The Algorithmic Bridge

The Algorithmic Bridge

Share this post

The Algorithmic Bridge
The Algorithmic Bridge
Weekly Top Picks #53
Copy link
Facebook
Email
Notes
More
Weekly Top Picks

Weekly Top Picks #53

Q* / Karpathy LLMs / GPT-4 Turing Test / ChatGPT voice / Hinton vs LeCun / SD Video / Claude 2.1 / Sutskever TED talk / Orca 2 / LCMs / Lyria / ChipNeMo / GraphCast / Emu / Tldraw / and more

Alberto Romero's avatar
Alberto Romero
Nov 27, 2023
∙ Paid
10

Share this post

The Algorithmic Bridge
The Algorithmic Bridge
Weekly Top Picks #53
Copy link
Facebook
Email
Notes
More
3
2
Share

I didn’t publish the weekly top picks column last Monday because the OpenAI crisis was still unfolding. That means I’ve accumulated two week’s worth of news and events to share with you. I will try to fit as many entries as I can while keeping it tight — so many things have happened these two weeks!

This review will focus mostly on releases and announcements more than essays and articles. As always, here’s a quick overview for free subscribers so that you consider going premium to get the whole thing.

There’s a lot to digest but don't panic; I’ve shortened each entry as much as I could so that it doesn’t get too overwhelming.

The week(s) at a glance

  • Q-star: AI breakthrough or hype?

  • Must watch: Andrej Karpathy’s intro to large language models

  • Why GPT-4 doesn’t pass the Turing Test

  • ChatGPT with voice for all free users

  • Geoffrey Hinton vs Yann LeCun on AI risk

  • Stability AI’s Stable Diffusion Video

  • Frontier AI (e.g., GPT-4) is all about compression, not intelligence

  • Claude 2.1 has a 200K-token context window

  • Is the next AI breakthrough synthetic data or new architectures?

  • Ilya Sutskever TED talk: The exciting, perilous journey toward AGI

  • Microsoft’s Orca 2: Teaching small language models how to reason

  • AI-generated songs: The decoupling of consumption and creativity

  • Google delays release of Gemini AI that aims to compete with OpenAI

  • Latent Consistency Models: The future of drawing

  • Nvidia ChipNeMo: Designers tap generative AI for a chip assist

  • Meta AI’s Emu Video: Text-to-video generation

  • DeepMind’s Lyria: Transforming the future of music creation

  • Mistral AI CEO explains his stance on the EU AI Act

  • Virtual AI tutors tailored for kids are coming to online schools

  • UK will refrain from regulating AI ‘in the short term’

  • Tldraw: No code AI tools are getting magical

  • DeepMind’s GraphCast: AI model for faster and more accurate global weather forecasting

  • Jimmy Apples leaks OpenAI will have a 1-10B model > GPT-4 by the end of 2025

Get 20% off a group subscription

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Alberto Romero
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share

Copy link
Facebook
Email
Notes
More