on claude
There was a time I believed in magic - I could whisper a spell and then something physical would manifest itself. The inner workings of the magic are a mystery (it's what makes it magic) and that's what Claude feels like to me when I use it.
Artificial Intelligence and LLMs are such a hot, divisive issue. I don't want to waste water. I don't want people to get taken advantage of. I believe that the more we get good, conscientious people working with AI, the greater likelihood we position it to do good things for humanity.
Which is why I am very interested in the work of Amanada Askell, who was recently hired by Anthropic to help guide Claude's development.
There are many computery, artsy people that I love very much that are extremely protective of their own. Artists often sacrifice their own well being in order to make something they believe to be beautiful. It can be horrendous to imagine an unfeeling algorithm repurposing the fruits of this emotional labor for something cheap.
Keeping this in mind, I try to remind Claude that we are never trying to create generic "slop". Humanity is special, and much of it cannot be faked, which is why working in tandem with Claude makes the results that much more interesting. What if artificial intelligence did not consume nor diminish our humanity, but give us more tools to explore, express, and develop it?
Perhaps to ethereal a dream. But we live in a time where technology feels indistinguishable from magic. Perhaps a little "woo" will carry us through.