K-12 Coding Platform similar to MIT Scratch, with enriched blocks for AI/3D game/2D physics etc. Also features an AI assistant for personalized learning support
Vibe coding is good enough for personal prototypes, but to build and deploy anything in production you need a few core engineering skills:
- Decomposition: break big problems into smaller pieces
- Iteration: start simple, improve step by step
- Design for change: write code that is easy to adapt when requirements change (they always do!)
- Simplicity first: avoid over-complicating solutions (AI often makes this mistake)
- Testing: check your code against both normal and edge cases
- Debugging: track down why actual behavior differs from what you expected
- Optimization: make programs fast and efficient
- Version control: manage changes and collaborate safely
- Error handling: recover gracefully instead of crashing
- Code reuse: build on existing solutions instead of reinventing
- User experience design: make apps feel natural
- Client–server architecture: understand how to split work between the frontend and the backend
AI coding agents can help with many of these, but you still need to know what to ask for and when to reject their default suggestions.
Education is completely broken in the age of AI.
Schools still insist that students shouldn't use LLMs for homework, which makes zero sense.
It's like asking humans not to use fire after it was discovered.
The best course of action is to embrace AI and teach them how to use it, but instead, all they are doing currently is encouraging them to cheat and avoid detection.
Unfortunately, this means our next generation are either going to be clueless because they never applied AI or cheaters because they figured out how to game the system.
#ScratchWeek is here! This annual celebration invites Scratchers worldwide to create projects around themed prompts all week long. Join us in the Scratch Olympics studio to kick start the fun! 🤸♀️⛹️♀️🏅 https://t.co/oXM76c5IOG
Did you know that you can create an AI chat with only 10 #Scratch blocks? Here is how to do it using the #ChatGPT and chat window widget blocks at https://t.co/wEg7wfBHjC:
https://t.co/WvPnquNH8X
Great initiative, @Scratch! This workshop on Scratch
and AI sounds fantastic. Important for educators to stay ahead. For those interested in exploring further, https://t.co/HexbzgkRNe offers Scratch-like blocks for AI tools like ChatGPT, speech recognition & more! #AI #ScratchEd #FutureofLearning
A new video tutorial was posted on how to make a sprite talk to us. It demonstrates how to use the #ChatGPT, text-to-speech, and speech-to-text #AI#Scratch blocks on the https://t.co/ZnWdVSSl10 platform:
https://t.co/lOYowbgerX
Curious how Scratch is approaching AI technologies? Join our free Exploring Scratch and AI: Pitfalls and Possibilities workshop to get hands-on with experimental Scratch blocks and explore this emerging space. Spaces are limited! https://t.co/qviPXMCRDS
@RickLamers Suggestion: now that inference latency is even shorter than RTT, maybe it makes sense to offer "agent co-location" service? Users can run their agents on your cloud that also runs your LLMs, so the agents can make multiple "local" LLM calls before sending the final response back.
@OfficialLoganK@OpenAI At https://t.co/wEg7wfBHjC, we have extended MIT Scratch to allow students to build AI-powered projects using ChatGPT and DALL-E blocks. Also, we are using ChatGPT for CreatiCode XO - the world's first K-12 coding assistant that understands Scratch blocks. Both features are free.