When someone asks me a very technical question, I provide a vague and ambiguous response. I learned this habit in grad school. No one unless they are a knowledge domain expert in the field cares about the specifics. Often people ask me about AI. I say that it’s worrying for many reasons but mainly that AI demands so much energy. This is enough for them to pick up and carry the conversation. I could lecture them on many aspects of AI. But you’d have to pay for this torture. 😁
I generally explain to people that the current state of technology for LLMs is that they’re larger and more complicated versions of the text prediction on your phone, you know, when it guesses what word you want to put next, but with whole sentences using the entirety of the public Internet as the base of reference for that information.
And they basically burn through kilowatts of power per inquiry.
When someone asks me a very technical question, I provide a vague and ambiguous response. I learned this habit in grad school. No one unless they are a knowledge domain expert in the field cares about the specifics. Often people ask me about AI. I say that it’s worrying for many reasons but mainly that AI demands so much energy. This is enough for them to pick up and carry the conversation. I could lecture them on many aspects of AI. But you’d have to pay for this torture. 😁
I generally explain to people that the current state of technology for LLMs is that they’re larger and more complicated versions of the text prediction on your phone, you know, when it guesses what word you want to put next, but with whole sentences using the entirety of the public Internet as the base of reference for that information.
And they basically burn through kilowatts of power per inquiry.