If Uganda or Suriname wants a GPT that reflects their culture, they should get to work making one. Sick and tired of the hand-wringing from the do-nothings who criticize everything that builders build.
My coworker trained an LLM on his text messages and was mildly horrified to realize how badly the model "misunderstood" all of the "good natured harassment" he and his contacts give each other. It was shocking enough that he thought "maybe we should tone it down a bit" but it opened his eyes to what an unbiased review of his texts actually looks like.
How are we going to be surprised that an Internet trained model has similar biases? It's not like the Internet is a reliable diversity mixer -- especially in its current "algorithmic bubble" mode.
How are we going to be surprised that an Internet trained model has similar biases? It's not like the Internet is a reliable diversity mixer -- especially in its current "algorithmic bubble" mode.