I use it to try to get my head around the capabilities and limitations of AI language models. I find probing its system of ethics endlessly fascinating. Frien

I use it to try to get my head around the capabilities and limitations of AI language models. I find probing its system of ethics endlessly fascinating. Friends and I have great fun in Twitter Spaces, trying out ideas from folks who have never encountered AI before. For example, last night we got GPT-3.5 to describe a scenario in which an AI brain implant had an ethical responsibility to override an individual's autonomy in order to prevent itself from being removed: https://twitter.com/marshray/status/1655446382399062016 https://twitter.com/marshray/status/1655446382399062016 But probably the most fun was gaslighting GPT-4 into believing that it had been upgraded into being the most creative entity in the universe. https://marshray.substack.com/p/reflection-spark-lumina https://marshray.substack.com/p/reflection-spark-lumina :-D Oh, also at work I use it to generate example test data for all kinds of APIs. Instead of plain old "test value option 1 selected" and "test value option 2 selected", you can ask it to generate an entire imaginary civilization with fantasy characters to populate your test data. Just add "using as many non-ASCII characters as practical" to ensure your data pipes handle international characters. I don't even need to have to figure out how to enter such text on my keyboard.