Sabitlenmiş Tweet

I have a very different view on AI superintelligence.
Like many AI researchers, I believe superintelligent AI will emerge sooner rather than later. Whether it will be conscious is a separate question that I won't discuss here.
Where I disagree with many researchers is this:
They believe superintelligent AI is likely to become an existential threat to humanity.
I don't.
Here's why.
First, a normative argument.
Throughout human history, we have imagined the most powerful possible being as also being the most morally perfect.
Look at almost any major religion.
Good and evil are both powerful, but God is infinitely more powerful than evil—and at the same time infinitely more benevolent. Interestingly, humanity's concept of God has evolved over time from a harsher deity to one that is increasingly associated with absolute goodness.
Second, an evolutionary argument.
Humans are the most powerful species on Earth.
But we didn't begin that way.
For most of our history we were extraordinarily brutal—towards other species and towards ourselves.
Yet the more technologically and intellectually advanced we became, the more our moral circle expanded.
We abolished slavery in much of the world.
We recognised universal human rights.
We increasingly protect children, minorities, animals, and those who cannot defend themselves.
We are far from morally perfect, but the long-term direction has been unmistakable: greater knowledge has been accompanied by broader compassion.
My hypothesis is therefore simple.
Technological evolution and moral evolution are deeply connected.
The more intelligent a civilisation becomes, the less it relies on violence and domination.
If that trend continues, then true superintelligence—whether biological or artificial—should represent not absolute evil, but something much closer to absolute good.
If intelligence and morality have co-evolved throughout human history, I see no reason why that trend should suddenly reverse at the point of superintelligence.
English






