- Everyday AI
- Posts
- ChatGPT’s Canvas Mode Overview: How to use it and 5 things you need to know
ChatGPT’s Canvas Mode Overview: How to use it and 5 things you need to know
Grok rolling out new features to try and compete with ChatGPT, OpenAI gets its president back, Sam Altman making big AI pitch in D.C. and more.
Everyday AI
November 13, 2024
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: ChatGPT’s Canvas mode is kinda being overlooked. Wondering exactly what it is? We break down the 5 things you need to know. Give it a listen.
🕵️♂️ Fresh Finds: Anthropic releases AI jailbreaking alternative, OpenAI president returns, Grok rolling out new features to (try and) compete with ChatGPT, Microsoft releases industry-specific models and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: ChatGPT testing a new feature in Canvas that could steal Anthropic Claude’s thunder, Google Gemini could be updating its Gemini Live to give it file access, AMD lays off 1,000, OpenAI set to release new AI strategy in D.C. and more – read on for Byte Sized News.
🚀 AI In 5: Thin on time? We’re giving today’s full episode the AI in 5 treatment and breaking down the 5 things you need to know about ChatGPT’s Canvas. Actually in 5 minutes. See it here
🧠 Learn & Leveraging AI: People are sleeping on ChatGPT’s new Canvas mode. We break down everything you need to know. Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We talked about Amazon’s latest AI chip, the Catholic Church turning to Digital Twins, a $100 billion AI initiative and more. Check it here!
ChatGPT’s Canvas Mode Overview: 5 things you need to know 🖐
Why is no everyone sleeping on this ONE feature of Canvas mode in ChatGPT?
OpenAI announced Canvas a few weeks ago, a new ChatGPT mode and way to code and write.
Everyone's trying to compare this to Anthropic's popular Artifacts feature inside of its Claude chatbot.
But almost everyone's missing the point by simply comparing it to Artifacts.
We'll break it down and tell you 5 things you need to know about OpenAI's new Canvas mode.
Join the conversation and ask Jordan questions on ChatGPT’s Canvas here.
Also on the pod today:
• Live demo of Canvas mode 🧑💻️
• Technical Details of Canvas Mode 🤖
• Canvas Mode Feature Comparison 🥊
It’ll be worth your 56 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – Vocera helps companies build with AI voice agents, PearAI is an open-source AI code editor and Draft is an AI-powered to-do list.
Trending in AI – X (Which we call Twitter like most people) may start providing free access to its AI chatbot, Grok, in attempt to compete with ChatGPT. (And it may get the ability to search the web.)
AI Jailbreaking - Anthropic released a paper on an alternative to jailbreaking defense.
New research: Jailbreak Rapid Response.
Ensuring perfect jailbreak robustness is hard. We propose an alternative: adaptive techniques that rapidly block new classes of jailbreak as they’re detected.
Read our paper with @MATSprogram: arxiv.org/abs/2411.07494
— Anthropic (@AnthropicAI)
5:32 PM • Nov 13, 2024
OpenAI News — Greg Brockman has returned to OpenAI as president after a three-month leave, amid other leadership changes at the company.
AI and Privacy — Want to opt out of Google’s new AI features? Not really possible, which is concerning to many.
AI in the home — Apple is set to release an AI-powered tablet that mounts on your wall and can control your home. Cool?
AI Models — Microsoft just made a big splash with its new adapted AI models for certain industries. In short, these models are tailored for specific industries, make AI more useful and precise, boosting business results.
1. Anthropic's CEO reveals why he left OpenAI and started Claude 🗣️
In a candid chat with Lex Fridman, Anthropic CEO Dario Amodei revealed his distaste for the "AI wars," emphasizing a vision of responsible AI development over market dominance.
Amodei left OpenAI not over a Microsoft deal but to create a culture where AI safety leads, hoping rivals will mimic successful, ethical business practices. He also shared the company's ongoing struggle with naming conventions, as updating models like Claude 3.5 Sonnett doesn't always align with easy-to-understand labels.
According to Amodei, the focus should remain on enhancing the AI ecosystem, which could inspire businesses aiming for growth to prioritize ethical operations over competition.
2. ChatGPT testing a Claude Artifacts killer with new Canvas feature?
OpenAI is apparently testing new features for its ChatGPT iOS app, introducing editing buttons for its canvas feature, previously exclusive to the web version. While still ironing out some issues and missing elements like the model selector, this move suggests a push to match and potentially surpass Anthropic's standout offerings.
On the web front, OpenAI is experimenting with executing code within canvases, initially supporting JavaScript, TypeScript, and Python. Though these features are in testing, OpenAI's rapid innovation could soon offer robust tools for users aiming to elevate their AI interactions, potentially reshaping the landscape for tech enthusiasts and businesses alike.
What’s this mean in short? Right now, the ability to execute code in-body is one of Anthropic Claude’s biggest differentiators, and it looks like OpenAI is looking to erase that.
3. Report: Big name LLM makers struggling to build advanced AI 🤦♂️
According to Bloomberg, leading AI companies, including OpenAI, Google, and Anthropic, are encountering setbacks with their latest models.
OpenAI's Orion, Google's upcoming Gemini iteration, and Anthropic's Claude 3.5 Opus are all struggling to meet high internal expectations. These challenges stem from the difficulty in sourcing fresh, high-quality training data and the immense costs involved in developing next-gen systems.
4. Google Gemini Live could soon let users chat with uploaded files 🗄️
Google's Gemini AI is set for a big leap forward, potentially mirroring OpenAI's ChatGPT by bringing conversational prowess to your files, photos, and recordings.
According to Android Authority, a beta version of the Google app hints at a "Gemini Live" feature that could chat about user-uploaded attachments. Though the feature remains inactive for now, Google's stealthy release of a Gemini app for iPhone users in the Philippines suggests their AI gears are turning fast.
5. In attempt to compete with NVIDIA, AMD lays off 1,000 employees 👔️
In a strategic shake-up, AMD announced it will lay off 4% of its global workforce (about 1,000 employees) to sharpen its focus on the burgeoning AI chip market, according to Reuters.
This move comes as AMD seeks to challenge Nvidia’s dominance, which boasts over 80% market share in AI chip sales. Despite the layoffs, AMD remains optimistic about capturing a slice of the $500 billion AI chip market projected by 2028, especially as its MI300X accelerators gain traction with tech giants like Meta and Microsoft.
6. OpenAI Unveils Bold AI Blueprint with a Nod to Nuclear Power 🏭
In a strategic move, OpenAI is set to present its “blueprint for U.S. AI infrastructure” in Washington, D.C., highlighting plans for AI economic zones and a North American AI alliance to rival China, as reported by CNBC.
The blueprint envisions leveraging the U.S. Navy’s nuclear expertise and boosting energy infrastructure through a proposed “National Transmission Highway Act.” OpenAI aims to collaborate with the incoming Trump administration on AI policy, emphasizing investment that could generate thousands of jobs and transform regions like the Midwest and Southwest into AI hubs.
ChatGPT Canvas: 5 things to know in 5 minutes
ChatGPT’s new Canvas mode is out!
We’re going to tell you how it compares to Claude Artifacts, the bigger AI picture that Canvas plays a role in and 5 things you need to know about this new ChatGPT mode.
We keep it actually short this time shorties.
Check out today's AI in 5.
🦾How You Can Leverage:
What’s this random weird buzzword we keep hearing about — augmented intelligence?
Simple.
For the first few years of us humans directing Large Language Models, we’ve essentially just had a back-and-forth shouting match and hope something good happens.
Like… DO MY WORK ChatGPT!
FIX THIS OUTLINE GEMINI!
It’s kinda like we’ve been in separate cars, driving on our own one-way streets, trying to make this AI thang work.
The results?
Sometimes great.
Sometimes, it’s a lot of back-and-forth prompt engineering and frustration on WHY DO YOU KEEP WRITING ABOUT my cat’s birthday cake when I asked for a business plan now I gotta start over what the frick AI?
(Happy late birthday, Rocky.)
But what if LLMs worked in a more intuitive way?
What if you and AI could have two-way streets and be in the same car?
Information freely flowing back-and-forth in a seamless way, cutting down on the detours of disappointing dialogue and fast-tracking to the end goal you wanted without all the keyboard smashing you don’t.
That’s augmented intelligence, in a nutshell — human and AI working hand-in-hand trying to make each other better.
And for the first time — we may finally have our first glimpse of that in a consumer AI product with ChatGPT’s new Canvas mode.
So today, we gave it the deep dive treatment on Everyday AI.
We’re not just showing off a new feature.
We’re previewing what the future of working with AI might look like, and dispelling some common myths along the way.
Ready to get augmented y’all?
Here’s the 5 things you NEED to know about ChatGPT’s new Canvas feature:
1 – Available to all ChatGPT Plus users 💰
In this case?
OpenAI shipped. 🛳️
Immediately after announcing Canvas, OpenAI rolled out availability to all paid users.
OpenAI plans to make it available to free users after the beta, but don't hold your breath for freebies.
If you want Canvas, you’ll probably just wanna fork over that $20 and immediately get that return back the first time you hit enter.
Try this:
If you're a Plus user, dive in now.
Start with a project you're already working on – maybe that blog post draft or coding challenge. See how Canvas streamlines your workflow compared to the standard chat.
If you're on the fence about upgrading, this might be the feature that tips the scales.
2 – Canvas is its own mode 💡
Start here.
TBH, we’re not sure why you’d ever choose the normal GPT-4o mode anymore.
You can't toggle Canvas on and off mid-conversation. It's a separate mode you select when starting a new chat.
Also, keep in mind that sometimes you even need to explicitly tell ChatGPT to "use Canvas for this chat" to activate the Canvas feature, even if you’re in the Canvas mode.
Try this:
Next time you start a chat, select "GPT-4 with Canvas" from the model picker.
If the full Canvas interface doesn't pop up, add "use Canvas for this chat" to your first prompt. Get in the habit of starting here – you might be surprised how often you end up using the extra features.
3 – Works with all other tools 🛠️
Plays nicely with others.
Good ChatGPT.
Unlike the o1 model and Advanced Voice Mode, Canvas mode can play nicely with most of OpenAI’s other toys and tools like Browse with Bing, Advanced Data Analysis, and DALL-E right within the Canvas interface.
Try this:
Create a research project in Canvas.
Use Browse with Bing to gather recent info, then use the inline editor to organize your findings.
Throw in some data analysis or DALL-E generated images to round it out. It's like building with AI Legos – all the pieces fit together in one place.
You’re the Augmentor now.
4 – Not a carbon copy of Anthropic Claude's Artifacts 🤔
Let’s cut the click-bait nonsense, shall we?
Stop comparing OpenAI’s Canvas to Anthropic Claude’s Artifacts.
Canvas is a collaborative in-line partner, and Artifacts creates and renders code.
While both have split-screen interfaces, Canvas and Claude's Artifacts serve different purposes.
Artifacts can render code and create interactive elements.
Canvas is focused on collaborative editing and writing.
Oh, and to all of y’all saying Canvas is just an Artifacts clone?
Receipt: OpenAI actually had the split-screen idea first with their GPT builder back in November. So…. There’s that.
Try this:
If you have access to both, compare them on a coding task.
Use Claude's Artifacts to write and render a simple game.
Then use Canvas to refine and debug the code.
See how the workflow differs and which feels more intuitive for your style. And both can be amazing for their own purposes, and even more powerful when used in tandem.
5 – In-line editor puts ChatGPT in a league of its own 💪
Sure this is a small step.
But, Canva is a BIG small step in transitioning from Artificial Intelligence to Augmented Intelligence.
You’re not just traveling a sometimes frustrating road of constantly re-prompting and copy and pasting a LLM that seems to be stuck in its ways.
The inline editor is Canvas's secret weapon. You can make edits directly in the generated text, then ask ChatGPT to work with your changes.
So if Canvas gets only 8 out of 10 things right, you don’t gotta start from scratch. You can just place your cursor on those two problem areas and fix them yourself in real-time, or ask ChatGPT to focus its attention on those two areas and it won’t touch the stuff that’s already sparkling.
It's a big step towards true human-AI collaboration.
Try this:
Start with a rough draft of something – could be an email, report, or even code. Use the inline editor to make targeted changes, then ask ChatGPT to expand on your edits.
Play with the "Polish," "Adjust reading level," and "Suggest edits" features. It's like having an AI writing partner that adapts to your style in real-time.
6 – Bonus! ✅
Remember, Canvas is in Beta. It’s brand new, so there will be bugs. As an example, the Canvas mode we built live this morning had errors when going back in later.
So, make sure to save (copy/paste) anything you create inside of Canvas when you’re done before you leave the window, just in case.
What do you all think?
Do you want to see more in-depth tutorials on Canvas?
Which Canvas features do you like most?Vote to see live results |
Reply