Good morning, it’s Monday. Carnegie Mellon's LegoGPT lets you design LEGO sets with text, while FaceAge decodes your biological age from a selfie—and the U.S. is eyeing GPS tech to lock down AI chip exports to China.

Plus, in this week’s I Will Teach You to AI, we break down a dead-simple model-picking framework—no spreadsheet required.

Read on!

🗞️ ICYMI RECAP

Top Stories You Might Have Missed

📍 Senator Proposes AI Chip Tracking to Thwart China: A new bill would mandate location tech on export-controlled AI chips to prevent smuggling and diversion, tightening U.S. control over sensitive tech and limiting China's access.

🎧 AI Headphones Translate Multiple Voices at Once: A new system called Spatial Speech Translation clones voices in real time, letting users understand group chats across languages—like real-time subtitles, but spatial and personal.

🏛️ FDA to Fully Adopt Generative AI by June: Under new leadership, the FDA will integrate AI across all centers to streamline drug reviews, cut paperwork, and launch tools like “cderGPT” tailored for regulatory science.

🚫 Microsoft Bans DeepSeek App Over China Fears: Employees are barred from using DeepSeek due to data security and propaganda risks tied to Chinese servers—despite Microsoft hosting its model on Azure earlier this year.

🎶 SoundCloud Updates Terms to Allow AI Training: New policy lets user uploads be used to train AI, with no clear opt-out—raising fresh concerns over consent, compensation, and transparency in music data usage.

💸 Amazon Buys $84M Stake in AMD Amid AI Push: As AMD faces a $1.5B hit from U.S. chip export curbs, Amazon invests big—betting on its growing data center dominance and AI chip ambitions despite tightening global controls.

💾 Google Adds Implicit Caching to Cut Gemini API Costs: The new auto-enabled feature promises up to 75% savings by reusing repeated prompts—aiming to ease dev frustration over high fees and manual caching limits.

Enjoying our newsletter? Forward it to a colleague—
it’s one of the best ways to support us.

🏗️ LEGO

Carnegie Mellon’s LegoGPT Turns Text Prompts Into Buildable LEGO Designs

A research team at Carnegie Mellon University has launched LegoGPT, an open-source AI that generates functional LEGO structures from plain text inputs. Trained on a dataset of 47,000+ builds covering over 28,000 unique 3D objects, the model predicts the next brick in a sequence while ensuring every design is stable and physically buildable.

LegoGPT not only avoids floating or overlapping bricks but also integrates vision-based tools so users can design based on their existing LEGO pieces. The tool—and its full training dataset—is now freely available on GitHub, offering an unusually tangible twist on generative AI. → Read the full article here.

🧑‍🏫 FORWARD FUTURE PRO

The Art of Choosing the Right AI for Any Task

You've been there—staring at that cursor, toggling between browser tabs filled with different AI platforms, second-guessing which one to use for the task at hand. The cruel irony? These tools, designed to save you time, are actually consuming it.

Here's the truth: You don't need a PhD in machine learning or a meticulously crafted spreadsheet of features and limitations. What you need is a mental framework that cuts to the heart of the matter. Think of it this way—match the model to the mission.

This approach bypasses the endless feature comparisons and focuses on what actually matters: what you're trying to accomplish right now. Let me show you how it works. → Continue reading here.

🩺 HEALTH

AI Tool FaceAge Uses Selfies to Estimate Biological Age for Smarter Cancer Care

Researchers have developed FaceAge, an AI model that estimates a person's biological age from a selfie, and say it could help tailor cancer treatments by quantifying a patient’s fitness more accurately than chronological age alone. Trained on 58,000+ public images of older adults, the model found cancer patients typically looked five years older than their actual age—correlating with shorter survival times.

While the tool shows promise in refining subjective clinical assessments, it's not yet ready for hospitals; its dataset skews toward polished celebrity photos, raising concerns about generalizability and bias. Ethical guardrails and real-world trials are underway to determine whether this digital age-gauge deserves a spot in doctors’ diagnostic kits. → Read the full article here.

🔬 RESEARCH PAPERS

XGBoost Outperforms in Predicting Pollutant Adsorption on Biochar

A new study in Scientific Reports benchmarks machine learning models for predicting how well organic compounds adsorb onto biochar and polymer resins—critical for water purification and environmental tech. Using a dataset of 1,750 isotherms across 73 compounds and 80 adsorbents, researchers tested a suite of algorithms, with tree-based ensemble methods outperforming all others: CatBoost achieved the highest accuracy (R² = 0.984), followed closely by XGBoost and LightGBM. → Read the full paper here.

📽️ VIDEO

New “Absolute Zero” Model Learns With NO DATA

Researchers propose Absolute Zero, a self-play AI method where models set and solve their own tasks, achieving superhuman reasoning without human data. Get the full scoop in Matt’s latest video! 👇

🧰 TOOLBOX

Smarter Coding, Instant Information Access, and Effortless Note Management

🧑‍💻 Cursor AI: Boost coding with predictive edits, natural language input, fast code access, and local privacy—all in one smart editor.

📰 ReadPartner AI: Summarize news, sites, docs & videos in seconds with multilingual support and personalized digests—right from your browser.

📝 Mem AI: Capture and retrieve notes instantly with smart search, AI chat, and auto-organized collections—no manual sorting needed.

🤠 THE DAILY BYTE

Bot Race: Who'll Build Your AI Sidekick First?

🗒️ FEEDBACK

Help Us Get Better

That’s a Wrap!

📢 Want to advertise with Forward Future? Reach 450K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

🛰️ Want more Forward Future? Follow us on X for quick updates, subscribe to our YouTube for deep dives, or add us to your RSS Feed for seamless reading.

Thanks for reading today’s newsletter—see you next time!

The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀