerGPT.

Why You Should Care: The @ mention feature represents a significant leap in the usability and versatility of ChatGPT, paving the way for more complex AI interactions and integrated uses.

High Level: Google Bard’s Leap with Gemini Pro and the Anticipation of Gemini Ultra

Image Credit: Google AI

The Bytes:

✣ Gemini Pro’s Leap: Google has significantly upgraded Bard with the Gemini Pro model, which now ranks higher than the regular OpenAI Chat GPT-4 in various benchmarks.

✣ Awaiting Gemini Ultra: While Gemini Pro has made strides, the AI community is eagerly awaiting the release of Gemini Ultra. Touted as Google’s best model yet, Gemini Ultra is expected to outperform its predecessors and set new standards in AI performance.

Why You Should Care: This is only their second-best offering, and it’s crushing expectations. The rapid improvements in AI models like Google Bard’s Gemini Pro signal the ramping up of an intense model race. I expect OpenAI and others to ship updates soon to stay ahead.


High Level: The Rising Challenge of AI Audio Deepfakes


The Bytes:

✣ Outpacing Detection: A recent article from Scientific American discusses the growing challenge of AI-generated audio deepfakes, which are multiplying and becoming increasingly difficult to detect.

✣ Case Study: The article references a case where a high school principal was allegedly recorded making racist comments. The authenticity of the recording was called into question and claimed to be a deepfake.

✣ Public Figures and Beyond: While deepfakes target public figures, this incident shows that anyone, even you can be a target of AI-generated impersonation.

✣ Robocalls and Elections: The article also details robocalls in New Hampshire that used a deepfake of President Joe Biden’s voice to discourage voter participation.

Why You Should Care: The proliferation of AI audio deepfakes poses significant risks to individuals and society, highlighting the urgent need for effective detection and authentication methods to maintain trust in digital media. Keep a close eye on this as it develops. With it being an election year, this is going to get annoying quickly.

High Level: Mixtral’s 8x7 Model - The Open Source LLM Surpassing ChatGPT-3.5

Image Credit: Mistral AI

The Bytes:

✣ Mixtral-8x7’s Superiority: The Mixtral-8x7, an open-source local Large Language Model (LLM) employing a Mixture of Experts (MoE) approach, has been confirmed to surpass ChatGPT-3.5 in performance across various tasks.

✣ Speed and Efficiency: Not only does Mixtral-8x7 outperform in terms of task handling, but it also operates faster, making it a highly efficient AI model for businesses to utilize.

✣ Business Integration: Companies can leverage Mixtral-8x7 to build their own AI-driven solutions, thanks to its open-source nature and the ability to deploy locally, offering a powerful AI directly on devices.

✣ Multilingual and Multitasking: With a context length of 32k tokens and support for multiple languages including English, Spanish, French, Italian, and German, Mixtral-8x7 is versatile and capable of code generation among other tasks.

Why You Should Care: If you’re a frequent reader, you’ll know I am an advocate for running your AI locally. It’s been out for about a month but has gone under the radar. The emergence of Mixtral-8x7 as a leading open-source LLM opens up new possibilities for businesses, creators, and individuals to integrate advanced AI capabilities into their operations without relying on external cloud services.

Read More


/Imagination of The Day: Let’s end the day on a creative note. Today, I’m showing you how to get this cool 3D character result that you can try for yourself. It is a mixture of prompts I’ve found online with elements of my testing, and it yields great results.

Prompt Share: 3D cartoon of an alien rapper wearing a Yeezy sweatsuit, diamond chains around his neck, full body, 8k, unreal engine, surreal quality --chaos 40 --ar 3:4 --style raw --stylize 750 --v 6

Stay Curious: Don't forget to follow me on Twitter at @jake_joseph and the brand on its fresh new page @slashimagineai for the latest AI news, my two cents, and discussions. Share your thoughts and join the conversation.

Share The Insights: Know someone else who's obsessed with AI? Forward this newsletter and let's expand the community.

Feedback: Your perspectives help shape the content, and hopefully helps me make the newsletter suck less every day. Share your feedback on this issue or what you'd like to see next by directly replying to this email or by reacting to the poll below.

How did we do with today's newsletter?

Share your vote to help us deliver the #1 resource on AI for our community.

Login or Subscribe to participate in polls.