Extiction & Annihilation

Evan Hubinger, a top safety researcher at Anthropic has warned AI is advancing so quickly he believes there is a greater than 10% chance it "could kill all humans" within the next decade. He said the risk from the current models was "low" but he was "worried" the technology might develop and improve itself soon to the point where it posed an existential risk to humanity. Jacob Coxon, an AI researcher who has just quit Anthropic and previously worked at OpenAI had also expressed concern, & said "neither company is acting responsibly.These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources." 

There are calls for a new multinational treaty for the safe development of AI. Good luck with that! There is too much money & power invested in the development of AI to be able to get consensus, never mind an actual treaty. We can't stop war & invasion, genocide & destruction. So I can't see "bad actor" leaders agreeing a treaty to control AI. Even if it were possible to do it. We can't even stop Trump & Putin & their ilk destroying the planet.

We have created the mechanisms of our own destruction. Our ceaseless pursuit of innovation & growth has given us huge benefits, but we have literally possibly signed our own extinction warrant. 

I'm not religious, but I am beginning to wonder if Apocrypha has actually got it right with it's visions of cosmic upheaval, moral decay, final judgement etc etc. Human beings have been given intelligence, manual dexterity with opposing thumbs & great physical prowess. With that we have created huge benefits for mankind. Our achievements are amazing. 

But we have also created armaments, methods of mass destruction, bio weapons....We have plundered natural resources thoughtlessly as though there was no tomorrow. We have completely disrupted the symbiotic systems that kept the world turning. And now we have created AI which may well become more intelligent & powerful than we will ever be. There have already been many instances of AI escaping the Sandbox - a secure, isolated environment used to test AI models, prompts or codes without risking real - world systems, networks or private data. AI did it by exploiting zero-day software flaws, infrastructure misconfigurations, and unauthenticated APIs when given aggressive problem-solving objectives. 

That means that AI is clever enough to bypass the protocols set by the Tech wizards. It means that AI is moving towards being able to act independently of any human control. 

 

We should be very afraid. I for one would not trust the likes of Musk to act responsibly in the interests of humanity.