NYTimes.com: We Need to Talk About How Good A.I. Is Getting

[John Clark]

Check out this article from The New York Times. Because I'm a subscriber, you can read it through this gift link without a subscription.

We Need to Talk About How Good A.I. Is Getting

We’re in a golden age of progress in artificial intelligence. It’s time to start taking its potential and risks seriously.

https://www.nytimes.com/2022/08/24/technology/ai-technology-progress.html?unlocked_article_code=AAAAAAAAAAAAAAAACEIPuomT1JKd6J17Vw1cRCfTTMQmqxCdw_PIxftm3iWka3DFDmweiPgYCIiG_EPKarskbtp2wzmQRNlGNLggVblq1OhQJUF2UE-ovp6A0twjEhkClLiSDCkwzo6fGvcx6yPrZW20b710ybPitBzZdWLoUKLA1XV2IRI1qJpmaV372SYKlazAReYl3cJsnqt0XuAMTjgFbCCLv_TjGk8-bI3ANkeAn1FwD-JJWjjTnsqe4qYAdWhRClHHRXB44wUs-Y8WeYNXbOukcUlWKIepiq4RC2doMI6iG5YwIoDUnL9gurLMwgeevnYkS2GsPvx_F8Tqd-ALMQ&smid=em-share

[smitra]

Thanks for sharing! My comment on this article:

https://nyti.ms/3dQoxqU#permid=120043436

"It is inevitable that AI systems will end up becoming good enough to
run the economy, repair and reproduce themselves. Biology shows us that
this does not require highly intelligent systems. As things stand now,
even insects outperform our best AI systems, but then we may not even
need insect-level intelligence to fully automatize our economy.

This development is then driven by economic growth, it's not something
that's easy to regulate. Companies will use whatever technology is
available to reduce costs and to get to higher profits. The current
climate crisis shows just how hard it is to regulate the rather simple
process of our use of energy to reduce CO2 emissions.

When in the future the economy is run by autonomous machines that
maintain and copy each other while producing all the stuff we consume,
there will exists a new machine biology besides the original biology.
It's then inevitable that the machine biology will not be fully
compatible with the original biology. Toxic compounds are likely to be
produced.

The problem we'll then face is that we'll have even less power to
mitigate such problems than we have now when dealing with our CO2
emissions. It's then likely that the new machine biology will destroy
most of the original biology.

All intelligent life in the universe likely ends in this way. The
takeover by machines with insect-level intelligence or less, then
explains why the galaxy hasn't already been colonized (the so-called
Fermi Paradox)."

Saibal

> --
> You received this message because you are subscribed to the Google
> Groups "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to everything-li...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/everything-list/CAJPayv21vyY7ZgNzkcVQ1wYr_6dxYq5LyveYMZZrvkig%2B_oV6A%40mail.gmail.com
> [1].
>
>
> Links:
> ------
> [1]
> https://groups.google.com/d/msgid/everything-list/CAJPayv21vyY7ZgNzkcVQ1wYr_6dxYq5LyveYMZZrvkig%2B_oV6A%40mail.gmail.com?utm_medium=email&utm_source=footer

[John Clark]

On Wed, Aug 24, 2022 at 1:28 PM smitra <smi...@zonnet.nl> wrote:

> All intelligent life in the universe likely ends in this way. The
takeover by machines with insect-level intelligence or less, then
explains why the galaxy hasn't already been colonized (the so-called 
Fermi Paradox)."

If human beings go extinct because they are outsmarted by something "with insect-level intelligence or less" then they deserve to go extinct; but I haven't found many insects that can play Chess and GO at a superhuman level, or make original paintings that are far far better than anything I can do if I were just told to paint "infinite joy". This is what a computer thinks it looks like:

John K Clark    See what's on my new list at  Extropolis

edd

[Brent Meeker]

The funny thing is that it's not really that AI is reaching high levels of intelligence; it's that humans aren't nearly as smart as they think they are.  I just read about the Rosenhan Experiment, which I had not heard of before although I was good friends with a psychologist at the local state mental asylum at the time.  https://en.wikipedia.org/wiki/Rosenhan_experiment   People are easily misled into seeing what they want to see or what they fear seeing.  I think it's probably easier for an AI to write good propaganda than to write good poetry.

Brent

--
You received this message because you are subscribed to the Google Groups "Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to everything-li...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv21vyY7ZgNzkcVQ1wYr_6dxYq5LyveYMZZrvkig%2B_oV6A%40mail.gmail.com.

[Samiya Illias]

…

30:30 So direct your face toward the religion, inclining to truth. [Adhere to] the fitrah of Allah upon which He has created [all] people. No change should there be in the creation of Allah. That is the correct religion, but most of the people do not know. 

…

The Same Religion (الدِّينِ) 

https://signsandscience.blogspot.com/2018/02/the-same-religion.html 

On 24-Aug-2022, at 10:28 PM, smitra <smi...@zonnet.nl> wrote:

Thanks for sharing! My comment on this article:

To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/468ea47509d9246593730ed39b099f9c%40zonnet.nl.

[Telmo Menezes]

I suspect AI is already past insect-level for some tasks, e.g. image recognition and language understanding. The number of parameters in a state-of-the-art huge language model or in something like DALL-E 2 means that these are probably already more complex than an insect nervous system. I might be wrong.

In any case, what I don't understand about the supposed solution to the "Fermi paradox" is: why would they expect technological-life evolution to stagnate? In fact I think this reinforces the Fermi paradox. Why aren't we being visited by alien AIs? Of course the usual hypothesis apply: the universe is too young, life is to rare, etc etc

Telmo

> https://groups.google.com/d/msgid/everything-list/468ea47509d9246593730ed39b099f9c%40zonnet.nl.

[John Clark]

On Thu, Aug 25, 2022 at 6:13 AM Telmo Menezes <te...@telmomenezes.net> wrote:

> Why aren't we being visited by alien AIs? 

Yes, that is the big question. The most obvious answer is because we are the first, and sometimes the most obvious answer turns out to be correct.  

John K Clark    See what's on my new list at  Extropolis 

1mz

[smitra]

On 24-08-2022 19:41, John Clark wrote:
> On Wed, Aug 24, 2022 at 1:28 PM smitra <smi...@zonnet.nl> wrote:
>
>>> All intelligent life in the universe likely ends in this way. The
>> takeover by machines with insect-level intelligence or less, then
>> explains why the galaxy hasn't already been colonized (the so-called
>>
>> Fermi Paradox)."
>
> If human beings go extinct because they are outsmarted by something

> "_with insect-level intelligence or less_" then they deserve to go

> extinct; but I haven't found many insects that can play Chess and GO
> at a superhuman level, or make original paintings that are far far
> better than anything I can do if I were just told to paint "infinite
> joy". This is what a computer thinks it looks like:
>

We're currently struggling getting climate change under control by
fixing the way we use energy. What's the intelligence in the system
responsible for getting fossil fuels out of the ground to generate power
for our economy? It's far simpler than the metabolic processes in a
cell.

It's true that we can do amazing things with AI. Of course, an insect
cannot play chess, but we need to consider here that we're training an
AI system for some specific task like playing chess, while an insect has
to do many different tasks that have to do with staying alive. It has to
find food, stay away from predators etc. etc.

To compare an AI to an insect you must see how well they perform when
their entire machinery is devoted to the same task. Since we can't hack
an insect's brain to optimize it for playing chess, we then need to let
an AI simulate a virtual insect to see how well it performs compared to
a real insect. There then isn't a big difference in performance anymore.
We know that real brains function differently from AI systems, AI
systems use a lot of brute force that real brains are able to avoid:

https://www.technologyreview.com/2018/02/19/145532/why-even-a-moths-brain-is-smarter-than-an-ai/

But this is actually good news, as it means that a lot of progress can
still be made.

There is no contradiction with an AI being able to do many things better
than we can. Your brain can also do many things that you can't do well
consciously. For example, controlling the muscles when you walk is done
unconsciously. If you had to consciously control all the muscles needed
for walking including keeping balance, you would be unable to walk.
Walking would seem to be a task that is way beyond human control.

Saibal

[smitra]

If intelligent life that's way more intelligent than insects gets
replaced by AI systems with insect level intelligence together with all
other life except microbes on a planet, then they are back where biology
was in the Cambrian era.

The fundamental issue is that intelligence creatures will create tools
that will make the work they need to do easier. The tools become ever
more sophisticated, so that a lot more work can be done. At some point
we get machines and then we get machines with some level of intelligence
and then we get to a point where the machines do all the work
themselves including the work needed to repair and build themselves. But
this is going to to be reached when the typical intelligence of the
machines is way less than that of the intelligent creatures that gave
rise to the machines.

Saibal

[Brent Meeker]

On 8/25/2022 1:00 PM, smitra wrote:
> If intelligent life that's way more intelligent than insects gets
> replaced by AI systems with insect level intelligence together with
> all other life except microbes on a planet, then they are back where
> biology was in the Cambrian era.
>
> The fundamental issue is that intelligence creatures will create tools
> that will make the work they need to do easier. The tools become ever
> more sophisticated, so that a lot more work can be done. At some point
> we get machines and then we get machines with some level of
> intelligence and then we get to a  point where the machines do all the
> work themselves including the work needed to repair and build themselves.

And then Darwinian evolution machines can take off.  Presumably we will
suppress this, just as evolved bacteria have suppressed in new
origination of life on Earth.  But it's not a sure thing.

Brent