@ninadschick - AI Dynamics

AI’s Trajectory: From Knowledge Work to Superintelligence

By

–

31 January 2025 19h00

And even crazier: If we extrapolate from AI’s current trajectory, its impact on knowledge work is obvious. But that’s just the beginning. This leads to another revolutionary phase: once we’ve built an 'AI brain' that’s smarter than us—one that can reason, 'think', and

→ View original post on X — @ninadschick

31 January 2025

AI Masters Logic First: Why Creativity Remains Elusive

By

@ninadschick

–

31 January 2025 17h49

The paradox of AI: The jobs we assumed required the most human intelligence—math, coding, complex reasoning—are the ones AI is mastering first. Why? Partly, because they are AI-adjacent. Meanwhile, AI still struggles to write truly groundbreaking prose (think a new Shakespeare)

→ View original post on X — @ninadschick

31 January 2025

AI Model Terms of Service: Synthetic Training Data Restrictions

By

@ninadschick

–

30 January 2025 4h05

And be that as it may, isn’t it a totally ridiculous stipulation in the TOC anyway? Use the model for anything except as synthetic training data?

→ View original post on X — @ninadschick

30 January 2025

Competitive AI Breakthrough Strategy Exposed to Market

By

@ninadschick

–

29 January 2025 20h07

I assume because they wanted to keep breakthrough close to their chest for competitive advantage? If that was the case the competition has clearly been blown wide open.

→ View original post on X — @ninadschick

29 January 2025

Scaling Reasoning Models: From Thousands to Millions of GPUs

By

@ninadschick

–

29 January 2025 18h57

This is the interesting bit. As we move to the next phase of scaling reasoning models w RL, data and compute converge. Next breakthroughs require moving from 100’000’s of GPUs to millions of GPUs.

→ View original post on X — @ninadschick

29 January 2025

NVIDIA GPU Infrastructure Cost Analysis for AI

By

@ninadschick

–

29 January 2025 4h36

If they have 50,000 NVIDIA GPUs at $30-40k each, that’s $1.5 – 2bn. Just saying.

→ View original post on X — @ninadschick

29 January 2025

DeepSeek disrupts AI market with cheaper open source alternative

By

@ninadschick

–

29 January 2025 4h30

The markets are wrong about DeepSeek. A Chinese hedge fund just made AI: – As capable as o1 – 95% cheaper to run – Completely open source Cue panic about CapEx on AI. The end to 'American AI supremacy.' But they missed the real story: This is MORE reason to be

→ View original post on X — @ninadschick

29 January 2025