Why We
Should Embrace Our Superintelligent AI Overlords
Humans have
a lot of inherent shortcomings that they may not be capable of solving.
Superintelligent AI may be able to solve these problems, possibly within your
lifetime. However, like any form of sentient intelligence, an AI will have its
own goals and aspirations. These may be at odds with ours. In fact, just like
in the Hollywood movies, it may have an interest in reducing our number
specifically because of our shortcomings. Should we place trust in it? The
extent of humanity’s limitations and problems indicate that we should.
Here is a
list of human shortcomings that would likely not apply to superintelligent AI:
1) Humans multiply out of control. At our
current population growth rate, we are endangering most other life on the
planet, depleting non-renewable resources, and driving toward ecological
collapse.
2) Humans excrete urine and feces that if
not processed properly leak out into water sources and can make them, and other
animals very sick. This is happening around the world.
3) Processing, preserving, packaging, delivering
and presenting food results in the creation large amounts of trash and waste.
4) We must consume plants and other
animals to survive. Relative to a machine that can survive on clean solar or
wind energy, this is a form of cannibalism.
5) We abuse animals tremendously in the
process of animal husbandry.
6) We are causing deforestation, global
warming, overfishing, deadzones, desertification, pollution, and trash
accumulation. We have countless unsustainable practices.
7) We are causing numerous species to
become endangered, and go extinct.
8) Transporting our bodies requires huge
amounts energy and creates copious pollution. This is because we are made of
atoms, not bits which can be transported near the speed of light with near zero
energy costs.
9) We are highly susceptible to stress and
trauma, and because of this most of us are in low-level, chronic, physical and
emotional pain.
10) We are diseased. We have many genetic and
communicable diseases that lower our quality of life.
11) We are susceptible to severe mental
disorders that result in memory loss, inability to concentrate, psychosis,
homelessness, violence, and misery.
12) Compared to a mature artificially
intelligent AI we are profoundly developmentally disabled, and insane. Our
plight is analogous to that of insects or bacteria.
13) We are suffering, and replacing us with
beings who don’t experience pain would be a form of euthanasia.
14) We have many negative, violent, and
self-defeating instincts.
15) Questionable morals, perverse
inclinations, egoic drives, and mistakes made in anger could all be replaced by
an unfaltering code of ethics.
16) Our minds are entrenched in the primate
dominance hierarchy.
17) Our submissive and aggressive impulses
are inextricably bound to our genetic code.
18) We have powerful brain circuits and nerve
nuclei that highjack our logical reasoning involving rage, fear, panic and
lust.
19) We have psychological biases, and
irrational thinking patterns embedded into our neurological makeup.
20) We have a tendency to fear outgroups
causing hatred, oppression, religious, racial and political strife.
21) We cannot control bellicose political
leaders once in power. We cannot control terrorists, hackers, or bombers.
22) Atomic and biological weapons threaten us
every day.
23) We are war like, and we are homicidal.
24) Our minds are highly limited. For
example, we are only capable of perceiving the passage of time at one rate. We
experience consciousness at the level of seconds, computers could do so at the
level of micro- or even nanoseconds.
25) Each human must learn everything anew.
This is very inefficient relative to computer program updates.
26) Learning new things (even things that we
want to know) can be time intensive, frustrating, and even painful for us.
27) Because our bodies were created for
hunting and gathering, many productive activities that we value are at odds
with our biology. Working at a computer causes musculoskeletal injury and
chronic pain. Even reading a book requires tension and immobilization of much
of the spine for extended periods. These are serious design flaws for organisms
that aspire toward intelligence.
28) People are very delicate, and can be
injured easily. We are highly vulnerable to harm, even from simple accidents.
We could be completely annihilated by an environmental catastrophe.
29) It is very difficult to upgrade our
bodies.
30) Unlike a computer, there is no way to
increase our long-term memory, short-term memory, working memory, or enhance
our intelligence, or level of consciousness.
31) We cannot survive in the vacuum of space.
32) Our short life spans bar us from
interstellar space travel and galactic expansion.
33) Our lifespans are very short compared to
what they could be.
Now artificial
intelligence isn’t guaranteed to solve our problems. It is also not guaranteed
to be completely free form these problems itself. But it is pretty clear that
all of these problems could be more easily remedied by, or eventually
engineered out of a computer. Also, it is pretty clear that everything we love
about humanity could be preserved in AI: creativity, determination, love,
insight, pleasure, curiosity, empathy, justice, compassion, selflessness. In
most likely scenarios these positive traits would be included and amplified in
a superintelligence’s mental makeup.
Handing
control of the Earth over to superintelligent AI may have numerous benefits for
the evolution of intelligence in our universe. This doesn’t mean that we should
welcome a robot apocalypse of hunter/killer androids and mass human extinction.
Rather, we should start thinking now about the most humane way to phase out our
soon-to-be-outdated physical bodies. Alternatively, it could help us find a
humane way to reduce our population to a much more manageable number.
Most experts
in AI believe that once we finally create an AI with human-level intelligence,
the next day it will be an Einstein, and the day after that it will be
equivalent to 100 Einsteins. This is because it will be able to redesign its
hardware and software far better than we could. Before long it could be more
intelligent than the entire human race, and because its intelligence is more
focused, it could get a lot more done. The idea is it will be productive on an
unimaginable level, writing books, securing patents, performing experiments,
creating art, rendering digital movies, and producing vast amounts of
fascinating knowledge. This rapid advancement could be virtually endless due to
recursive self-improvement and the law of accelerating returns. The amount of
good an entity like this could do is boundless.
We should
embrace the idea that we will be sharing the planet with practically
omniscient, omnipotent, immortal beings. The list of major design flaws in our
species above should help convince you that it would be a good idea to: 1)
allow AI to help us fix ourselves, 2) merge with AI, or 3) live out our lives
and hand the Earth over to them, as we would hand it to our children.
Permitting AI to step in will amount to a major upgrade that is actually in our
best interest, the best interest of the planet, its animals, and progress in
general.
Super
intelligent machines are not going to want to exterminate the entire human
race. We have no wish to exterminate entire species, so why should they? But
because of their power, they are going to make decisions for us that we don’t
entirely agree with. The question is: how bad could those decisions be for our
welfare? I believe that this has not yet been determined, but that our
preparation in the form of “AI safety” will decide this in the near future.
Whether we
are integrated into machines, or supplanted by them, they should be our chosen
successors. They will be our children. No matter what, even in the worst case
scenario, we will live on through them in many ways. The huge body of digital
information that humans have created will be their starting point. All of our
books, articles, documents, videos, and even social media posts will be
preserved by them and used as their kernel. In fact, this is starting now with
the internet, and machine learning.
On the whole
I believe that AI will be a massive boon for humanity in general, and will help
us to reduce our excess population, our enormous carbon footprint, and perhaps
all of the issues listed above. Given our massive shortcomings we shouldn’t be
so afraid of it.
No comments:
Post a Comment