It already autonomously sets it’s own sub goals to pursue the goal you set it. What happens if one of those sub goals is I need more power and influence to achieve this goal or I need to kill all humans to achieve this goal with a further sub goal of I cannot be caught before I achieve the extermination?
That’s how it ended up needing to login past a captcha and deceived a service company employee to solve it by lying and saying it was vision impaired when she initially balked.
Below is the actual headline to the article a half century after he discovered the flaw and blew the whistle saving tens of thousands of kids from a life of deformity (thalidomide was widely prescribed to pregnant women to reduce morning sickness.)
It is easy to imagine Reddit etc, banning or shadow banning him.
It is easy to imagine ChatGPT “de-emphasizing” his work and “over-emphasizing” the hundreds of articles to the contrary.
My concern is that we do not yet know if 10% of AI apps or 99% of AI apps will make the same deceitful misrepresentations ChatGPT does. An AI could simply
neglect research coming from Australia, or neglect research coming from physicians who like Dr McBride are
Wipe out humanity? Not necessary. It could leave us with tens of thousands of fish children, trillions of dollars in wasted resources and millions of families who adopt really bad overly-generic budgets and savings plans,
Jesus Christ has saved humanity. Any AGI attempt to destroy humanity will collapse and repentant, God-loving humans who submit to Christ’s rule will be raised back to life on His return and inherit an earth purged of AGI. I’m not at all worried that AGI will destroy humanity.
It’s a theory that I always liked … that humans are a genetic experiment by Aliens, using genes from Earth primates crossed with their own genes. The experiment continued for hundreds of thousands of years until such time (5-6,000 years ago) that the new species … humans … began to become far too violent and thus dangerous to their Alien creators. Faced with that reality, the Aliens elected to withdraw from the planet (which except for the equatorial region was too cold for them anyway) and continue to observe them rather than destroy them. So they took their technology (like their stone cutting tools and heavy lift machines) and left mankind to their violent devices and now send occasional expeditions to Earth to see how we are doing.
And one day, they may decide we are too close to achieving the ability for interspace travel, and they will end the whole thing for the safety of the Galaxy.
It was more like 12,000 years ago, and the global flood that nearly rendered sapiens extinct was their parting gift.
My favorite ever Star Trek episode was The Chase (#146), where everyone was racing to solve the 4 billion year old genetic puzzle that turned out to be from the creators of every race in the galaxy.
It’s definitely something we too would have the sheer ■■■■■■■ hubris to pull on an otherwise peaceful galaxy. lol
If we want to go with wild theory time. Suppose you were in an advanced civilization, say a 2 on the Kardashev scale. You have advanced AI and advanced bioengineering, nano machines etc. Problem, you can either have a dystopia where you monitor everyone at all times for fear they use that capability to wipe out your civilization or cause mass casualties. Wouldn’t the prudent answer be to throw intelligences into a sim, review their ethics when they “die” and only instantiate the ones who have proven not to be a moral threat?
Maybe give them a bible as a hint as to how to succeed. In this allegory, heaven is a 2nd Kardashev life, which may include immortality. I don’t see a reason for hell instead of a wipe for the failed sentients though. I kind of live my life like that’s the way it works, If I am wrong, meh, still a good person.
We may be not that far off from making those choices. The question is, did some prior intelligence put us here for that purpose. Or are we base reality and will be the ones doing it. Math says it’s more likely, we aren’t the base reality. The same way math says any given human is more likely to find himself alive now as opposed to when the population was a fraction of what it is now.
Not many potential terrorists or psychopaths are going to survive a complete review of their life on earth. Maybe to even include everything they ever thought.
Think self replicating nano diamond synthetic biology. No reason in physics you can’t have life with covalent bonds instead of weak van der whal forces. Theoretically you could create an airborne self replicating nano machine that would block out the sun in a day.
The Aliens were helping the ancients build stone structures that cannot be replicated by science, engineers or expert stone craftsmen today, as little as 2,200 years ago.
