Today LLMs are everywhere - But are they the future of AI?

Backpropagation algorithm for neural network training works well in todays AI tasks from pattern recognition to LLMs. But do they really?

LLMs need tons of data. And energy.

I would like to discuss and develop alternative learning mechanisms for AI in the next months. We could meet on the third day of Congress or afterwards - but please contact via Email hello@blufoo.de

References: The algorithms behind todays AI: https://radical.vc/geoffrey-hinton-on-the-algorithm-powering-modern-ai/

"Self Models of Loving Grace" by Joscha https://fahrplan.events.ccc.de/congress/2024/fahrplan/talk/3TSPQW/

-> See also his former presentations at Congress: https://media.ccc.de/search/?q=joscha