ChatGPT Goes Back to School

ChatGPT's new Study Together mode is changing how we learn, AI models are finally learning to work as teams with 30% better results, and Canva hopped on the Veo 3 train too…

In this edition we’ll be covering…

• ChatGPT's mysterious new educational feature that might fight cheating

• A tutorial on creating professional videos with Veo 3 in Canva

• How Sakana AI's multi-model teams are outperforming solo performers

• 6 trending AI signals

• 3 AI tools to boost your productivity

• And much more…

The Latest in AI

ChatGPT Goes Back to School?

ChatGPT is quietly rolling out a mysterious new feature called "Study Together," and it might be the smartest move OpenAI has made for education yet.

Instead of just giving students answers, this feature flips the script entirely, and it couldn't come at a better time.

The numbers tell a stark story: 43% of college students have used ChatGPT or similar AI tools, with 89% using it for homework and 53% for essays. The Guardian's investigation revealed thousands of UK university students caught misusing AI tools, while traditional plagiarism shows a marked decline.

Okay, but how does THIS actually help?

• Rather than providing direct answers, Study Together requires students to engage by responding to follow-up questions, essentially forcing them to think through problems instead of copying solutions.

• Some users speculate it might evolve into a group study mode where multiple students can join the same chat session, creating collaborative learning experiences that encourage discussion over plagiarism.

• The feature appears to be OpenAI's strategic response to what experts deem is “killing higher education.”

So What?

OpenAI is walking a fascinating tightrope here. They've created the most powerful homework-completion tool in history, and now they're trying to turn it into something that actually makes you learn.

It's like building the ultimate cheat sheet and then programming it to make you study instead of just copying the answers…

Whether students will embrace this more challenging approach remains to be seen, but it signals a greater shift toward AI that enhances learning rather than replacing it.

Get Your Hands Dirty!

Creating Cinematic Videos with Veo 3 Directly in Canva

Google's Veo 3 video generation model recently landed in Canva, and it's surprisingly easy to create professional-looking video content.

The Create a Video Clip feature can generate 8-second cinematic clips with sound from simple text prompts.

Here's how to access and use it:

1. Log into Canva and from the homepage, select Canva AI from the main menu.

2. Click on Create a video clip to open the Veo 3-powered interface (must have a Pro plan).

3. Describe exactly what you want to see. Be specific about setting, action, and mood.

4. The system will create an 8-second video clip complete with appropriate sound effects and background audio.

5. Once satisfied with the result, you can download the clip directly or integrate it into larger Canva projects for more complex video content.

Of course, some prompts to get you started:

A steaming cup of coffee on a wooden desk with morning sunlight streaming through a window, slow motion steam rising, cozy café atmosphere
------------------------------------------------------
Golden hour sunset over a calm ocean, waves gently lapping at the shore, seagulls flying in the distance, peaceful and serene
------------------------------------------------------
Busy city street at night with neon lights reflecting on wet pavement, people walking with umbrellas, rain drops on camera lens, cyberpunk aesthetic

Industry Intel

The AI Dream Team Shows Up

Japanese AI lab Sakana AI just proved that sometimes the whole really is greater than the sum of its parts.

Their new Multi-LLM AB-MCTS system lets multiple AI models work together on complex problems, and the results are pretty remarkable.

This approach creates what researchers call a "dream team" of AI agents. Each model brings its unique strengths to the table, whether that's coding expertise, creative writing skills, or logical reasoning abilities.

Here's how it works in practice:

• The system uses Adaptive Branching Monte Carlo Tree Search to intelligently decide whether to refine existing solutions or generate completely new approaches at each step.

• Different models like OpenAI's o4-mini, Google's Gemini 2.5 Pro, and DeepSeek-R1 are dynamically assigned tasks based on which one performs best for specific problem types.

• In testing on the challenging ARC-AGI-2 benchmark, the multi-model approach solved over 30% of problems, which significantly outperforms any individual model working alone.

So What?

This just might fundamentally change how we think about AI deployment. Instead of being locked into one provider's ecosystem, businesses could soon mix and match the best models for different parts of complex workflows.

The real breakthrough here is that the system can solve problems that were impossible for any single model. In one case, o4-mini generated a flawed solution, but DeepSeek-R1 and Gemini identified the error and corrected it.

That's the kind of collaborative intelligence that could transform how AI tackles real-world challenges.

Sakana has open-sourced the framework as TreeQuest, so expect to see this approach spread quickly across the AI landscape…

Quick Bites

Stay updated with our favorite highlights, dive in for a full flavor of the coverage!

Google rolled out Batch Mode for the Gemini API, allowing developers to process large AI workloads asynchronously at 50% of standard pricing with 24-hour turnaround times.

A massive study analyzing over 15 million biomedical papers found that at least 13.5% of 2024 publications show signs of AI assistance.

UC Davis launched the first AI-powered self-monitoring satellite that uses onboard digital twins to predict system health in real-time, completing the entire project from design to launch in just 13 months.

A couple is pregnant with their first child after 18 years of trying, thanks to Columbia University's STAR system that uses AI to find hidden sperm in azoospermic men.

Capgemini announced its acquisition of WNS for $76.50 per share, creating a global leader in "Agentic AI-powered Intelligent Operations" to transform business process automation.

Trending Tools

🔍 Devi AI - Helps you find organic leads by monitoring Facebook private and public groups, LinkedIn, Twitter, Bluesky, Reddit, and Telegram posts. Save 3 hours a day and grow your business. Get started here.

⚡ Unsloth AI - Open-source fine-tuning platform that's 30x faster than traditional methods with 90% less memory usage.

💬 IllumiChat - Transform your customer support ticket backlog into instant AI-powered responses. Designed to crush wait times, improve NPS scores, and reduce churn for growing customer service teams.

The Neural Network

Gemini product lead Logan says the next 6 months of AI are likely to be the most wild we will have seen so far.

Coming from someone who's been at the center of AI development, that's either really exciting or mildly terrifying. Maybe both.

Meanwhile, our last poll revealed something we probably all suspected but didn't want to admit: most of you think the kitchen needs the most AI design help.

Apparently, we've all collectively decided that if AI is going to redesign our lives, it might as well start with the room where we pretend to be adults who cook real food…

Until we Type Again…

Thank you for reading yet another edition of Digestible AI. Be sure to give us a follow on X, Instagram, and LinkedIn too!

If you have any suggestions or specific feedback, simply reply to this email. Additionally, if you found this insightful, don't hesitate to engage with us on our socials and forward this over to your friends!