The Inside View

Raphaël Millière Contra Scaling Maximalism

Jun 24, 2022

02:27:12

Creator website

Highlights

AI Chapters

Episode notes

The Impact of Atis Systems in the Real World

03:47

Is Scaling All You Need?

04:19

Compositionality in the Linguistic Domain

04:11

Introduction

5min

Is There a Superhuman Intelligence?

4min

What Is the Difference Between Behavior and Knowledge?

5min

The Chinese Room Experiment, Is It Intelligent or Not?

4min

The Impact of Atis Systems in the Real World

4min

How Can an Octopus Hack the Communication System?

3min

The Octopus Is a Deep Sea Creature That Has No Interaction With the World Above the Surface

3min

The Language Model Is Bringing Together Different Issues

4min

I'm Not Sure What You Mean, but I Think It's a Reasonable Take.

6min

10.

Do You See a Future for Philit Professors in Am, Correcting Essays?

4min

11.

The Sarcastic Parrot's Metaphor Is Short Selling What These Models Are Doing

3min

12.

Isal Monkey Can Learn Fusota Tasks Very Well

2min

13.

The Limitations of Dhep Learning

5min

14.

The Difference Between Local Generalization and Brodgenation

6min

15.

Do You Have a Cofe Test?

3min

16.

Is It Possible to Scale Carotactectures?

4min

17.

Is There a Difference Between Garry and Guarmarcus?

2min

18.

The Scaling Laws of Scaling Models

4min

19.

Scaling Lows for Relationships

3min

20.

Scaling Maximumists - Scaling Maximalists

3min

21.

We Can Learn a Lot From Rodeta and Through Bigger Models

4min

22.

Is There a Need for Human Features in Deep Learning?

2min

23.

How to Train a Model to Play a Setant Game

2min

24.

Is Scaling All You Need?

4min

25.

Is Supervised Learning Really Working?

1min

26.

Is That What Humans Learn?

6min

27.

Scale Is All You Need

2min

28.

I'm Impressed by the Plearning Models

5min

29.

How Much Scaling Is Going to Change the Architecture?

2min

30.

Is Scale Really All You Need?

5min

31.

Is There a Catastrophe From Climate Change?

3min

32.

Is Scaling a Good Thing?

4min

33.

The Impact of Nuclear Weapons

2min

34.

Compositionality in the Linguistic Domain

4min

35.

The Compositional Cementics of a Text to Text Model

4min

36.

The Compositional Structure of a Complex Expression

6min

37.

Deep Learning Is Setting a Wall

4min

38.

Is There a Way to Do Chain of Thought Prompting?

3min

39.

Is There a General Solution to the Problem of Generating Images?

5min

40.

Compositional Intelligence

3min

Raphaël Millière is a Presidential Scholar in Society and Neuroscience at Columbia University. He has previously completed a PhD in philosophy in Oxford, is interested in the philosophy of mind, cognitive science, and artificial intelligence, and has recently been discussing at length the current progress in AI with popular Twitter threads on GPT-3, Dalle-2 and a thesis he called “scaling maximalism”. Raphaël is also co-organizing with Gary Marcus a workshop about compositionality in AI at the end of the month.

Transcript: https://theinsideview.ai/raphael

Video: https://youtu.be/2EHWzK10kvw

Host: https://twitter.com/MichaelTrazzi

Raphaël : https://twitter.com/raphaelmilliere

Workshop: https://compositionalintelligence.github.io

Outline

(00:36) definitions of artificial general intelligence

(7:25) behavior correlates of intellience, chinese room

(19:11) natural language understanding, the octopus test, linguistics, semantics

(33:05) generating philosophy with GPT-3, college essays grades, bullshit

(42:45) Stochastic Chameleon, out of distribution generalization

(51:19) three levels of generalization, the Wozniak test