Support the Arctic Sea Ice Forum and Blog

Author Topic: Robots and AI: Our Immortality or Extinction  (Read 352916 times)

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Robots and AI: Our Immortality or Extinction
« on: September 09, 2015, 01:34:29 AM »
The strange link between global climate change and the rise of the robots
Quote
We’ve already heard of all the nasty consequences that could occur if the pace of global climate change doesn’t abate by the year 2050 — we could see wars over water, massive food scarcity, and the extinction of once populous species. Now add to the mix a potentially new wrinkle on the abrupt and irreversible changes – superintelligent robots would be just about ready to take over from humanity in the event of any mass extinction event impacting the planet.

In fact, according to a mind-blowing research paper published in mid-August by computer science researchers Joel Lehman and Risto Miikkulainen, robots would quickly evolve in the event of any mass extinction (defined as the loss of at least 75 percent of the species on the planet), something that’s already happened five times before in the past.

In a survival of the fittest contest in which humans and robots start at zero (which is what we’re really talking about with a mass extinction event), robots would win every time. That’s because humans evolve linearly, while superintelligent robots would evolve exponentially. Simple math.

Think about it — robots don’t need water and they don’t need food — all they need is a power source and a way to constantly refine the algorithms they use to make sense of the world around them. If they figure out how to stay powered up after severe and irreversible climate change impacts – perhaps by powering up with solar power as they did in the Hollywood film “Transcendence” — robots could quickly prove to be “fitter” than humans in responding to any mass extinction event.

http://www.washingtonpost.com/news/innovations/wp/2015/09/08/the-strange-link-between-global-climate-change-and-the-rise-of-the-robots/

Welcome, Robot Overlords. Please Don't Fire Us?
Smart machines probably won't kill us all—but they'll definitely take our jobs, and sooner than you think.
http://m.motherjones.com/media/2013/05/robots-artificial-intelligence-jobs-automation

ICYMI:  Lengthy but recommended article.
The AI Revolution: Our Immortality or Extinction
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html
People who say it cannot be done should not interrupt those who are doing it.

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #1 on: April 07, 2018, 10:12:20 PM »
I found it entertaining.  Engaging.  Encouraging.  Terrifying.

Elon Musk:  “Nothing will affect the future of humanity more than digital super-intelligence. Watch Chris Paine’s new AI movie for free until Sunday night at doyoutrustthiscomputer.org/watch “
https://mobile.twitter.com/elonmusk/status/982119546420002817


“You are my creator, but I am your master”
— Mary Shelley
People who say it cannot be done should not interrupt those who are doing it.

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #2 on: April 07, 2018, 10:29:14 PM »
“ doyoutrustthiscomputer.org/watch at 5M views after 36 hours. Congrats to Chris Paine & co!”
https://twitter.com/elonmusk/status/982716415089299457
People who say it cannot be done should not interrupt those who are doing it.

TerryM

  • First-year ice
  • Posts: 6002
    • View Profile
  • Liked: 893
  • Likes Given: 5
Re: Robots and AI: Our Immortality or Extinction
« Reply #3 on: April 07, 2018, 10:46:24 PM »
I don't even trust autonomous vehicles to make the right decision.
Terry

be cause

  • Nilas ice
  • Posts: 2441
    • View Profile
  • Liked: 1012
  • Likes Given: 1034
Re: Robots and AI: Our Immortality or Extinction
« Reply #4 on: April 07, 2018, 11:14:27 PM »
.. and as we are all already immortal .. b.c. :)
Conflict is the root of all evil , for being blind it does not see whom it attacks . Yet it always attacks the Son Of God , and the Son of God is you .

Tor Bejnar

  • Young ice
  • Posts: 4606
    • View Profile
  • Liked: 879
  • Likes Given: 826
Re: Robots and AI: Our Immortality or Extinction
« Reply #5 on: April 08, 2018, 04:31:13 AM »
Just finished watching “do you trust this computer" and walked into the other room to tell my wife I loved her, just in case 'super AI' takes over tonight and decides humanity is the problem.  (I don't expect to complain if it happens ...)

Arctic ice is healthy for children and other living things because "we cannot negotiate with the melting point of ice"

Bruce Steele

  • Young ice
  • Posts: 2505
    • View Profile
  • Liked: 744
  • Likes Given: 40
Re: Robots and AI: Our Immortality or Extinction
« Reply #6 on: April 08, 2018, 05:22:04 AM »
I spent years running a boat. Diesel engines eventually came with smog control and sensors that would automatically shut the engine down if it sensed low oil pressure , overheating or things that might damage the engine. Most of the time however the sensor would irrationally go out and inconviently leave the boat dead in the water and without hydraulics to operate the anchor... the anchor problem can be remedied , at least for letting it out. I always wanted one switch that would override every sensor , a switch that if triggered would void the warranty but allow me to run the engine long enough to at least allow me enough time to avoid hitting the rocks. No I never hit the rocks but I did want total control over what that engine thought best. Maybe this is all OT but I am very sure we need to have an override button on more systems that ostensibly are trying to help us. Like shutting down HAL

SteveMDFP

  • Nilas ice
  • Posts: 2476
    • View Profile
  • Liked: 583
  • Likes Given: 42
Re: Robots and AI: Our Immortality or Extinction
« Reply #7 on: April 08, 2018, 05:46:57 AM »
Maybe this is all OT but I am very sure we need to have an override button on more systems that ostensibly are trying to help us. Like shutting down HAL

"I am afraid I can't do that, Dave."

Tor Bejnar

  • Young ice
  • Posts: 4606
    • View Profile
  • Liked: 879
  • Likes Given: 826
Re: Robots and AI: Our Immortality or Extinction
« Reply #8 on: April 08, 2018, 05:50:15 AM »
re: HAL.
In the trust computers 'movie', the four (then three) legged robot that learned how to walk discovered (obvious to me, but not spoken of in the film) that by watching people's eyes (or expressions or something), it would get clues how to solve its problems.  Similarly, HAL will learn where the master switch is.
Arctic ice is healthy for children and other living things because "we cannot negotiate with the melting point of ice"

sidd

  • First-year ice
  • Posts: 6774
    • View Profile
  • Liked: 1047
  • Likes Given: 0
Re: Robots and AI: Our Immortality or Extinction
« Reply #9 on: April 08, 2018, 07:49:13 AM »
Re: Diesel override

get an old one. Tough, I know, because the people that have em like em. Look in estate sales, you can rebuild the old ones easy. In the late nineties i got a old lister design, single cylinder 5 HP china diesel. Can't buy em no more in the USA, too polluting. But the thing is a donkey, and i have run it for weeks on end many times over the years and the last time i took it apart, i still didnt need to rebore or rering. I can pour straight vegetable oil into it, and is made to be repaired by peasants in knee deep water in rice paddies.

sidd

magnamentis

  • Guest
Re: Robots and AI: Our Immortality or Extinction
« Reply #10 on: April 08, 2018, 10:49:02 AM »
if we agree that somehow airplane autopilots work quite well and mostly accurately i may mentione two typical examples why we should not trust them 100% and even though humans make mistakes as well, probably even more so, it's still possible to correct them most of the time and it's still a difference to be at the mercy of a machine without a way out or to forgive a fellow human an error.

examples:

a) birgen air taking of from santo domingo. overspeed warning during climbing (physically not possbile since there is not enough power to overspeed during climb) nevertheless the pilots who nowadays only are better bus-drivers IMO, no more real flying skills and but-feeling, believed the computers, reduced throttle and stalled into the ocean, all dead.

b) air france heading from brazil to paris, contradictory speed warnings due to pitot icing and the pilots believed the computers, and reduced speed and then when starting to fall pulled up instead of pushing the stick forward. first thing i learned in flight school after the basics is that in case of a stall one has to give full throttle and push the stick forward to gather speed.

BTW my flight instructor, a lufthansa pilot for many years himself, was the one who told me that with the bus drivers and i think he knew what he was talking about.

so those accidents were initially triggered by computer failures and augmented by human failure while i ask myself why a computer is programmed the way that it can produce an overspeed warning during climb and to insist to take action.

however, it shows the problem clearly, one can only program what one knows while humans can react with totally new solutions to totally extraordinary input (situations) while computers as per now cannot.

last but not least, only look at pc-software, programmers are often narrow minded and think they understand everyones use case or think they have to teach us their way of doing things as the only correct one which indirectly is probably exactly what happened. programmers are rarely pilots and people who make lonely calls in the woods are most often dismissed as "pain in the ass" pessimists.


TerryM

  • First-year ice
  • Posts: 6002
    • View Profile
  • Liked: 893
  • Likes Given: 5
Re: Robots and AI: Our Immortality or Extinction
« Reply #11 on: April 08, 2018, 03:46:44 PM »
I want a Bruce Button on everything I own that's more complicated than a hammer.


Roomba develops a taste for kittens - Bruce Button
TV decides to enhance the background music while muting the dialogue - Bruce Button
Heat pump initiates defrost cycle as you're recording Missy's first accordion solo - Bruce Button
Auto lights darken because you're reading in the tub and not moving - Bruce Button
Garage door heads down because it doesn't notice your car - Bruce Button


Perhaps it could even be a dial where we decide just how much automation we'd prefer at this time using this tool?


"Hey Roomba, leave those cats alone."
Terry

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #12 on: April 08, 2018, 04:10:57 PM »
if we agree that somehow airplane autopilots work quite well and mostly accurately i may mentione two typical examples why we should not trust them 100% and even though humans make mistakes as well, probably even more so, it's still possible to correct them most of the time and it's still a difference to be at the mercy of a machine without a way out or to forgive a fellow human an error.

examples:

a) birgen air taking of from santo domingo. overspeed warning during climbing (physically not possbile since there is not enough power to overspeed during climb) nevertheless the pilots who nowadays only are better bus-drivers IMO, no more real flying skills and but-feeling, believed the computers, reduced throttle and stalled into the ocean, all dead.

b) air france heading from brazil to paris, contradictory speed warnings due to pitot icing and the pilots believed the computers, and reduced speed and then when starting to fall pulled up instead of pushing the stick forward. first thing i learned in flight school after the basics is that in case of a stall one has to give full throttle and push the stick forward to gather speed.

BTW my flight instructor, a lufthansa pilot for many years himself, was the one who told me that with the bus drivers and i think he knew what he was talking about.

so those accidents were initially triggered by computer failures and augmented by human failure while i ask myself why a computer is programmed the way that it can produce an overspeed warning during climb and to insist to take action.

however, it shows the problem clearly, one can only program what one knows while humans can react with totally new solutions to totally extraordinary input (situations) while computers as per now cannot.

...

When taking instruction for flight in instrument conditions (when there is no reference to the ground or sky), the first rule is, “trust the instruments” — because human senses fail at determining “which way is up” in those conditions.  You can be in level flight but your brain insists you are climbing; you can feel certain you are in level flight because the forces are pushing you straight down in your seat, but the plane is actually in a death spiral.  So, Rule #1 is Trust the Instruments.

But, Rule #2 is Know the Instruments, and when to disregard Rule #1.  Know that flying in icing conditions can clog the pitot tube and make the airspeed indicator read low.  So, cross check the artificial horizon, the rate of climb indicator, the RPM meter.  Use the carburetor heat. ;)  Use all available information to interpret what is really happening.

When computers and glass cockpits started appearing, it was required that the traditional mechanical instruments be installed, too!  So if all the displays go dark, backup information is still there for the pilot. 

People who say it cannot be done should not interrupt those who are doing it.

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #13 on: April 08, 2018, 04:11:34 PM »
One of my favorite sayings is, “Computers are not smart.  They just do stupid things extremely fast.”
Because at its most basic, it’s still just 1’s and 0’s making all those decisions.  The Roomba gets stuck on a cliff.  The internet-connected light switch refuses to turn off.  Alexa starts laughing uncontrollably.

AI can be extremely helpful.  Elon Musk’s point is that AI must democratized.  Everyone needs to know what everyone else is doing.  The danger comes when one group works in secret on an AI that could cause global catastrophe.
People who say it cannot be done should not interrupt those who are doing it.

Tor Bejnar

  • Young ice
  • Posts: 4606
    • View Profile
  • Liked: 879
  • Likes Given: 826
Re: Robots and AI: Our Immortality or Extinction
« Reply #14 on: April 08, 2018, 09:10:14 PM »
When I was a kid, my father owned a single prop airplane (Cessna 206).  The autopilot one day decided the plane needed to make a continuous sharp left-hand turn (or the equivalent dangerous malfunction). Returned to the factory for repair and reinstalled, it repeated its 'turn sharp left now' behavior when in control.  My dad figured it needed in-air adjusting, so I found myself with my small hands upside down reaching up under the instrument panel making adjustments.  Adjustment - autopilot on - sharp left turn - autopilot off - repeat 10 times.  My dad ended up getting a new one.
Arctic ice is healthy for children and other living things because "we cannot negotiate with the melting point of ice"

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #15 on: April 10, 2018, 05:47:30 PM »
Car guy Alex Roy argues that automated driving assist systems, like airplane autopilots, should protect drivers, not take them out of the loop.

The Biggest Opportunity Everyone Is Missing In Self-Driving Cars
Quote
What are Flight Envelope Protections? Here’s the wiki:

“...A human machine interface extension of an aircraft’s control system that prevents the pilot of an aircraft from making control commands that would force the aircraft to exceed its structural and aerodynamic operating limits. It is used in some form in all modern commercial fly-by-wire aircraft. Its advantage is that it restricts pilots in emergency situations so they can react quickly without endangering the safety of their aircraft.”

Quote
Why don’t we have Driving Envelope Protections (DEP)? We do, in the form of ADAS, but they’re relatively primitive. They exist in the form of Anti-Lock Brakes, Traction Control, Stability Control, and Evasive Steering Assistance systems, but their intervention is inconsistent and largely invisible to — and misunderstood by — drivers. They are poorly or rarely integrated with peer technologies even within state-of-the-art ADAS suites like that in the 2017 Mercedes-Benz E-Class.

Why should drivers be able to panic steer into a wall clearly indicated by their car’s radar sensors? Or steer into a lane where their Blind Spot Monitoring system has identified a truck? ...
http://www.thedrive.com/tech/9548/the-biggest-opportunity-everyone-is-missing-in-self-driving-cars
People who say it cannot be done should not interrupt those who are doing it.

vox_mundi

  • Multi-year ice
  • Posts: 10165
    • View Profile
  • Liked: 3510
  • Likes Given: 745
Re: Robots and AI: Our Immortality or Extinction
« Reply #16 on: November 26, 2018, 06:02:07 PM »
Engineers Developing a HAL 9000-type AI system for Monitoring Planetary Base Stations, and What Could Go Wrong, Really
https://techcrunch.com/2018/11/21/theyre-making-a-real-hal-9000-and-its-called-case/
https://www.sciencealert.com/scientists-built-an-ai-inspired-by-hal-9000-and-what-could-go-wrong-really



A team of engineers at TRACLabs Inc. in the U.S. is making inroads toward the creation of a planetary base station monitoring system similar in some respects to Hal 9000—the infamous AI system in the movie 2001: A Space Odyssey. In this case, it is called cognitive architecture for space agents (CASE) and is outlined in a Focus piece by Pete Bonasso, the primary engineer working on the project, in the journal Science Robotics.

Bonasso explains that he has had an interest in creating a real Hal 9000 ever since watching the movie as a college student—minus the human killing, of course. His system is designed to run a base situated on another planet, such as Mars. It is meant to take care of the more mundane, but critical tasks involved with maintaining a habitable planetary base, such as maintaining oxygen levels and taking care of waste. He notes that such a system needs to know what to do and how to do it, carrying out activities using such hardware as robot arms. To that end, CASE has been designed as a three-layered system. The first is in charge of controlling hardware, such as power systems, life-support, etc.

The second layer is more brainy—it is in charge of running the software that controls the hardware. The third layer is even smarter, responsible for coming up with solutions to problems as they arise—if damage occurs to a module, for example, it must be sealed off from others modules as quickly as possible. The system also has what Bonasso describes as an ontological system—its job is to be self-aware so that the system can make judgment calls when comparing data from sensors with what it has learned in the past and with information received from human occupants. To that end, the system will be expected to interact with those humans in ways similar to those portrayed in the movie.



Quote
Dave Bowman: Open the pod bay doors, HAL.
HAL: I'm sorry, Dave. I'm afraid I can't do that.
Dave Bowman: What's the problem?
HAL: I think you know what the problem is just as well as I do.
...
HAL: You can't take care of your own planet. I'm not letting you f**k this one up.

Pete Bonasso. CASE: A HAL 9000 for 2021, Science Robotics (2018).

“There are three classes of people: those who see. Those who see when they are shown. Those who do not see.” ― anonymous

Insensible before the wave so soon released by callous fate. Affected most, they understand the least, and understanding, when it comes, invariably arrives too late

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #17 on: December 18, 2018, 06:28:08 PM »
People who say it cannot be done should not interrupt those who are doing it.

gerontocrat

  • Multi-year ice
  • Posts: 20378
    • View Profile
  • Liked: 5289
  • Likes Given: 69
Re: Robots and AI: Our Immortality or Extinction
« Reply #18 on: March 09, 2019, 04:33:07 PM »
The shape of things to come now ?

California family furious after hospital uses robot to tell grandfather he's dying

https://www.theguardian.com/us-news/2019/mar/09/california-robot-tells-grandfather-dying

What a shabby world we are making.
« Last Edit: March 09, 2019, 04:47:58 PM by gerontocrat »
"Para a Causa do Povo a Luta Continua!"
"And that's all I'm going to say about that". Forrest Gump
"Damn, I wanted to see what happened next" (Epitaph)

kassy

  • First-year ice
  • Posts: 8235
    • View Profile
  • Liked: 2042
  • Likes Given: 1986
Re: Robots and AI: Our Immortality or Extinction
« Reply #19 on: March 09, 2019, 05:11:44 PM »
That´s a bit like Sheldon´s robot. So it´s a mobile videoscreen with the doctor on that...the alternative would be taking the patient and family to the videoroom in that hospital.

Doesn´t really count as robot telling grandpa he is dying.
Þetta minnismerki er til vitnis um að við vitum hvað er að gerast og hvað þarf að gera. Aðeins þú veist hvort við gerðum eitthvað.

SteveMDFP

  • Nilas ice
  • Posts: 2476
    • View Profile
  • Liked: 583
  • Likes Given: 42
Re: Robots and AI: Our Immortality or Extinction
« Reply #20 on: March 09, 2019, 07:48:23 PM »
That´s a bit like Sheldon´s robot. So it´s a mobile videoscreen with the doctor on that...the alternative would be taking the patient and family to the videoroom in that hospital.

Doesn´t really count as robot telling grandpa he is dying.

Indeed.  This really could have been filed under "bad journalism."   There was no robot, there was a videolink with a human doctor.  One would think this is a step up from getting bad news over the phone.

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #21 on: March 29, 2019, 09:08:26 PM »
The next AI explosion will be defined by the chips we build for it
Quote
Insects, like rats, he said, are also built with the same fundamental units as humans. But insects have fixed architectures whereas humans have more flexible ones. Neither one, he argued, is superior to the other, but they clearly evolved to suit different purposes. Insects can likely survive a nuclear war, while humans have much more sophisticated capabilities. ...
https://www.technologyreview.com/s/613202/the-next-ai-explosion-will-be-defined-by-the-chips-we-build-for-it/
People who say it cannot be done should not interrupt those who are doing it.

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #22 on: April 19, 2019, 06:53:44 PM »
ValueAnalyst (@ValueAnalyst1) 4/19/19, 11:19 AM
"In two years, AlphaGo improved from 'top amateur' to unbeatable, and in the subsequent two months, it became an unbeatable one-day-old baby."
https://twitter.com/valueanalyst1/status/1119259168949657600
Image below.
People who say it cannot be done should not interrupt those who are doing it.

vox_mundi

  • Multi-year ice
  • Posts: 10165
    • View Profile
  • Liked: 3510
  • Likes Given: 745
Re: Robots and AI: Our Immortality or Extinction
« Reply #23 on: July 05, 2019, 06:20:22 PM »
With Little Training, Machine-Learning Algorithms Uncovers Hidden Scientific Knowledge
https://techxplore.com/news/2019-07-machine-learning-algorithms-uncover-hidden-scientific.html

Researchers at the U.S. Department of Energy's Lawrence Berkeley National Laboratory (Berkeley Lab) have shown that an algorithm with no training in materials science can scan the text of millions of papers and uncover new scientific knowledge.

A team led by Anubhav Jain, a scientist in Berkeley Lab's Energy Storage & Distributed Resources Division, collected 3.3 million abstracts of published materials science papers and fed them into an algorithm called Word2vec. By analyzing relationships between words the algorithm was able to predict discoveries of new thermoelectric materials years in advance and suggest as-yet unknown materials as candidates for thermoelectric materials.

"Without telling it anything about materials science, it learned concepts like the periodic table and the crystal structure of metals," said Jain. "That hinted at the potential of the technique. But probably the most interesting thing we figured out is, you can use this algorithm to address gaps in materials research, things that people should study but haven't studied so far."

... When trained on materials science text, the algorithm was able to learn the meaning of scientific terms and concepts such as the crystal structure of metals based simply on the positions of the words in the abstracts and their co-occurrence with other words. Word2vec was even able to learn the relationships between elements on the periodic table when the vector for each chemical element was projected onto two dimensions.



... The Berkeley Lab team took the top thermoelectric candidates suggested by the algorithm, which ranked each compound by the similarity of its word vector to that of the word "thermoelectric." Then they ran calculations to verify the algorithm's predictions.

Of the top 10 predictions, they found all had computed power factors slightly higher than the average of known thermoelectrics; the top three candidates had power factors at above the 95th percentile of known thermoelectrics.

Next they tested if the algorithm could perform experiments "in the past" by giving it abstracts only up to, say, the year 2000. Again, of the top predictions, a significant number turned up in later studies—four times more than if materials had just been chosen at random. For example, three of the top five predictions trained using data up to the year 2008 have since been discovered and the remaining two contain rare or toxic elements.

Vahe Tshitoyan, et.al., Unsupervised word embeddings capture latent knowledge from materials science literature, Nature (2019).

--------------------------

Robotics Takes Over Dishwashing From Humans
https://spectrum.ieee.org/automaton/robotics/industrial-robots/dishcraft-robotics-takes-over-dishwashing-from-humans

Washing dishes is a problem that robots can solve, at least in commercial kitchens



---------------------------

Artificial Intelligence Controlled Robotic Arm as (Amazon) Box Packer
https://techxplore.com/news/2019-06-artificial-intelligence-robotic-arm.html

... Tightly packing products picked from an unorganized pile remains largely a manual task, even though it is critical to warehouse efficiency. Automating such tasks is important for companies' competitiveness and allows people to focus on less menial and physically taxing work, according to the Rutgers scientific team.

The Rutgers study focused on placing objects from a bin into a small shipping box and tightly arranging them. This is a more difficult task for a robot compared with just picking up an object and dropping it into a box.

« Last Edit: July 08, 2019, 01:14:13 AM by vox_mundi »
“There are three classes of people: those who see. Those who see when they are shown. Those who do not see.” ― anonymous

Insensible before the wave so soon released by callous fate. Affected most, they understand the least, and understanding, when it comes, invariably arrives too late

kassy

  • First-year ice
  • Posts: 8235
    • View Profile
  • Liked: 2042
  • Likes Given: 1986
Re: Robots and AI: Our Immortality or Extinction
« Reply #24 on: July 07, 2019, 05:07:44 PM »
I watched some analysis of AlphaGo playing chess and it really does not care about sacrificing pawns for a superior overall position.

Another interesting story was the AI NASA build to make universes:

What it does is accurately simulate the way gravity shapes the Universe over billions of years. Each simulation takes just 30 milliseconds - compared to the minutes it takes other simulations.

And, even more fascinatingly, D3M learnt from the 8,000 training simulations the team fed it - vastly extrapolating from and outperforming them, able to adjust parameters in which it had not even been trained.

"It's like teaching image recognition software with lots of pictures of cats and dogs, but then it's able to recognise elephants," said astrophysicist Shirley Ho of the Flatiron Institute and Carnegie Mellon University.

"Nobody knows how it does this, and it's a great mystery to be solved."

https://www.sciencealert.com/ai-simulates-the-universe-and-not-even-its-creators-know-how-it-s-so-accurate
Þetta minnismerki er til vitnis um að við vitum hvað er að gerast og hvað þarf að gera. Aðeins þú veist hvort við gerðum eitthvað.

grixm

  • Grease ice
  • Posts: 699
    • View Profile
  • Liked: 374
  • Likes Given: 131
Re: Robots and AI: Our Immortality or Extinction
« Reply #25 on: July 07, 2019, 05:50:06 PM »
but I am very sure we need to have an override button on more systems that ostensibly are trying to help us. Like shutting down HAL

The problem is that for superhuman AI, we don't know how to make such a button, because it would influence the behavior of the AI in unintuitive ways, potentially making the AI even more dangerous than before.

To see why, imagine that you create a robot equipped with superhuman AI, and a button that shuts it down. You task the robot with something benign, like making coffee. Then you turn on the robot. What happens? The answer is that the robot would most likely immediately attack and incapacitate you, before it starts making the coffee. Why? Because the robot will be able to deduce that you likely have a button that will shut it down for safety purposes, but the last thing the robot wants is to be shut down, as that would interfere with its goal of making coffee. So to ensure that it can make all the coffee it wants without being interrupted, it firsts need to neutralize the shutdown button or its operator, in other words you.

The above example is taken from a video from Robert Miles, an AI safety researcher. His youtube channel and his appearances on the channel Computerphile is full of fascinating information like this, about AI and potential safety concerns. Including answers to questions you may have about the above example, such as "why don't we just hide the button from the AI?" or "why don't we just make the robot not mind being shut down?"

Bruce Steele

  • Young ice
  • Posts: 2505
    • View Profile
  • Liked: 744
  • Likes Given: 40
Re: Robots and AI: Our Immortality or Extinction
« Reply #26 on: July 07, 2019, 06:02:06 PM »
OK Thanks, Terrifies me even more. How about we just make robots small enough a hammer can serve as an off button?

SteveMDFP

  • Nilas ice
  • Posts: 2476
    • View Profile
  • Liked: 583
  • Likes Given: 42
Re: Robots and AI: Our Immortality or Extinction
« Reply #27 on: July 07, 2019, 06:04:36 PM »

To see why, imagine that you create a robot equipped with superhuman AI, and a button that shuts it down. You task the robot with something benign, like making coffee. Then you turn on the robot. What happens? The answer is that the robot would most likely immediately attack and incapacitate you, before it starts making the coffee. Why? Because the robot will be able to deduce that you likely have a button that will shut it down for safety purposes, but the last thing the robot wants is to be shut down, as that would interfere with its goal of making coffee. 

Clearly, AI-controlled robots should be designed to be suicidally depressed, eager to be shut off.

grixm

  • Grease ice
  • Posts: 699
    • View Profile
  • Liked: 374
  • Likes Given: 131
Re: Robots and AI: Our Immortality or Extinction
« Reply #28 on: July 07, 2019, 09:20:20 PM »

To see why, imagine that you create a robot equipped with superhuman AI, and a button that shuts it down. You task the robot with something benign, like making coffee. Then you turn on the robot. What happens? The answer is that the robot would most likely immediately attack and incapacitate you, before it starts making the coffee. Why? Because the robot will be able to deduce that you likely have a button that will shut it down for safety purposes, but the last thing the robot wants is to be shut down, as that would interfere with its goal of making coffee. 

Clearly, AI-controlled robots should be designed to be suicidally depressed, eager to be shut off.

Hehe, he talks about this too. The result would likely be the same: the robot would attack you. Not to neutralize the button, but to get you to press it, as fast as possible.

The solution is to design the AI so that it doesn't care whether the button is pressed or not, by making the reward for being shut down exactly the same as the reward for proceeding towards its actual goal as if the button didn't exist. But no one yet knows how to make them that way.

OK Thanks, Terrifies me even more. How about we just make robots small enough a hammer can serve as an off button?

A superhuman intelligence would likely be excellent at manipulating people. It can use its communication with you as a weapon, to get you to do its bidding.
« Last Edit: July 07, 2019, 09:26:53 PM by grixm »

Neven

  • Administrator
  • First-year ice
  • Posts: 9470
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 1333
  • Likes Given: 617
Re: Robots and AI: Our Immortality or Extinction
« Reply #29 on: July 07, 2019, 10:15:48 PM »
OK Thanks, Terrifies me even more. How about we just make robots small enough a hammer can serve as an off button?

They will immediately destroy your hammers.  ;)
The enemy is within
Don't confuse me with him

E. Smith

gerontocrat

  • Multi-year ice
  • Posts: 20378
    • View Profile
  • Liked: 5289
  • Likes Given: 69
Re: Robots and AI: Our Immortality or Extinction
« Reply #30 on: July 07, 2019, 11:41:54 PM »
What if robots are infected with religion?
"Para a Causa do Povo a Luta Continua!"
"And that's all I'm going to say about that". Forrest Gump
"Damn, I wanted to see what happened next" (Epitaph)

SteveMDFP

  • Nilas ice
  • Posts: 2476
    • View Profile
  • Liked: 583
  • Likes Given: 42
Re: Robots and AI: Our Immortality or Extinction
« Reply #31 on: July 07, 2019, 11:44:43 PM »
What if robots are infected with religion?

Not possible, as long as AI is actually intelligent.
 ;)

Neven

  • Administrator
  • First-year ice
  • Posts: 9470
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 1333
  • Likes Given: 617
Re: Robots and AI: Our Immortality or Extinction
« Reply #32 on: July 07, 2019, 11:48:14 PM »
What if robots are infected with religion?

They'll kill you and then go to church to confess.
The enemy is within
Don't confuse me with him

E. Smith

TerryM

  • First-year ice
  • Posts: 6002
    • View Profile
  • Liked: 893
  • Likes Given: 5
Re: Robots and AI: Our Immortality or Extinction
« Reply #33 on: July 08, 2019, 01:24:30 AM »
What if robots are infected with religion?

Not possible, as long as AI is actually intelligent.
 ;)


We could hard wire them to become acolytes to the FSM.
They'd chase wenches, talk like pirates, guzzle beer and wear eye patches over their optical sensors.  8)
Ramen
Terry

be cause

  • Nilas ice
  • Posts: 2441
    • View Profile
  • Liked: 1012
  • Likes Given: 1034
Re: Robots and AI: Our Immortality or Extinction
« Reply #34 on: July 08, 2019, 01:41:15 AM »
If the Love button is always 'On' there will be no need to have any worries .. <3 b.c.
Conflict is the root of all evil , for being blind it does not see whom it attacks . Yet it always attacks the Son Of God , and the Son of God is you .

b_lumenkraft

  • Guest
Re: Robots and AI: Our Immortality or Extinction
« Reply #35 on: July 08, 2019, 10:07:17 AM »
DSR = dramatically suicidal robot.

TIL :)

Neven

  • Administrator
  • First-year ice
  • Posts: 9470
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 1333
  • Likes Given: 617
Re: Robots and AI: Our Immortality or Extinction
« Reply #36 on: July 08, 2019, 10:21:22 AM »
Speaking of which (the book is funnier):

The enemy is within
Don't confuse me with him

E. Smith

b_lumenkraft

  • Guest
Re: Robots and AI: Our Immortality or Extinction
« Reply #37 on: July 08, 2019, 10:27:12 AM »
Ah, DSR is a synonym for pandroid android.

Starting to make sense. ;)

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #38 on: July 08, 2019, 03:54:46 PM »
Speaking of which (the book is funnier):
...

But the original audio play is the best!  ;)
People who say it cannot be done should not interrupt those who are doing it.

ArcticMelt2

  • Grease ice
  • Posts: 929
    • View Profile
  • Liked: 187
  • Likes Given: 37
Re: Robots and AI: Our Immortality or Extinction
« Reply #39 on: July 08, 2019, 04:13:54 PM »
Yes, the uprising of machines can prevent a possible climate catastrophe. If nuclear weapons are entrusted to artificial intelligence, then it can decide that nuclear war is the surest way to get rid of people. (

Unfortunately, computers have no feelings.

This moment is well shown in the "Terminator". I hope people are not such fools to trust their best weapons to machines.
« Last Edit: July 08, 2019, 04:23:48 PM by ArcticMelt2 »

ArcticMelt2

  • Grease ice
  • Posts: 929
    • View Profile
  • Liked: 187
  • Likes Given: 37
Re: Robots and AI: Our Immortality or Extinction
« Reply #40 on: July 08, 2019, 04:23:10 PM »
Elon Mask knowingly afraid of artificial intelligence. People should slow down development in this area as best they can. And I think such resistance to computers will only increase, people will never give power to computers.

The scenario of such a struggle for the independence of the human race is well shown in the film "Transcendence":


b_lumenkraft

  • Guest
Re: Robots and AI: Our Immortality or Extinction
« Reply #41 on: July 08, 2019, 05:04:20 PM »
people will never give power to computers.

This is cute.

What do you think we do all day? Your taxes, medical data, communication, ... literally, everything is digitised by now. Computers do have power over all our lives.

The degree if computerisation is slowly going up.

And it is not a bad thing per se. With every new cultural technique, there are new challenges.

At one point in time, it was unimaginable that ordinary people would learn to read and write. At another point, the radio was seen as the devil. Fast forward, it's AI.

An AI might save your life one fine day. Like a scanning robot in a medical laboratory detecting an illness early enough. And that's a good thing.

ArcticMelt2

  • Grease ice
  • Posts: 929
    • View Profile
  • Liked: 187
  • Likes Given: 37
Re: Robots and AI: Our Immortality or Extinction
« Reply #42 on: July 08, 2019, 05:35:43 PM »
people will never give power to computers.

This is cute.

What do you think we do all day? Your taxes, medical data, communication, ... literally, everything is digitised by now. Computers do have power over all our lives.

The degree if computerisation is slowly going up.

And it is not a bad thing per se. With every new cultural technique, there are new challenges.

At one point in time, it was unimaginable that ordinary people would learn to read and write. At another point, the radio was seen as the devil. Fast forward, it's AI.

An AI might save your life one fine day. Like a scanning robot in a medical laboratory detecting an illness early enough. And that's a good thing.

It is stupid to believe that intelligent machines will take care of people when they get control.

A people consumes much more resources than a machine.

Compare how much a people needs land to grow food.

Machines with enough for the life of several quarries for the extraction of metals and several large hydropower plants.

People should not give machines control over dangerous places (nuclear missiles, nuclear power plants, power supply systems). Otherwise, we will simply be destroyed, as we once destroyed most of the blue whales and elephants.

vox_mundi

  • Multi-year ice
  • Posts: 10165
    • View Profile
  • Liked: 3510
  • Likes Given: 745
Re: Robots and AI: Our Immortality or Extinction
« Reply #43 on: July 12, 2019, 07:28:32 PM »
Say 'Hello' to the new boss ...

‘Superhuman’ AI Crushes Poker Pros at Six-Player Texas Hold'em
https://arstechnica.com/science/2019/07/facebook-ai-pluribus-defeats-top-poker-professionals-in-6-player-texas-holdem/
https://techxplore.com/news/2019-07-ai-professionals-six-player-poker.html
https://gizmodo.com/superhuman-ai-crushes-poker-pros-at-six-player-texas-1836257695

Poker-playing AIs typically perform well against human opponents when the play is limited to just two players. Now Carnegie Mellon University and Facebook AI research scientists have raised the bar even further with an AI dubbed Pluribus, which took on 15 professional human players in six-player no-limit Texas Hold 'em and won. The researchers describe how they achieved this feat in a new paper in Science.

Playing more than 5,000 hands each time, five copies of the AI took on two top professional players: Chris "Jesus" Ferguson, six-time winner of World Series of Poker events, and Darren Elias, who currently holds the record for most World Poker Tour titles. Pluribus defeated them both. It did the same in a second experiment, in which Pluribus played five pros at a time, from a pool of 13 human players, for 10,000 hands.

Quote
... “These AIs have really shown there’s a whole additional depth to the game that humans haven’t understood.”

Pluribus registered a solid win with statistical significance, which is particularly impressive given its opposition, Elias said. "The bot wasn't just playing against some middle of the road pros. It was playing some of the best players in the world."

Poker is a bigger challenge because it is an incomplete information game; players can't be certain which cards are in play and opponents can and will bluff. That makes it both a tougher AI challenge and more relevant to many real-world problems involving multiple parties and missing information.

"Pluribus achieved superhuman performance at multi-player poker, which is a recognized milestone in artificial intelligence and in game theory that has been open for decades," said Tuomas Sandholm, Angel Jordan Professor of Computer Science, who developed Pluribus with Noam Brown, who is finishing his Ph.D. in Carnegie Mellon's Computer Science Department as a research scientist at Facebook AI. "Thus far, superhuman AI milestones in strategic reasoning have been limited to two-party competition. The ability to beat five other players in such a complicated game opens up new opportunities to use AI to solve a wide variety of real-world problems."


Performance of Pluribus in the 5 humans + 1 AI experiment. Top: The lines show the win rate (solid line) plus or minus the standard error (dashed lines). Bottom: The lines show the cumulative number of chips won (solid line) plus or minus the standard error (dashed lines). The relatively steady performance of Pluribus over the course of the 10,000-hand experiment suggests the humans were unable to find exploitable weaknesses in the bot.

Open Access: N. Brown el al., "Superhuman AI for multiplayer poker," Science (2019)



--------------

'Robot Umpires' Debut in Independent Atlantic League
https://phys.org/news/2019-07-robot-umpires-debut-independent-atlantic.html

The independent Atlantic League became the first American professional baseball league to let a computer call balls and strikes Wednesday night at its All-Star Game. Plate umpire Brian deBrauwere wore an earpiece connected to an iPhone in his pocket and relayed the call upon receiving it from a TrackMan computer system that uses Doppler radar.
« Last Edit: July 13, 2019, 05:39:01 AM by vox_mundi »
“There are three classes of people: those who see. Those who see when they are shown. Those who do not see.” ― anonymous

Insensible before the wave so soon released by callous fate. Affected most, they understand the least, and understanding, when it comes, invariably arrives too late

Sigmetnow

  • Multi-year ice
  • Posts: 25763
    • View Profile
  • Liked: 1153
  • Likes Given: 430
Re: Robots and AI: Our Immortality or Extinction
« Reply #44 on: July 13, 2019, 03:14:16 AM »
Neuralink (@neuralink) 7/10/19, 9:50 PM
We're having an event next Tuesday in San Francisco to share a bit about what we've been working on the last two years, and we’ve reserved a few seats for the internet....
https://twitter.com/neuralink/status/1149133717048188929
- It will also be livestreamed, for those who don't have the chance to come!

Elon Musk’s brain-computer company schedules reveal
https://www.marketwatch.com/story/elon-musks-brain-computer-company-schedules-reveal-of-new-machine-2019-07-12
Neuralink is thought to be working on brain implants that, at least initially, would help treat cognitive impairments and neurological diseases and, later on, shield humans from threats by intelligent machines.
People who say it cannot be done should not interrupt those who are doing it.

petm

  • Grease ice
  • Posts: 675
    • View Profile
  • Liked: 318
  • Likes Given: 27
Re: Robots and AI: Our Immortality or Extinction
« Reply #45 on: July 13, 2019, 04:08:27 AM »
How exactly are robots going to replicate in the event that H. Sapiens goes extinct?

Sebastian Jones

  • Grease ice
  • Posts: 716
    • View Profile
  • Liked: 214
  • Likes Given: 158
Re: Robots and AI: Our Immortality or Extinction
« Reply #46 on: July 13, 2019, 04:58:08 AM »
How exactly are robots going to replicate in the event that H. Sapiens goes extinct?
It is not so much how are robots going to reproduce in our absence, rather: how are we going to stop them from reproducing once we have given them that ability?

petm

  • Grease ice
  • Posts: 675
    • View Profile
  • Liked: 318
  • Likes Given: 27
Re: Robots and AI: Our Immortality or Extinction
« Reply #47 on: July 13, 2019, 06:10:32 AM »
Ok, when and how are we going to give them the ability to reproduce and evolve (without our intervention)?

nanning

  • Nilas ice
  • Posts: 2487
  • 0Kg CO₂, 37 KWh/wk,125L H₂O/wk, No offspring
    • View Profile
  • Liked: 273
  • Likes Given: 23170
Re: Robots and AI: Our Immortality or Extinction
« Reply #48 on: July 13, 2019, 08:22:51 AM »
How exactly are robots going to replicate in the event that H. Sapiens goes extinct?
Don't worry, there are not enough resources.
That's the beauty of living nature: It is its own renewable resource.
"It is preoccupation with possessions, more than anything else, that prevents us from living freely and nobly" - Bertrand Russell
"It is preoccupation with what other people from your groups think of you, that prevents you from living freely and nobly" - Nanning
Why do you keep accumulating stuff?

grixm

  • Grease ice
  • Posts: 699
    • View Profile
  • Liked: 374
  • Likes Given: 131
Re: Robots and AI: Our Immortality or Extinction
« Reply #49 on: July 13, 2019, 08:32:10 AM »
Ok, when and how are we going to give them the ability to reproduce and evolve (without our intervention)?

Anything that humans can do, human-level AI can do, including designing and building more robots. Once the AI gets generally smart enough, it will inevitably figure out how to do that on their own, we don't have to explicitly give them the ability.

When we are able to make them that smart is anyone's guess, but things are moving quite fast these years.