I’m no expert in artificial intelligence, but I’ve spent hours reading up and was lucky enough to speak with @unconst1 and @DoctorM_DO whose passion is palpable
In this thread we’ll talk the basics of AI, the threats of centralization, and where #Bittensor fits in
2/
Let’s start with #OpenAI so we can translate their structure to Bittensor’s model
Open AI is the company behind the last 3 iterations of #GPT
GPT is the neural network, essentially a fine tuned artificial human brain, which is the infrastructure that apps are built on
Apps are built on GPT to leverage the learning model. In general, more data with quality "inputs" and "weights" result in a better learning model to build on
But think
#GPT-3 cost at least $10 million to train on 175 billion parameters > 10x any other learning model...
5/
And this model is siloed to one private company and its own data set. Even within #OpenAI, each new model has to be completely re-trained
So all the previous value from GPT-1 and 2 was lost
Plus, each iteration costs exponentially more money, time and computing power/energy