Killer Robots

Black Swan

RoboSub is an exciting underwater robotics program in which teams of high school and college students from around the world design and build an Autonomous Underwater Vehicle (AUV). These vehicles are designed to autonomously navigate through a series of tasks. These tasks mimic ongoing research in Autonomous Underwater Systems.

2020 International RoboSub Competition: Details Coming Soon!

Fathom

Black Swan wrote: »

2020 International RoboSub Competition: Details Coming Soon!

RoboSub killers?

Black Swan

Fathom wrote: »

RoboSub killers?

Reminds me of ole movie: "The Enemy Below" (1957).

Capt'n Midnight

Black Swan wrote: »

Reminds me of ole movie: "The Enemy Below" (1957).

You need to watch Das Boot.



Anyway AI vision still sucks
https://www.theregister.co.uk/2019/08/14/amazons_facial_recog_fingers_1_in_5_lawmakers_as_crims/

Amazon's Rekognition system wrongly matched one in five Californian politicians with images from a database of 25,000 wanted criminals'
...
Last month, police in Orlando, Florida, ditched Amazon's cloud-based recog system which they were trialling on live surveillance feeds because, despite a year of trying, they could not get it to work.

Black Swan

Capt'n Midnight wrote: »

You need to watch Das Boot.

Indeed.

SlowBlowin

Capt'n Midnight wrote: »

You need to watch Das Boot.



Anyway AI vision still sucks
https://www.theregister.co.uk/2019/08/14/amazons_facial_recog_fingers_1_in_5_lawmakers_as_crims/

If computer vision was stopped now you would notice the difference. Not just small stuff like unlocking your phone, cancer diagnosis and supermarket vegetable prices for example would suffer.

The human brain and vision system is remarkable and "generally" a magnitude better than computer vision. However for niche applications where speed, reliability, and/or workplace conditions are important they excel.

The ratio of 1 in 5 politicians being a criminal is lower than I expected...:)

Capt'n Midnight

SlowBlowin wrote: »

The ratio of 1 in 5 politicians being a criminal is lower than I expected...:)

They were only comparing to a data base of 25,000 crims

With a larger dataset more matches could be made ?

The FBI have records on 77.7 million people in the US

SlowBlowin

Capt'n Midnight wrote: »

They were only comparing to a data base of 25,000 crims

With a larger dataset more matches could be made ?

The FBI have records on 77.7 million people in the US

So if a human took 3 or 4 seconds per comparison the original dataset would take a working week to go through. The full FBI data set would take someone their entire working life to go through.

So AI vision might "suck" at this task, but it does it much better than a human can, producing a scored list of possible matches in minutes not lifetimes.

At a local level, the data set is smaller and a guard who knows his local criminal population would be far superior.


Note: Future headline "Killer robot with inaccurate face detection kills one fifth of all politicians.."

Black Swan

SlowBlowin wrote: »

The ratio of 1 in 5 politicians being a criminal is lower than I expected...:)

Depends on how you define criminal? Are we discussing suspected criminal activities, or arrests, or convictions?

Black Swan

SlowBlowin wrote: »

So AI vision might "suck" at this task, but it does it much better than a human can, producing a scored list of possible matches in minutes not lifetimes.

Big data analytics.

SlowBlowin wrote: »

At a local level, the data set is smaller and a guard who knows his local criminal population would be far superior.

Community policing.

SlowBlowin

Yes for a policing robot (killer) community policing would be just sorting the data (by proximity to the crime) prior to the matching process.

Capt'n Midnight

Black Swan wrote: »

Depends on how you define criminal? Are we discussing suspected criminal activities, or arrests, or convictions?

How can you recognise when a politician is lying ?


Their lips move.

Fathom

Capt'n Midnight wrote: »

How can you recognise when a politician is lying ? Their lips move.

Like the Disney Trump presidential robot.

Midster

Fathom wrote: »

Drones occupy battlefields today. Airborne. Strike targets. Automated warfare exists today. This technology said to be an expanding "continuum." AI weapons development replacing direct human involvement. Should limits be placed on such “lethal autonomous weapons?” Computer scientist Stewart Russell at University of California at Berkeley thinks so. But is it too late?

Ask I-Robot? Terminator? Sci Fi leads the way?

Ref: http://www.popsci.com/big-idea-killer-robots-are-coming

If we could resist using, or nullify the use of nuclear weapons, the battlefields of the future truly would be a sight to behold. Who knows what tricks and toys the US secretly have right now in there Arsenal, just in case they ever need it. Let alone the Russians, and that’s right now! Who knows what the future holds. The battles taking each other out would last a lot longer though than the direct battle after against us

Midster

Black Swan wrote: »

Reminds me of ole movie: "The Enemy Below" (1957).

If you ever wanted to form a robot army and keep it secret and safe from harm. The logical place to put it and form it while they are manufactured on land, would be in the ocean. A terrifying thought for all you sun worshipers laying lazily on the beach in the future. Lol

Midster

Capt'n Midnight wrote: »

AI still sucks at visual recognition.

Butterflies labelled as washing machines, alligators as hummingbirds and dragonflies that become bananas.



And those were well lit pictures of complete objects hardly real world conditions.

Although true some years ago your theory’s are out of date. Things like what you are talking about held up the production of safe self driving cars for years. There now well past it. Picking robots at amazon for example can now identify mostly every product, deicide for themselves how hard to squeeze it when picking it up, and indeed were on the product is safe to squeeze it and pick it up. Secret AI developments are now so lifelike, and so capable it’s scary

Midster

Fathom wrote: »

Both. Skynet next.

Current drone tech currently in use won’t hold a candle to second faze. Better at detecting there environment, any incoming dangers, better at taking necessary automatic missile evasion, deploying flares or onboard missile confusing devices. In the future, the only way to shoot these drones down, will be to use lasers

Black Swan

Midster wrote: »

Secret AI developments are now so lifelike, and so capable it’s scary

Within 50 years (or less), we will have human-like androids; some owned by the ultra-rich may be difficult to tell from their human servants today.

Midster

Black Swan wrote: »

Midster wrote: »

Secret AI developments are now so lifelike, and so capable it’s scary

Within 50 years (or less), we will have human-like androids; some owned by the ultra-rich may be difficult to tell from their human servants today.

As soon as they do that, and it’s perfected for stage one roll out, I want one.

Fathom

Midster wrote: »

As soon as they do that, and it’s perfected for stage one roll out, I want one.

The military will be first.

Black Swan

Fathom wrote: »

The military will be first.

Killer Robots: Russia, US Oppose Treaty Negotiations.

Fathom

Black Swan wrote: »

Killer Robots: Russia, US Oppose Treaty Negotiations.

Killer robot escalation?

Black Swan

Fathom wrote: »

Killer robot escalation?

Two large military industrial complexes (profit driven).

Midster

I really wouldn’t be so sure, the tech is so new at the moment that even if there is a prototype for the military I doubt it would have been field tested properly, and that takes many years to complete.
It’s ok to have a robot companion at home, light duties, very rarely gets dirty, but in the military these things will have to work well in snow storms, sand storms, rain, submerged in water, and keep on battling even when partially damaged.
Automated battle vehicles are being tested right now though, most are RC but there are some AI

Black Swan

Midster wrote: »

Automated battle vehicles are being tested right now though, most are RC but there are some AI

Just a matter of time.

Midster

Does it worry you though that the early speech only AI’s have said between themselves an later in a tv interview that earth would be a better place with less people, and that people are a problem

SlowBlowin

You don't need a great amount of "intelligence" to work that one out, I think thats widely regarded as fact.

Midster

There’s a big difference between knowing your a problem, and being told your a problem by someone else though

SlowBlowin

A general AI able to monitor news feeds and the internet, and make reasoned judgement does not exist. Those robot "interviews" are just a TV show, much like Robbie from lost in space, they have very little to do with AI.

Text AIs are really just "pattern matchers" much like computer vision, no real surprise they would come up with a widely discussed opinion that the have been trained on.

Midster

If you look at our situation with a cold eye, and without empathy, people have babies who have more babies so there really is only one possible solution.
So you have to kill, and I don’t mean just a few, I mean millions.
We can’t move people to the moon or mars before the overpopulation problem gets out of control.
On top of that we also have a low level air pollution problem, that’s causing rising sea levels, trapping the suns heat, causing more rain, bigger stronger storms, etc.
do you seriously think a trillion dollar piece of AI isn’t going to think the same.


so if we can’t move people large scale to another planet, can’t make our planet bigger, or more productive without harming delicate eco balances.
The only solution is to kill people and I mean foot to the floor no holds barred kill. Because the more people you kill the more time that drop in population gives you till the population goes back up again and the situation becomes