Retrain, Relearn, Remodel: Decentralising Reinforcement Learning

[mozfest-bot]

[ UUID ] c06435ac-9ab6-4736-9833-21939e85b467

[ Session Name ] Retrain, Relearn, Remodel: Decentralising Reinforcement Learning [ Primary Space ] Decentralization [ Secondary Space ] Web Literacy

[ Submitter's Name ] Oliver Spall [ Submitter's Affiliated Organisation ] BBC / Self [ Submitter's Github ] @oliverspall

[ Additional facilitators ] TBD

What will happen in your session?

We can't decouple AI from how we live our lives, so why not figure out how to live with it?

Current networks developing AI offer limited scope for change, beyond specific purposes controlled by a small minority.

Why not work out how different (human and machine) networks can affect how deep learning learns, making more fluid, dynamic systems which change based on a much more open network of influence.

We'll think about what role AI should play in relation with humans, and vice versa.

We'll play with deep learning models and explore what humans need to do differently to live in tune with AI.

We'll build speculative systems where humans with less skills than mathematicians can interact with the training process of ANNs.

What is the goal or outcome of your session?

Speculative, networked approaches to living with deep learning systems that aim to reduce inequality.

Discussion and open commentary.

Proposals to decentralised approaches to developing AI with ANNs.

Increased literacy around deep learning methods.

Better ways to live with AI than we have now.

Time needed

All weekend, as an installation, exhibit or drop-in session

[cubicgarden]

Nice!

[cubicgarden]

@marcwalsh I think we can shut this one, as @oliverspall can't make it anymore

[jontutcher]

Oli can't do this, so I'm closing the issue :)