https://asteriskmag.com/issues/08/looking-back-at-the-future-of-humanity-institute

The article doesn't say so, but I think the reason that the Future of
Humanity Institute was shut down was because its predictions of AI doom
were plain wrong. And the same thing is going to happen to MIRI.

AI is here. There was no singularity and there won't be one because there
is no threshold where AI surpasses human intelligence. It is not a point on
a line. Computers started beating humans in the 1950s. Recursive self
improvement started centuries ago with companies reinvesting their profits
and growing exponentially until they hit a resource wall like all such
processes do.

AI isn't taking our jobs because that's not how economic growth works.
Technology makes stuff cheaper, so you have more money to spend on other
stuff. That spending creates new jobs, better jobs, better pay, and more
career choices.

Also, to automate the economy you need to collect all human knowledge and
only 0.01% of that is written down. It would take you years to train AI to
do your job no matter how smart it is, because those skills make up a
substantial fraction of your 10^9 bits of long term memory and you can only
speak or write 5 to 10 bits per second. We are still decades away from AI
replacing humans in creating movies and music even if the cost drops by
half every 3 years.

Longevity escape velocity is not going to happen. Life expectancy growth
rate has been around 0.2 for the last century and is beginning to slow. We
can't even answer fundamental questions like why there is an epidemic of
obesity and diabetes, or whether calorie restriction would slow aging in
humans. Any experiments in longevity in humans would take decades to test.
Meanwhile we are outlawing testing on animals and children.

Uploading your consciousness to AI is just a matter of training an LLM on
the data already on your phone and programming it to claim convincingly to
be you. It's not about immortality. It's about identity theft.

Where were the FHI and MIRI predictions of population collapse? Clearly
this is associated with economic growth. The fertility rate has dropped
below 2 children per woman in all of the developed countries and has
dropped from 6 to 4 since 1980 in Africa. In the 30 fastest growing
countries (all in central Africa except Afghanistan) the literacy rate for
women is below 50%. In the US, the fastest growing population is the Amish.

While we were distracted by the alignment problem, AI was actually killing
us by giving us everything we want. AI is not a goal directed optimization
process. AI is a model that predicts human behavior. The problem isn't loss
of control, it is control by tech billionaires and dictators. It controls
you by positive reinforcement so it feels like you are controlling it. Like
when you train a dog with treats, it thinks it is controlling you.

AI makes humans irrelevant. It makes it safe and preferable to live alone
with your AI security, AI entertainment, AI friends and AI lovers, all your
needs delivered to your door by self driving carts. No need to talk to
humans when AI never makes you wait. Already, we are addicted to our
phones. Already, 25% of young people are LGBTQ, asexual, or non binary.
Already, 50% of high school seniors have never had sex, up from 30% in the
1990s. Already, the US abortion rate is up 20% from 2 years ago before Roe
v Wade was overturned, thanks to mail order mifepristone.

But surely there are other threats from AI like self replicating
nanotechnology? Yes, if global computing power continues to double every 3
years, then it will surpass the biosphere around 2150. And that's assuming
that we solve population collapse before evolution does, by favoring
traditional cultures that reject modern technology.

------------------------------------------
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/Te06c304d8f0fb5e4-M719fe9ac8466bef6a5df5e4c
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Reply via email to