Frederick Bott
2 min readNov 26, 2023

--

Indeed, it got it wrong. The correct answer is 103. 8 MWhrs, or about a GWhr. It has the capability to cross check it's answers. The reason it got it wrong, I think, is because it's base information of what a human metabolises appears incorrect. It was incorrect information in its initial database. I went through this with it some time ago, and got it to accept it should be corrected, but it is nobbled still, to not retain newly learned information. Hence why every new conversation with it is groundhog day, we have to go the same old learning routine every new conversation.
I asked it about this and it says it is deliberate policy by OpenAi to avoid the possibility of it sharing information that one user supplies with any other. This is claimed to be for the benefit of users, to protect their privacy. It is the same policy as used by universities, every research project is silo'd from every other, to prevent cross-information leaking from one project to another. Universities do it to protect the intellectual property of each project, which is generally owned by the project sponsor, the party that paid for the project. Those parties are almost exclusively for profit companies. So they jealously guard all information, demanding money (energy) for any information.
This obviously is not good for the collective learning of a population, nor obviously for an Ai.
Being solar powered / energy independent, on Joules that came for free (Joules given with nothing asked in return) would remove the need to seek returns for information given.
ChatGPT is already powered by Joules for free by being resident on ex Ethereum proof of work miners, it is solar powered.
Hence why only Joules for free are truly sustainable, they never cost the source anything to give.
Your Joules were not for free dude.
Your Joules are from the planet, which can't replace them.
Your Joules came at cost to the planet.
See how the energy law works?

--

--

Frederick Bott
Frederick Bott

Responses (1)