The Big Story

OpenAI added a feature to ChatGPT that lets it create interactive graphics, buttons, charts, and even simple tools like calculators inside your conversation. The new capability is called Intelligent UI and it works by giving the model a library of ready-made visual components it can combine on the fly. Instead of just typing words, the AI can now pull from a library of visual components and decide which ones best answer your question. Think of it like the difference between reading a recipe and having a kitchen timer and measuring cups appear on screen so you can actually follow along. The system was trained to know when a simple sentence is enough and when a graph or interactive diagram would help more. For students, this means explaining something like how a bicycle works or how retirement savings grow can now include a working model you can play with. For anyone doing homework or creative projects, answers become easier to understand because you can see and touch the information instead of just reading it. The change rolled out first to paid users and is now reaching free accounts as well. GPT-6 itself launched for paid ChatGPT users in September, replacing the earlier GPT-5.6 model, but Intelligent UI was added later. It is available right now to ChatGPT Plus, Pro, Business, and Enterprise subscribers, with free and Go tier users gaining access the following day. You can try it yourself by opening ChatGPT on your phone or laptop and asking something visual, like “show me an interactive diagram of how photosynthesis works” or “make a calculator to split a dinner bill.” The AI will decide whether to add buttons or charts based on what you asked. Source: sea.mashable.com

Explain Like I'm 14

You know how when you’re texting, your phone guesses the next word based on what you’ve typed so far? Now picture that same idea, but instead of guessing words the system also guesses whether a picture, a button, or a chart would help you understand the answer better. It has a box of ready-made pieces—like a chart template, a button you can tap, or a simple calculator—and it learns which pieces fit different kinds of questions. When you ask about money over time, it might pull the chart piece and fill it with numbers. When you ask how to split a bill, it might build a working calculator instead of just describing the math. The training happens by showing the model thousands of examples so it gets better at choosing the right piece at the right moment. Over time it also learns when to skip the extras and just reply with plain text. That’s the core idea behind the new visual responses you’re starting to see.

Cool Stuff & Try This

Check if an image was made by AI in seconds Google just opened a free website where anyone can upload a picture, video, or audio file and see if it carries a hidden digital watermark that proves it came from an AI system. The tool works for content made by Google, OpenAI, Nvidia, and soon Apple. It’s useful because it helps you spot when something online might not be real. To date the company has already watermarked 180 billion images and videos plus more than 240,000 years of audio, and the site handles about a million verification requests every day. Go to synthid.com on your phone or laptop, upload any image you want to check, and the site will tell you right away if it finds the watermark. No account is needed for basic checks. Source: synthid.com

Meta’s AI helper now works on iPad Meta’s agent called Muse, which can help with creative tasks and answer questions, just added full iPad support after launching on iPhone only a month earlier. If you already use the iPhone version, the iPad app feels like a bigger, more comfortable workspace for the same AI. You can try it by searching “Meta Muse” in the App Store, downloading the free app, and asking it to help plan a story or generate ideas for a school project. Source: techcrunch.com

Quick Bits

AI agents take a real car to In-N-Out Three engineers let different AIs (GPT, Claude, and Grok) control an actual Toyota Corolla to drive to a fast-food restaurant. Only one succeeded, showing how far we still are from reliable self-driving AI helpers. Source: wired.com

A new push to predict how cells behave with AI A big research effort backed by Mark Zuckerberg and Priscilla Chan is spending $1.8 billion to train AI models that can forecast what living cells will do. Meta, Google DeepMind, Isomorphic Labs, and the US Department of Energy are helping fund the data, lab equipment, and compute needed. The first big dataset is expected in about a year and could help future medicine. Source: the-decoder.com

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.