Threat Analysis - AI, Another Threat Item To Be Tracked?

[TWP]

Since I last updated my threat analysis spreadsheet, I found this article about possible Artificial Intelligence(s) (AI) being placed in control of our weapons systems.

Since this is (IMO) well within the realm of possibility, I've added a new threat to my spreadsheet:  "AI Control".

At this point, I rate the threat level as low, 1 on a scale of 10.  Our personal (family) preparedness level is also low, 1 on a scale of 10.

[edit] The reason our family preparedness level is so low is because we use the Internet, a lot, and there are no real protections (IMO) available now.  If the 'net goes down, there is not much we, individually, can do.  Of course we can try to protect our own computers, but there is an unknown risk for which we don't yet have any remedies in place.

This needs more research and, as always, continual monitoring of science news and research.

For an introduction to the problem, I'll link to this article.  It is not intended as a "scare" tactic, but rather to show where we are, as a world society, at this moment in time.

http://bgr.com/2017/11/08/ai-weapons-systems-military-destruction-apocalypse/amp/

You can expect to see more reports and, of course, media spin on this topic.  We, all of us, will need to apply some rational logic to what we are going to be reading, ie. Who published it, What are their motives, How reliable are they as a source, When will something take place, etc.  Where is literally anywhere in the world, but particularly where there is computer access and control.

I'll be honest and say that I have my own opinions on AI.  You probably read about my opinion here...

[Jerry D Young]

There are multiple and intertwined risks and outright dangers to Artificial Intelligence. Some are not at all likely, though. But some of the others, more for humans trusting the programming that created the AI, are very real, and if they were to occur, could be very widespread and rather devastating.

My story in which several super computers are linked together, with three of them having AI programming left in them that was not supposed to be there, is one of the more far fetched ones. But when humans are taken out of the loop in decision making processes that can result in tremendous negative impacts, the AI becomes dangerous.

Just my opinion.

[WolfBrother]

Isaac Asimov explored the issues of unfettered AI in his Robot series of books.

He came up with the 3 laws of Robotics which kept robots "fettered".

He explored a number of what if's in several stories.  One in particular dealt with the modification of one of the laws because robots were exposing themselves to a radiation that fried them at a level much lower than could cause harm to humans. 

Until I retired, I was a Senior Oracle Database Administrator.  Within the internals of the database management system I could "twiddle bits".  Very hard core low level access.  I came up always in control of everything that way. 

I find that the more something does for me the more I worry about what else it does without my knowledge and more importantly my permission.  Backdoors?  Sending info to vendors? Letting SkyNet know where I am?

So, unfettered AI is scary as heck to me.

Read the 1966 science fiction novel Colossus, by Dennis Feltham Jones (as D. F. Jones) , or watch the movie Collossus: The Forbin Project for an example of unfettered AI.
The Wiki on it is here - https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project

[TWP]

Jerry and WolfBrother;

Thanks for your responses.

I'd like to develop some kind of prepper plans with respect to AI, but I'm still researching what is available in public reports and papers.  Frankly, I would not trust ANY of the Main Stream Media (MSM) "news" on this subject because it is far to easy to manipulate (money talks loudly).

Do any of you have any thoughts on how to go about measuring this risk factor?

I've noticed, and posted here about, the problem of "personal digital assistants" in the home and office.  This is one factor in the risk analysis for developing AI.  They represent a direct, two way interface to the global internet (the WEB) which is NOT in your control.  The Luddite in me feels fear about this trend.

What else can we look for?

[Jerry D Young]

A 2008 publication by the organization Future Of Humanity Institute gave Super Intelligent AI as a 5% probability of causing human extinction by 2100. Molecular nano-technology weapons were also given a 5% probability. They were the two highest percentage probabilities in a 19% overall probability of human extinction by 2100. Nuclear war was only listed as a 1% probability.

Here is a list of some of the things I am planning to help protect myself from some forms of AI use. Please take them with a grain of salt, but do not discount all of them out of hand.

AI Defenses

1.   Stock up on theatrical makeup. May need to fool facial recognition software.
2.   Have some alternative theatrical face masks made for the same reason.
3.   Have some powerful 1,000+ lumen flashlights to blind sensors.
4.   Ditto high power lasers and audio generators.
5.   Get a few large dome type umbrellas and line them inside with space blanket material to hide and mask your face to make identification more difficult.
6.   Have several more space blankets cut in various shapes for decoys.
7.   Have human shaped manikins with surface material that can be heated to look like a human form for diversions.
8.   Get a .338 Lapua or .408 Cheytac rifle for anti-drone/anti-AI device destruction.
9.   Keep up with the designs of new military and police drones and mobile AI devices so you can figure out vulnerabilities.
10.   Invest in a lot of helium capable balloons, helium, and light bank line to put up 'barrage balloons' to keep flying AIs/drones from close range observation distances.
11.   Keep self-activating ice packs handy to break up the heat signature when out and about and under surveillance.
12.   Have heat and cooling sources in the home to emulate humans, and to mask humans from AI controlled surveillance.
13.   Monitor as many forms of local communication as you can to get reports of actions taking place around the area so you have time to prepare to hide or evade.
14.   Be prepared to alter your movement patterns to change the way you walk and move.
15.   Get some walking aids, especially walking sticks, to help in the above, and as role camouflage.
16.   Read John T. Molloy's Dress For Success books to learn about how people perceive people, which are some of the sames thing that AI surveillance considers when tracking people.
17.   Learn your AO so you will know useful places to hide from AI based surveillance, places to ambush drones, and AI machines.
18.   Create alternative personalities using some of the above techniques, with good locations to change from one to another.
19.   Get the best observation and monitoring equipment you can to test your defensive actions as well as you can. They will not be as good as what the owners of AI have, but will give you some idea if you are wasting your time.
20.   Practice with a voice changer now so you can use it in the future when it might be needed.
21.   Practice using cyphers and codes to communicate, in case you need to keep information private in the future using alternative methods as well as current standard methods.
22.   Do not use ‘standard’ methods of encoding, such as using the Bible or other common books as reference works to develop cyphers and codes. The computers can analyze books like that in seconds to minutes.
23.   Always remember that you leave behind a clear trail visible to one or more of their sophisticated sensors, so you have to really think outside the box to defeat them, or overcome the results of them locating you.
24.   Remember that most (not all) visual and audio sensors can be blinded or masked. Even some higher tech ones can be as well. So metallic paints, high power sound systems and foam sprays, dry powders (flour, barite, bentonite), oily sprays, and such can be useful armaments against drones and some AI devices.
25.   Remember π. If you can get a computer system to start calculating π to a finite decimal point, that should either lock it up or keep it busy for a very long time. Or the old computer BASIC language routine 10 CLS  20 GOTO 10 which simply keeps looping through, keeping the screen clear until interrupted, which is easier to stop than the π routine. (Just kidding. Sort of.)
26.   Fog machines, water misters, and some other theatrical and agricultural devices can help deter some sensors and AI devices, but usually only partially.
27.   Though some can fly, very few of the drones and AI units can swim. Water can be your friend. Keep that in mind.
28.   If you can get a device trapped, burning bars can quickly disable them, even if lightly armored.
29.   Try to have several different decoy systems available, since any one device may have multiple sensor types and you may have to lay many different trails to confuse them.
30.   Consider the mobility method of the device, if it is mobile, and determine a way to stop its movement.
31.   Consider the type of fuel it uses and determine a way to make it lose fuel, use a great deal more than usual, or contaminate it. Ditto the type of engine and traction if applicable.
32.   Prepare as many spider holes as possible that will counter known sensors. Have at least some that have escape routes through which you can get outside the range of the sensors long enough to change the look of whatever sensors are being used is monitoring.
33.   Radio jammers that would be effective on police/military drone and mobile AI equipment are hard to come by and of limited range. But if you know someone that can build one, that is good. But you will only get one use out of it, hopefully that will be enough to let you make a permanent getaway.
34.   As ‘smart’ as AI can be, it currently and into the near future, is still stupid compared to humans and cannot reason beyond a certain set of parameters programmed into it. If you can do the really unexpected, you have a better chance of evading them, if they are one of the predictive activity types. Basic tracking devices are more difficult to evade.
35.   Always have a heavy hammer handy.

Just my opinion.

[TWP]

Jerry, I see that you are several steps ahead of me on this:  Ok, several miles ahead...

RE #28;  What is a "burning bar"?

I found this comic which points at one possible direction of AI: Assimilation...

https://www.smbc-comics.com/comic/specialization

[WolfBrother]

22.   Do not use ‘standard’ methods of encoding, such as using the Bible or other common books as reference works to develop cyphers and codes. The computers can analyze books like that in seconds to minutes.

Jerry,
Yes you're correct if your using page line word coding

0850511
xlates to P85 L5 W11

You just have to have identical copies of the books (down to publisher and date published)

Same same if you code a sentence using a line from a book:
Book - Now is the time for all good men to come to the aid of their country
MSG  - Bugging Out Now

N is the 14th letter in the alphabet
B is the 2n letter
result is 16 or P

O is the 15th letter
U is the 21st letter
result is 36 or 10 (35 - 26) or J

First two entries in the code would be:
1636
or
1610
or
PJ

Remember this is a simple code and it can be brute force broken fairly quickly. 

Now if you encrypt the numbers with a line from a one time pad it makes it fairly hard to break
1636

first 4 numbers from onetime pad page numbered 4899321 are:
4938 is the encoded result

add them without carrying:
1636
4938
------
5564   is the encrypted result

send it (page number is the first sent and it's plain text i.e.

receiver writes the numbers out and finds sheet 4899321
he then writes out the numbers from the sheet
and subtracts without borrowing

5564
4938
------
1636 is the decrypted result

Reverse the first steps and you get
BU the decoded message.

Note - never reuse page 4899321 or you render it VERY, VERY,m VERY, VERY easy to break.


So you can use standard works to encode with - you just have to encrypt them also.

IF you're careful and do not follow rules of capitalization you can have 52 possible encoded values for a letter.
IF you're very careful and don't mistake the letter l for the number 1 and the capital O for zero 0 you can have 62 possible combinations to encode a letter or a number.

THEN you can have the same number to encrypt with giving 62 x 62 = 3864 possible combinations for a single letter.

the word bugging is 7 characters.  To brute force it would take a maximum of  3864x3864x3864x3864x3864x3864x3864 possible combinations to ensure you got every letter.

In reality, once you get buggixx a smart computer would probably go to the next.

A properly encoded, encrypted message (done as above) is extremely difficult to break.

A page line number code matrix that changes daily is good for radio.

SOI would state the unique page id for the day.

Then all messages would be page/line/column

example

page 05 line 16  column 08 could read stragglers
page 06 line 08 column 18 could read 15
page 01 line 07 column 05 could read coming.

a bit cumbersome but your radio message would be
051508break
060818break
010705end

Your SOI can get very complex but it can be secure.


 

[TWP]

WolfBrother;

I've worked with these cyphers before, as have some other members.

Do you have a good reference document on the use of these?  It would help to have this in our individual libraries for practice.

Jerry;  Do you have some references?

[WolfBrother]

i have a one page line number/column code sheet that was used by a specific tank company in GWI and II.  Sometimes the code just has to secure enough for you to take/not take action.

I have put it in an Excel spreadsheet to be able generate unique sheet id and new line and column headers.

same same a pseudo-random onetime pad generator


Most of my knowledge is from grunt level from when I was a Medic and a Ranger.

I have read Bruce Schneier's book Applied Cryptography about 5 times.  I understand maybe 10%.
I've read explanations of how the Enigma machine works and why it was breakable.

A good fiction to read is Neal Stephenson's Cryptonomicon.  It is a fictionalized history/fiction that starts pre-WWII and ends in modern day.  You will pick up a bunch about cryptography in it.  The WWII part is full of good stuff.

[TWP]

Coming back to the original theme of this thread:  Artificial Intelligence (AI) as a prepper threat factor.

At this point in time, we are just beginning to see the use of AI.  Mostly we find "super" computers which are being "educated" or "trained" to understand human language and to interpret images.  Another use is in the autonomous driving vehicles that are now on the road.  I've watched a few videos which purport to show an AI running a humanoid robotic body.  I"m not sure how real that was... It could have been an attempt to get funding for more AI research.

While I could extrapolate these into some scary scenarios, a la "Terminator" the movie and numerous sci-fi books, I'd rather address what can be done to plan for this as an eventuality (future way-of-life).

Predicated on the internet and computers being available, we can expect to see progressively more autonomous operation in our electronic systems.   We already have the "personal digital assistants" which, while not fully AI, are certainly a functional part of such an entity.  We have lots of pieces, not fully connect to each other, yet...

1)  I question the wisdom of allowing such devices to become a necessary part of our lives.  I advise planning on how NOT to use such devices, even when they are readily available.

2)  There is a trend toward the use of cameras in both public and private places which have advanced facial recognition built in.  If you want to move around in public spaces, these are becoming hard to avoid.  Think traffic  cameras, security cameras, personal computer cameras... You get the picture? pun intended.

3)  In personal homes, the "personal digital assistant" is already monitoring all speech and some of this is being transmitted outside of the home space and stored in databases.  This leads to a problem if you visit someones home and fail to inquire about the use of such devices.   Remember; "loose lips sink ships"...

4)  We must take it as a "given" that our online activity is being tracked, monitored and recorded.  Even if you take precautions to use various methods to stay anonymous while online (VPN, TOR, encryption, etc.),  the entire internet is inherently insecure and mostly accessible to the random, curious AI which might be cruising the 'net at any given time.

The above are only part of how we are exposed to the rest of the world.

Of course, the meme of going "hermit" and retreating into the woods, severing all or most contact with society, remains an option.  Most of us don't actually have the freedom to go that direction.

So, given that most of us live inside human society,  how do we go about protecting our security and privacy in an world which is moving to AI integration (assimilation  ) ?

Yes, I get that this is mostly hypothetical right now.  It is a specific scenario which may never come to pass...  I'd rather have plans in place to deal with it, than be caught un-prepared...

So; Plans?

[WolfBrother]

Yes, I get that this is mostly hypothetical right now.  It is a specific scenario which may never come to pass...  I'd rather have plans in place to deal with it, than be caught un-prepared...

I agree its mostly hypothetical right now.  I suspect it will get worse.
I agree we need plans to deal with it.

Apologize for allowing my inner nerd to take over and hijack the thread. 

[TWP]

...
Apologize for allowing my inner nerd to take over and hijack the thread.

Well, really now, I'VE never done that before...

The internet gestapo will be knocking on your door shortly? 

That is why we are still free, mostly, kinda, sorta...

So what plans can be made for this hypothetical AI threat?

[Jerry D Young]

WolfBrother, every step helps make it harder for computerized systems to break codes. I was aware of the cascading of methods, but I still believe it better to not use any of the most commonly used books for any step of the procedure. I am sure that TPTB already have pretty much all versions of the bible and other religious texts (US Constitution, Declaration of Independence, and similar documents that have meaning to the people that use them) in their code breaking systems, as well as many other texts that have commonly been used as a source of page/line/word coding, and other methods of 'same book' coding.

They can run through all of those extremely quickly, applying many other coding systems in conjunction, in both before and after use techniques. I prefer to use a book or document that is not particularly common, but is available in same edition copies. And not things, like the 'Freedom Documents' listed above, that the logic systems can predict that a given person might use. For me, I will never use any of my own works, any other Prep/PAW fiction, any of the science fiction I am prone to read such as Robert A. Heinlein, and many other types of works that even a simple investigation would find.

Not about to say what I will/would use, but it will not be anything that will be easy to associate with my thought processes.

TWP, I have WolfBrother's spreadsheet, but I will let him distribute it as he wishes. Actually, I think I got it here.

My experience with cyphers and codes is a bit limited. WolfBrother is a much better source than I. Most of the coding systems I have used have been very old hat, but using very modern versions or twists on the originals. Some date back to ancient Greece, Rome, Egypt, and the far east civilizations. And while the following might give a hint, I do not think it is a big risk for me to say that I make up my own systems, not based on any other system (Of which I am aware. I could be reinventing things at times.), that use a different logic pattern than I am known for, or any of the most common logic systems in use.

Burning bars are also known as thermal lances. They are an exothermic cutting and breaching tool used in mining, demolition, and several other businesses and industries. No links, so you can use your own search methods, but look for burning bars, thermal lances, Oxylance, Caldo burning bars, Thermolance, Broko-Rankin.

Well, as stated in my earlier post, there are certain things that can be done about many of the aspects of AI being incorporated into several systems that could affect us in the future. Part of my plan is to acquire the items listed, practice the techniques listed, and extremely limit my exposure to many of the devices that are becoming so common.

As to going hermit, I do not think this is a viable option for many. However, a person (not me, of course) could make certain preparations now, for the time when it will not only be a viable option, but a very desirable one. I do not want to post the details here, but there are quite a few things that can be done to have things ready when the time is right to fall into the ground clutter of the radar, as long as 'they' are not actively looking for you from the start of the event.

If you can clarify just a bit, TWP, on what types of preps you are talking about, I might be able to explain some of my ideas a bit better.

Just my opinion.

[WolfBrother]

A long winded demonstration of Fiction AI and how AI is being presented now:

 
The following is primarily made up with cut/past from this wiki https://en.wikipedia.org/wiki/Three_Laws_of_Robotics .
 
The rest are my comments.
 
 
The Three Laws, quoted as being from the "Handbook of Robotics, 56th Edition, 2058 A.D.", are:
1.  A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2.  A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3.  A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.^[1]
 
In a much later story:
Asimov also added a fourth, or zeroth law, to precede the others:
0.  A robot may not harm humanity, or, by inaction, allow humanity to come to harm.
 
Many of Asimov's robot-focused stories involve robots behaving in unusual and counter-intuitive ways as an unintended consequence of how the robot applies the Three Laws to the situation in which it finds itself.
 
In an earlier post I mentioned this story briefly:
First Law modified
In "Little Lost Robot" several NS-2, or "Nestor", robots are created with only part of the First Law.^[1] It reads:
1. A robot may not harm a human being.
This modification is motivated by a practical difficulty as robots have to work alongside human beings who are exposed to low doses of radiation. Because their positronic brains are highly sensitive to gamma rays the robots are rendered inoperable by doses reasonably safe for humans. The robots are being destroyed attempting to rescue the humans who are in no actual danger but "might forget to leave" the irradiated area within the exposure time limit. Removing the First Law's "inaction" clause solves this problem but creates the possibility of an even greater one: a robot could initiate an action that would harm a human (dropping a heavy weight and failing to catch it is the example given in the text), knowing that it was capable of preventing the harm and then decide not to do so.^[1]

The problem with that, is mentioned in that story.  Robots, with their “Positronic” brains KNOW they have superior strength and intellect and therefore are essentially superior to humans.  The 3 laws keep them in control, “fettered” as I noted above.  The modification of the first law changed the dynamic.

The Will Smith move I Robot was a horrible mishmash of Asimov’s works.  The dual purpose robots were Nestor series robots.  You saw how well that turned out.

In 2001: A Space Odyssey the HAL 2000 series computer went crazy when told to keep a secret. 
Trivia – H. A. L. are the letters immediately before I. B. M. in the alphabet.  In 1998, we retired our IBM 9000 series computer because it was outdated.  It didn’t make it to 2001.

Ambiguities resulting from lack of definition
The Laws of Robotics presume that the terms "human being" and "robot" are understood and well defined. In some stories this presumption is overturned.
 Definition of "human being" The Solarians create robots with the Three Laws but with a warped meaning of "human". Solarian robots are told that only people speaking with a Solarian accent are human. This enables their robots to have no ethical dilemma in harming non-Solarian human beings (and are specifically programmed to do so). By the time period of Foundation and Earth it is revealed that the Solarians have genetically modified themselves into a distinct species from humanity — becoming hermaphroditic^[36] and telekinetic and containing biological organs capable of individually powering and controlling whole complexes of robots. The robots of Solaria thus respected the Three Laws only with regard to the "humans" of Solaria. It is unclear whether all the robots had such definitions, since only the overseer and guardian robots were shown explicitly to have them. In "Robots and Empire", the lower class robots were instructed by their overseer about whether certain creatures are human or not.

Definition of "robot"
As noted in "The Fifth Law of Robotics" by Nikola Kesarovski, "A robot must know it is a robot": it is presumed that a robot has a definition of the term or a means to apply it to its own actions. Nikola Kesarovski played with this idea in writing about a robot that could kill a human being because it did not understand that it was a robot, and therefore did not apply the Laws of Robotics to its actions.

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

All of this points out how “unfettered” AI can and will cause problems.  Ford has an AI parallel parking feature.  The AI feature decides how to park, turning the wheel, applying the brake, applying accelerator etc.  What do you do when you’re going down the Interstate at 70mph and an errant electronic signal makes the car decide to take over and parallel park right then?  How do you override? 
Amazon has an AI system made up of Echo smart speakers and Alexa.

A smart speaker is a type of wireless speaker and voice command device with an integrated virtual assistant (artificial intelligence) that offers interactive actions and handsfree activation with the help of one "hot word" (or several "hot words"). Some smart speakers can also act as a smart device that utilizes Wi-Fi, Bluetooth and other wireless protocol standards to extend usage beyond audio playback, such as to control home automation devices. This can include, but is not be limited to, features such as compatibility across a number of services and platforms, peer-to-peer connection through mesh networking, intelligent personal assistants, and others. Each can have its own designated interface and features in-house, usually launched or controlled via application or home automation software. ^[1]
https://en.wikipedia.org/wiki/Smart_speaker
 
The devices connect to the voice-controlled intelligent personal assistant service Alexa, which responds to the name "Alexa". This "wake word" can be changed by the user to "Amazon", "Echo" or "Computer".^[1][2] The device is capable of voice interaction, music playback, making to-do lists, setting alarms, streaming podcasts, playing audiobooks, and providing weather, traffic and other real-time information. It can also control several smart devices using itself as a home automation hub.
https://en.wikipedia.org/wiki/Amazon_Echo
 
 
So, what do we do?  I know I’m not going to use Alexa. What are you going to do?
 
 

[WolfBrother]

WB edited to specific:

TWP, I have WolfBrother's spreadsheet, but I will let him distribute it as he wishes. Actually, I think I got it here.

Jerry,
Domino Nabisco - you have been blessed and may share those with whomever you choose.
I appreciate the courtesy you showed.

ALL
If you want copies of the one page code sheet and/or the one time pad generator, either bug Jerry or contact me.

I am working on an authentication sheet and a callsign/freequency sheet still.

Will share those when they're done.