In partnership with

"We're not building animals. We're building ghosts."

That's Andre Karpathy explaining why AI won't look like biology—and why timelines are longer than the hype suggests.

After 15 years building AI (GPT-2, Tesla self-driving, now education), he's got frameworks that actually work.

• The march of nines.

• Model collapse.

• Why coding is different.

• What humans will do post-AGI.

and more…

But first, quick word from Hubspot -

Turn AI into Your Income Engine

Ready to transform artificial intelligence from a buzzword into your personal revenue generator

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

• A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential

• Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background

• Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get Your Guide

AI progress is way slower than you think.

The "intelligence explosion" already happened. And humans won't become obsolete—they'll become superhuman.

Andrej Karpathy has the receipts. And after watching him nail prediction after prediction while others got it wrong, I extracted every framework from his latest interview.

Including the one about why reinforcement learning is "terrible" (but we need it anyway) and why the geniuses of today are "barely scratching the surface" of what human minds can do.

Full post →

The Simplest Way To Create and Launch AI Agents

Imagine if ChatGPT, Zapier, and Webflow all had a baby. That's Lindy.

With Lindy, you can build AI agents and apps in minutes simply by describing what you want in plain English.

From inbound lead qualification to AI-powered customer support and full-blown apps, Lindy has hundreds of agents that are ready to work for you 24/7/365.

Stop doing repetitive tasks manually. Let Lindy automate workflows, save time, and grow your business.

Get $20 Worth of Free Credits

Key Takeaways: What You'll Learn

Understanding AI Reality:

• The March of Nines - Why every 9 of reliability (90% → 99% → 99.9%) takes constant effort, explaining why demos ship fast but products take decades

• Ghosts vs Animals - We're training AI on internet documents (mimicking humans), not running evolution—fundamentally different intelligence

• The Demo-to-Product Gap - Self-driving looked "done" in 2014, still isn't economical in 2025; same pattern will repeat with AI agents

What Actually Works Now:

• Autocomplete is magic, VIP coding is slop - High-bandwidth communication (type 3 letters, get completion) beats low-bandwidth (describe in English)

• The Autonomy Slider - Stop thinking "replacement," start thinking "what % of this job can AI handle?" (call centers: 80%, novel coding: 20%)

• Coding works because it's text-native - Pre-built infrastructure (IDEs, diffs, tests) makes coding the perfect first domain; everything else is harder

Why Timelines Are Longer:

• 5 missing pieces - Continual learning, true multimodality, context management, AI culture/collaboration, actual reasoning—each takes years

• Reinforcement learning is terrible - "Sucking supervision through a straw"—upweighting entire trajectories based on one final number

• Model collapse is real - AI trained on its own output gets dumber; maintaining entropy is an unsolved problem

The Cognitive Core Insight:

• Smaller might be better - Future AI could be 1B parameters (vs current trillions), focused on thinking algorithms instead of memorized facts

• Memory is a bug, not a feature - Models being too good at memorization prevents generalization and off-manifold thinking

Human Future (The Optimistic Part):

• Education becomes fun, not useful - Post-AGI, learning is like gym (we don't need physical strength, but people still work out)

• Geniuses are at 10% capacity - Current bottleneck is bad teaching, not human limits; perfect AI tutors unlock superhuman learning

• The intelligence explosion already happened - We've been in recursive self-improvement since the Industrial Revolution; AI continues the trend

Practical Frameworks:

• First principles thinking - Find the first-order terms (like micrograd: 100 lines captures ALL of neural network training)

• Present pain before solutions - Best teaching: let people try, fail, then appreciate the answer

• Translation invariance in time - Look back 10 years to predict next 10; progress requires algorithms + data + compute + systems improving together

Timeline Reality Check:

• Decade of agents, not year - Fundamental capabilities need multiple breakthrough cycles

• No discrete "foom" moment - Gradual diffusion, like every technology before it (can't find iPhone or internet in GDP curve)

• Physics > hype - Train yourself to see spherical cows; approximate, then refine

Effortless Tutorial Video Creation with Guidde

Transform your team’s static training materials into dynamic, engaging video guides with Guidde.

Here’s what you’ll love about Guidde:

1️⃣ Easy to Create: Turn PDFs or manuals into stunning video tutorials with a single click.
2️⃣ Easy to Update: Update video content in seconds to keep your training materials relevant.
3️⃣ Easy to Localize: Generate multilingual guides to ensure accessibility for global teams.

Empower your teammates with interactive learning.

And the best part? The browser extension is 100% free.

Check out Guidde

That’s it for this week

See you next week

Cheers,

Ayush & Aditi