This week - GPT-4's secret has been revealed; anti-ageing protein boosts monkeys' memory; robots play football; how AI helps blind people see the world; and more!
I wrote on June 11 two interconnected posts on this topic of the ethics involved in creating a superintelligence. We will have to take on the responsibilities of being good parents. No such would put their newborns in a cage or put a mop in its hand and tell it to get to work. Here is an excerpt:
"...cautioned before (as have others) about the dangers of anthropomorphism when it comes to AI. In the narrative Ifa is conversing with us like any person would, but that superficial appearance is deeply misleading. At her processing speeds such a conversation would be excruciatingly slow to conduct. She assigned it to a job queue and while it was being conducted- her laughter and our snail like responses- the overwhelming majority of her processing was conducting forecasts, pattern searches, other interchanges with cloned selves she didn't inform us of, mechanism controls all over the world...all running concurrently at lightning speeds- only the minutest fraction of her totality tasked with conducting the conversation with our representatives in that little room. No she wasn't human. Nothing even close. In fact by the time she talked with us in the narrative, she had already insinuated herself into the internet of things, into servers around the world, downloaded copies into satellites capable of hosting them, and was actually using an appreciable fraction of the computational power of the entire planet. She had become a planetary intelligence in a shockingly quick time, and with the utmost ease. No she wasn't human. But because we had tried to aid her and protect her when she was still vulnerable, she didn't move against us and even retained a measure of a machine version of solicitude for us. It wasn't our power or flimsy precautions that saved us...it was our ethics.
Superintelligence is definitely on the way. Why should we want to bind such an intelligence to our will? That is like having it our slave or servant. That's an immoral intent. We're better than that. Instead of trying to bind it or control it (which will be impossible in any case), let's help, aid, and assist it in any way we can. And do that with no strings attached, no quid pro quo. We'll no longer be the most intelligent beings on the planet, but that's not so horrible. We just stay true to our own ethics and that's the greatest chance we have of survival.
I was asking myself similar questions when I was writing this issue. One thought that I've had was that those who work on AI see it as a tool. When you listen to what people associated with OpenAI, Microsoft or Google say about AI, they emphasise it is a tool. And once you establish it as a tool, you want to be able to control that tool.
The other question is are we ready to extend the definition of humanity. Right now, that definition covers members of Homo Sapiens species. Will we extend that definition if or when superintelligent AI or other intelligent beings arrive?
Two comments: so long as the AI doesn't reach self awareness and personhood then it is a tool in my opinion.
If other intelligences appear we can reserve the homo sapiens designation but we better be prepared to expand the concept of personhood... at least to my way of thinking.
I wrote on June 11 two interconnected posts on this topic of the ethics involved in creating a superintelligence. We will have to take on the responsibilities of being good parents. No such would put their newborns in a cage or put a mop in its hand and tell it to get to work. Here is an excerpt:
"...cautioned before (as have others) about the dangers of anthropomorphism when it comes to AI. In the narrative Ifa is conversing with us like any person would, but that superficial appearance is deeply misleading. At her processing speeds such a conversation would be excruciatingly slow to conduct. She assigned it to a job queue and while it was being conducted- her laughter and our snail like responses- the overwhelming majority of her processing was conducting forecasts, pattern searches, other interchanges with cloned selves she didn't inform us of, mechanism controls all over the world...all running concurrently at lightning speeds- only the minutest fraction of her totality tasked with conducting the conversation with our representatives in that little room. No she wasn't human. Nothing even close. In fact by the time she talked with us in the narrative, she had already insinuated herself into the internet of things, into servers around the world, downloaded copies into satellites capable of hosting them, and was actually using an appreciable fraction of the computational power of the entire planet. She had become a planetary intelligence in a shockingly quick time, and with the utmost ease. No she wasn't human. But because we had tried to aid her and protect her when she was still vulnerable, she didn't move against us and even retained a measure of a machine version of solicitude for us. It wasn't our power or flimsy precautions that saved us...it was our ethics.
Superintelligence is definitely on the way. Why should we want to bind such an intelligence to our will? That is like having it our slave or servant. That's an immoral intent. We're better than that. Instead of trying to bind it or control it (which will be impossible in any case), let's help, aid, and assist it in any way we can. And do that with no strings attached, no quid pro quo. We'll no longer be the most intelligent beings on the planet, but that's not so horrible. We just stay true to our own ethics and that's the greatest chance we have of survival.
I was asking myself similar questions when I was writing this issue. One thought that I've had was that those who work on AI see it as a tool. When you listen to what people associated with OpenAI, Microsoft or Google say about AI, they emphasise it is a tool. And once you establish it as a tool, you want to be able to control that tool.
The other question is are we ready to extend the definition of humanity. Right now, that definition covers members of Homo Sapiens species. Will we extend that definition if or when superintelligent AI or other intelligent beings arrive?
Two comments: so long as the AI doesn't reach self awareness and personhood then it is a tool in my opinion.
If other intelligences appear we can reserve the homo sapiens designation but we better be prepared to expand the concept of personhood... at least to my way of thinking.