ChatGPT just got easier to use

Good morning. Are you hungry? Because we’ve prepared a giant feast of AI news and insights just for you! 🍽️

🤯 MYSTERY AI LINK 🤯

(The mystery link can lead to ANYTHING AI-related: tools, memes, articles, videos, and more…)

🔨 AI TOOLS OF THE DAY

🎁 Cool Gift Ideas: Use AI to find gifts for everyone on your list. → Check it out

💬 CaptionGen: Generate the perfect image captions for social media. → Check it out

🏠 RoomGPT: Your personal AI interior designer. → Check it out

OPENAI RELEASES CHATGPT “PROJECTS” FEATURE 🗂️

OpenAI continues to roll out fancy AI Christmas gifts. 😁

What’s new? OpenAI is rolling out a new feature for ChatGPT called “Projects,” designed to help users stay organized.

How does it work? Projects is a folder-based system that allows users to group related files, chats, and instructions together on ChatGPT, making it easier to manage multiple tasks or ideas within the chatbot. As shown in the demo (above), Projects appear in a sidebar where you can create, name, and customize folders with colorful icons. You can also attach files, set specific instructions for how ChatGPT responds within each project, and even include past conversations for quick reference. Projects are now available for Plus, Pro, and Teams users, with free and educational users set to receive access in early 2025.

Why is this significant? ChatGPT can help with projects of all kinds, from designing workflows to helping design budgets. Until now, however, it has been difficult to organize conversations and content pertaining to different projects. This new feature will make that a lot easier, and it will allow people to more easily use ChatGPT for both short-term and long-term tasks.

MICROSOFT UNVEILS PHI-4 FOR RESEARCH TASKS 🤓

Did you hear about the new tech sorority? It’s called Micro Phi. 👨‍💻

What’s up? Microsoft has unveiled Phi-4, the latest addition to its Phi series of generative AI models, aimed at efficiency for research tasks.

What does it do? Designed as a small, more efficient language model with 14 billion parameters, Phi-4 focuses on enhanced problem-solving, especially in mathematics. In tests, Phi-4 outperformed much larger models, including Gemini Pro 1.5, on math competition problems. This improvement is credited to higher-quality training data, including synthetic datasets and refined post-training techniques. Currently, Phi-4 is available for research purposes only through Azure AI Foundry, Microsoft’s development platform. Access is restricted under a Microsoft research license agreement.

Why is this important? The new Phi-4 model enters a competitive field of smaller AI models, such as GPT-4o Mini, Gemini 2.0 Flash, and Claude 3.5 Haiku, which prioritize speed and cost efficiency. As advanced AI models get more expensive to run and traditional training data reaches its limits, researchers are exploring these more focused, cost effective models that can play a part in training themselves.

HARVARD TO RELEASE 1 MILLION BOOKS FOR AI TRAINING 📚

Q: What’s a writer’s favorite snack?

A: Synonym rolls. 📕

What’s happening? Harvard University, in collaboration with Google, plans to release a groundbreaking dataset of 1 million public-domain books. This collection will feature works from iconic authors like Dickens, Dante, and Shakespeare—texts no longer protected by copyright.

Why is this significant? The release of this dataset is part of the Institutional Data Initiative, which focuses on collaborating with libraries and museums to publish their collections as data. This dataset will make high-quality training material for AI widely accessible. This is important for developers of all sorts. As large tech companies spend hundreds of millions of dollars on AI training data, the little guys are falling behind. While the release date and specifics are still unclear, this initiative could benefit researchers, startups, and organizations who are lacking the resources of tech giants like Google and Microsoft.

TWITTER (X) TUESDAY 🐦

HAS AI REACHED SINGULARITY? CHECK OUT THE FRY METER BELOW: