Ask Grok or any other ''artificial intelligence'' about the Scottish COVID inquiry health and social care hearings closing statments. I'm sorry that does not compute. Like many people i approach about it. lol
AI works by generating probability distributions that match the training dataset. Once trained, for any query it tries to find patterns of words, sentences and paragraphs (or sets of ideas/notions) that align with both the query and the prior probability distributions generated as well as the dialogue history, or even your personal profile, in many instances. If you give it a very specific question it will answer with the most frequent set (e.g. the covid narrative). The only way for it to be objective is if you ask something like "Assume this scenario is true. What would happen then?". Then it will give a better answer.
PS: Some AI systems might have been trained to be quite biased in certain topics.
It's not really artificial intelligence in my view (that's sales talk) it's just a computer program specifically designed by the establishment to generate answers that suit them. Just ask it about COVID jab safety or lockdowns and masks saving lives. Of course it's also so smart and independent it knows very little to nothing about the world's only official COVID inquiry to give real families a voice.
It is just a tool. The only way for it to be useful in the future would be via decentralized AI, where the user can be certain that there are no biases during the training process. This can become feasible with more affordable hardware, better chips, smaller supercomputers... If not, it will be monopolized, controlled and dangerous, as you describe. For now, it is not even working properly yet. All the available systems are beta versions.
Ask Grok or any other ''artificial intelligence'' about the Scottish COVID inquiry health and social care hearings closing statments. I'm sorry that does not compute. Like many people i approach about it. lol
https://biologyphenom.substack.com/p/scottish-covid-19-inquiryclosing
AI works by generating probability distributions that match the training dataset. Once trained, for any query it tries to find patterns of words, sentences and paragraphs (or sets of ideas/notions) that align with both the query and the prior probability distributions generated as well as the dialogue history, or even your personal profile, in many instances. If you give it a very specific question it will answer with the most frequent set (e.g. the covid narrative). The only way for it to be objective is if you ask something like "Assume this scenario is true. What would happen then?". Then it will give a better answer.
PS: Some AI systems might have been trained to be quite biased in certain topics.
It's not really artificial intelligence in my view (that's sales talk) it's just a computer program specifically designed by the establishment to generate answers that suit them. Just ask it about COVID jab safety or lockdowns and masks saving lives. Of course it's also so smart and independent it knows very little to nothing about the world's only official COVID inquiry to give real families a voice.
It is just a tool. The only way for it to be useful in the future would be via decentralized AI, where the user can be certain that there are no biases during the training process. This can become feasible with more affordable hardware, better chips, smaller supercomputers... If not, it will be monopolized, controlled and dangerous, as you describe. For now, it is not even working properly yet. All the available systems are beta versions.
I noticed it is actually very good at writing letters.