AI just turned physics class into a playable game with… · M&A Beginners 🎓
![]() Models & Agents for BeginnersAI explained simply — for beginners and teens.
|
🎧 If you only have 10 minutes this week Episode 32 · AI just turned physics class into a playable game with snap, crackle, and pop. 2026-05-08 ▶ Listen now |
| > **AI just turned physics class into a playable game with snap, crackle, and pop.**
> **A fun new demo lets you play with the hidden rules of motion that most people never notice. We’ll also look at why real AI helpers still trip up in surprising ways and check out two big companies testing AI for games and dating. By the end you’ll have something you can open right now on your laptop and experiment with.**
### The Big Story Someone posted on Reddit about the messy ways AI agents actually break when people try to use them for real work. These aren’t problems with the AI being “dumb.” They’re problems with how the AI keeps track of what it’s doing across many steps. Think of it like trying to have a long group chat about planning a school project while also texting your friend about weekend plans at the same time. Bits of one conversation leak into the other, and suddenly the AI starts mixing up details. It might invent a fact that sounds believable, like referencing a website that doesn’t exist, because it’s just trying to keep the conversation going smoothly. This matters because more people are starting to use AI helpers for homework, creative projects, or even small side jobs. When the helper gets confidently wrong, it can waste time or create awkward situations with teachers or clients. The post points out that the fix often isn’t a smarter model—it’s adding better ways for humans to check the work before it goes out. For you specifically, this means the AI tools you might use for essays or game ideas still need a quick human eye. It’s not a reason to avoid them, but it’s a reason to treat their answers as helpful drafts rather than final truth. Right now you can’t try the exact broken workflows from the post, but you can test how AI handles memory yourself. Open any chatbot and give it two different tasks in the same chat, then ask it to combine them. Watch what details it keeps or mixes up. Source: [reddit.com](https://www.reddit.com/r/artificial/comments/1t6yo2f/ai_agents_fail_in_ways_nobody_writes_about_heres/) ### Explain Like I'm 14 You know how when you’re riding your bike down the street, your speed keeps changing in tiny ways even if you’re not touching the brakes? First you notice you’re going faster (that’s velocity). Then you feel yourself speeding up or slowing down (that’s acceleration). Then there’s the jerk when you suddenly hit the brakes and everything lurches. Now imagine those changes themselves start changing. The fourth level of change is called “snap.” The fifth is “crackle.” The sixth is “pop.” It sounds like cereal, but it’s just a way to describe how motion can get more and more complicated the longer you track it. An AI can build a little world where you drag sliders and watch a dot move with all these layers of change at once. It doesn’t magically understand physics the way a scientist does. It learned patterns from tons of examples of moving objects, then used those patterns to guess what should happen next when you change the sliders. So next time someone says “higher-order derivatives,” you can tell them it’s basically the AI keeping track of how the changes are changing, just like noticing that your bike’s wobble gets worse the faster you go. Not so scary, right? ### Cool Stuff & Try This **Play with snap, crackle, and pop in real time:** Ethan Mollick (One Useful Thing) A simple webpage lets you control velocity, acceleration, jerk, snap, crackle, and pop and watch how a moving object reacts. It was built so anyone can see these physics ideas without doing math. This is cool because it turns something that usually feels like boring textbook stuff into a toy you can mess with. If you like games or animation, seeing how tiny changes in speed create different feels helps you understand why characters in games move the way they do. Go to https://motion-derivatives-exhibit.netlify.app/ on your laptop or phone. Drag the sliders for snap, crackle, and pop while watching the dot. Try making only snap high and everything else low, then switch to making pop high. Notice how the motion changes personality. Source: [x.com](https://x.com/emollick/status/2052605991078756658) **Bumble wants to replace swiping with an AI helper:** Engadget Bumble is building an AI that can suggest matches for you instead of making you swipe through hundreds of profiles. The idea is that the AI learns what you usually like and brings those people to the top. This matters if you or your friends use dating apps. It could save time, but it also raises questions about whether the AI will show you a narrow slice of people or help you discover new types. You can’t try the new feature yet, but you can open the Bumble app and think about what details you’d want an AI to notice about your past likes and dislikes. Source: [engadget.com](https://www.engadget.com/2167487/bumble-will-replace-swiping-right-with-something/) ### Quick Bits **Meta wants an AI that can shop for you on Instagram:** Engadget Meta is reportedly working on an AI agent that could browse Instagram and actually buy things for you, similar to other shopping agents. It would live inside the apps you already use every day. Source: [engadget.com](https://www.engadget.com/2167475/metas-ai-agent-plans-reportedly-include-an-openclaw-competitor-that-can-shop-on-instagram/) **The US and China may start official AI talks:** The Decoder Leaders from both countries are exploring formal conversations about how to handle powerful AI systems. This is still early, but it shows governments are starting to treat AI like something that needs international rules. Source: [the-decoder.com](https://the-decoder.com/the-us-and-china-are-considering-formal-talks-on-ai/) |
💬 Reply to this email — Patrick reads every one. |
Catch up on more Models & Agents for Beginners: Nerra Network · AI-narrated voice (Grok TTS) · Editorial by Patrick You're receiving this because you subscribed to Models & Agents for Beginners on nerranetwork.com. |
| Issue #32 · Models & Agents for Beginners · May 8, 2026 |
Don't miss what's next. Subscribe to Nerra Network:
