Startseite
Podcasts
Giants Shoulder Podcast
#121 Meet The Neuroscientist Teaching Neuron's in a Dish to Play Video Games

#121 Meet The Neuroscientist Teaching Neuron's in a Dish to Play Video Games

Giants Shoulder vor 3 Monaten

0:00

1:17:03

Join the #1 Community Dedicated to Rigorous & Radical
Biology: https://www.skool.com/the-giants-shoulder-6526/about

Brett Kagan is the Chief Scientific Officer at Cortical Labs, the
Australian startup that fuses living human neurons with silicon
chips to create a fundamentally new kind of intelligence. His
team's DishBrain paper showed that neurons in a dish learned to
play Pong in under five minutes — without any reward signal —
providing some of the strongest experimental evidence for Karl
Friston's Free Energy Principle.

Expect to learn how neurons on a chip actually play Pong and
Doom, why unpredictable stimulation drives learning better than
reward, what the Free Energy Principle really says and where
Brett disagrees with Friston, why 200,000 neurons outperform
800,000 on certain tasks, how Brett's baby daughter staring at
fractals led to a new theory of intelligence, why no AI system
can walk into your kitchen and make a cup of tea, what biological
data centres look like and why they already exist, and much more…

Timestamps:

00:00 Trailer

02:30 Brett Kagan — Neurons Playing Doom

03:04 The stepping stones from Pong to Doom

04:00 What is the CL1 and Cortical Cloud?

05:20 The moonshot goal of Cortical Labs

06:46 Giant buildings of neurons as computation hubs

07:13 The biocomputing data centre already running in Melbourne

07:38 Not building a brain in a dish

09:00 What does a biological computer actually look like?

09:29 Electricity as the shared language between biology and
silicon

10:49 How neurons on a chip play Pong

11:10 Topographic mapping inspired by the whisker barrel system

13:10 Why the neurons hit the ball — and what happens when they
miss

14:33 The Free Energy Principle explained simply

16:03 The trap door analogy for unpredictable stimulation

18:42 How the structure of neuron cultures shapes performance

19:19 Bees vs elephants: why bigger brains aren’t always better

21:37 Growing hippocampal place cells in a dish

24:12 Teaching neurons Morse code

28:27 The Free Energy Principle — cups, predictions and surprise

32:29 Where Brett disagrees with Friston

37:03 The dog treat problem — how do you reward a neuron?

39:40 Complexity as a hidden driver of intelligence

41:53 His baby daughter’s obsession with fractals

43:57 Defining agency — the three orders of information
processing

50:33 Change how you change over time

53:27 Why scientists are incentivised to be vague instead of
wrong

59:42 J.D. Bernal’s 1929 vision for brains beyond bodies

1:02:38 Standing on the shoulders of iPSC pioneers

1:03:14 Can neurons play chess? Moravec’s Paradox

1:04:12 No AI can walk into your kitchen and make tea

1:05:16 Augmenting what silicon is bad at

1:06:35 Is general intelligence even possible for silicon?

1:09:40 Why 200,000 neurons beat 800,000

1:11:16 Complexity and structure matter more than scale

1:11:58 Could biological neurons end up in your laptop?

1:13:36 Edge robotics and the future of biological computing

...

Email for sponsorship enquiries: evan@thegiantsshoulder.com