Why should AI be decentralized?

From the podcast Infinite Machine Learning ~ Compressed

Issue no 1
• AI is non-compounding
• In itself, AI is non-compounding
• Research however is compounding
• Siloed Models stop knowledge sharing

Issue no 2
• No individual access to powerful computers or AI resources
• Only big companies or academia can provide resources
• No chance of the individual contribution

Issue no 3
• The world is moving towards decentralized models
• AGI cannot be achieved by a singular entity
• Requires more resources for innovation
• Innovation is scarce

It seems all well and good when it's in theory but when it comes to practical terms that's when we are faced with problems. Onemajorproblem i see in decentralized system is less learning opportunity from experienced people. I understand the need for decentralized systems oo but in a centralized system it allows new people to work under experienced and learn from them which should also be included in the decentralized systems.

I agree with Lambda, although it seems all well and good to decentralise networks to allow open access to people but it does come with its fair share of problems, such as Co-Ordination Difficulty which is almost impossible to achieve through online mediums even if you have best platforms available. Also, a lot of resources will go to waste in such a model.

For me, the potential challenge with decentralized AI is the lack of learning opportunities from experienced individuals which are quite understandable. In a centralized system, new people can work under experienced individuals and learn from them. While there may be some challenges to implementing decentralized AI systems, such as a lack of learning opportunities from experienced individuals, the potential benefits outweigh the drawbacks. Decentralization allows for greater access, innovation, and progress in the field of AI.

Such powerful technology as AI has no right to be kept in the hands of a few corporations.