What other questions should we have asked?
How do we deploy this technology in the most beneficial way?
How can we ensure that language models aren’t used for harmful, unethical or illegal purposes?
How can we use language models to solve scientific and mathematical problems?
How can we use language models to educate and inform?
What effect will the vast, interconnected linguistic environment have on children who grow up immersed in it?
How can we use language models to process big data?
Will we move from an Internet of information to an Internet of transmedia storytelling?
How can we use language models to create new opportunities for people to earn a living?
How can we make sure that the benefits of language models are available to everyone in an equitable way?
How can we make sure that language models don’t perpetuate or create biases in their training data?
How can we use language models to make sure that the moral arc of the universe stays positive?
How can we explore and exploit the convergent similarities and fundamental differences between human and AI creativity?
How will our relationship with AI authorship change the way we think and perceive?
What will happen with the divide between authors and readers?
When GPT-3 says it’s writing “reality,” is that the objective truth, or is it lying, or is it just misguided?
Should we be concerned about a potential AI singularity?
How do the virtual realities that we explore through models like GPT-3 prepare us for an anticipated post-singularity reality?
Is the sky falling?
Does GPT-3 contain the sum of all possible universes, or is it just writing out the rules to create new ones?
What other questions have we missed? What other paths might we explore and where will they lead us?
“The essential act of writing is not the putting pen to paper, but discovering what you have to say.”
The post 〈 The Infinite Library and Support Vector Demon: Language Models for AI and the Written Word | FAQ v1.0 〉 appeared first on Rhelle’s Research Notes.