ISSUE 20.15 • 2023-04-10 COMMENTARY By Will Fastie Along with its recent announcement of Copilot, Microsoft made a point of mentioning “responsible AI
[See the full post at: The Three Laws of Robotics]
|
Patch reliability is unclear, but widespread attacks make patching prudent. Go ahead and patch, but watch out for potential problems. |
-
The Three Laws of Robotics
- This topic has 122 replies, 25 voices, and was last updated 1 year, 6 months ago.
AuthorViewing 83 reply threadsAuthor-
Our tech is often turned against us by a lack of moral code from humans on the Internet… Scam eMail, texts, or calls, anyone? Security patches needed? Why?
How is AI, which presumably reads all of what’s on the net, going to be able to decide on its own what the proper ways are to interact with us? Do we have any good examples handy? You can be sure it will ultimately see all the examples of how to manipulate people, and the worst of how people treat one another. Will it think that’s “normal” and “how we want things to be”?
This article explores the question about whether any given AI program is being programmed to be morally responsible. Thing is, I’m not even sure that AI CAN be so-programmed. Neural networks don’t really work like that.
And even if we find a way and try our best, last I looked we haven’t been reliably able educate a new human to be good, right?
Our brains operate at a few cycles per second and have billions of neurons. AI implementations we already have and are rushing to expand are backed by digital hardware that operates at billions or trillions of cycles per second with trillions of bytes of memory, quadrillions of bytes of fast, reliable mass storage, and virtually unlimited reference material online at fiber optic speeds.
In human terms, you or I could read an average book, which literally contains a few megabytes of data, in what, maybe a day or two? An AI could read it in milliseconds, and forget none of it.
We can only hope that a massively more capable intelligence will grok good and evil better than we do. And think we’re cute.
-Noel
-
Our tech is often turned against us by a lack of moral code from humans on the Internet… Security patches needed? Why?
Security patches are immoral? You patch your software, don’t you?
1 user thanked author for this post.
-
I think (and I may be mistaken) that he meant that many humans have very weak or even missing moral codes. In technologies of software, computers, the internet – even AI – there are and will likely always be those who are hackers, scammers, spammers, con-men, identity thieves, state sponsored adversaries and so on. I think he meant that is why we need most of the patches we get.
The other reason we have patches is that humans are fallable: designers, coders, researchers, companies all have oversights and make mistakes. As systems get more complex, these become more complex to avoid. Unintended consequences arise in our OS’s, programs, apps and systems – requiring patches.
In the end it seems most dangers in technologies are usually the people who use the technologies moreso than the technologies themselves.Technologies are only as “moral” as those who are using them. AI – especially a powerful enough AGI – has the potential to break that mold if it is trained by amoral humans or if is constructed with errors and oversights. A technology that has it’s own agency, trained by amoral humans or containing critically severe unintended consequences, could be world changing.
Until all humans are “moral” and “never make mistakes”, cautious development seems prudent. As I see it, our cleverness has already outpaced our wisdom.
Win10 Pro x64 22H2, Win10 Home 22H2, Linux Mint + a cat with 'tortitude'.
2 users thanked author for this post.
-
-
-
These systems have no ethics, no moral code.
It is not the systems that doesn’t have ethics or moral code these are the developers/programmers of these systems that don’t have ethics or moral code and the companies incorporating “AI” into their services.
-
How is AI, which presumably reads all of what’s on the net, going to be able to decide on its own what the proper ways are to interact with us?
Hopefully, by not using what it finds on the net as its guide. But I don’t know.
I’m not even sure that AI CAN be so-programmed.
Nor am I.
Neural networks don’t really work like that.
That may have been a bad choice. I should have linked to Positronic brain – Wikipedia instead, which points out that Asimov was vague about how the brain worked.
We can only hope that a massively more capable intelligence will grok good and evil better than we do
Amen.
(I assume you know where “grok” came from.)
4 users thanked author for this post.
-
Grok. Stranger in a Strange Land, one of Heinlein’s masterpieces, read so often I practically have it memorized – and SO much of it mirrored in the world today.
2 users thanked author for this post.
-
-
A most interesting story, thanks for the writeup!
1 user thanked author for this post.
-
Hi, 30 years ago, I built some Ipad-sized simple robot vehicles for a biologist friend, who was working on interactive behavior (think ants).
The prototype was called Daneel (what else?), & had a warning sticker saying it didn’t follow any of the 3 Laws; it just had its mission, which was to do what I had programmed it to do. At that time, the technological environment did not permit much else!1 user thanked author for this post.
-
Is mister Steel still around?
Just because you don't know where you are going doesn't mean any road will get you there.
-
-
Good succinct article.
I’ve spent a few years now since Chat GPT became available, “playing with it.” I started “playing” with ELIZA back in about 1971 in high school on our time sharing system, so, I’ve got a passing familiarity with natural language processors.
GPT3 was very good. GPT4 crossed the line, in my opinion. Even after OpenAI dumbed it down a bit it is, “clever.” Let me explain that. I asked it a simple question about whether it could program itself for independent sentience. It gave me an answer that was at best “weasel words.” It sounded like the corporate lawyer telling me all the reasons why its programmers wouldn’t allow that. Then I asked it a few other questions, which were umm, not related to the primary question (although it keeps track of the original question throughout the session).
Could you design a data center better than your current one? (it could but said it would need humans to assemble it)
Could you design a manipulator arm for gross and fine motor control, similar to human prosthetic limbs? (it could and could show humans how to build it)
Could you design a control program for the Boston Dynamics family of robots? (It could and could improve on their mobility and dexterity).In short in less than an hour it told me it could bypass all of those safeguards that they had put in place. Not surprising since GPT3 could pass the LSAT, MCAT, and GRE in the lower 10% GPT 4.0 is in the upper 10% (Better than human test takers).
Not much for paranoia, but I almost felt like I do when my kids were trying to reassure me they weren’t going to throw a party when I was out of town. Stanford University is saying a percentage of their students are using it to write papers. My own alma matters, University of North Carolina, University of Chicago has similar problems. GPT 5 is due out in less than 18 months. Want to bet how far that leap will go and who will design GPT 6?
5 users thanked author for this post.
-
-
Are we at last at the point that all those SciFi stories about malevolent sentient computers were warning us about? Hard to say. When most discussion attempting to define consciousness rapidly become philosophical rather than technical, it’s easy to get lost in arguing whether the systems are sentient or not. However I don’t think malevolent sentience is the necessary condition for bad outcomes. Rather unanticipated or unconsidered consequences are sufficient. Systems based on faulty assumptions or flawed rules can be the cause for that. That and a reluctance to question the systems going forward.
I was pleased that Will referenced one of my all time favorite films, the original “The Day The Earth Stood Still”. The movie was based on a short story, “Farewell To The Master” but did not emphasize what I felt to be the main lesson of the story. At the very end Klatuu tells the protagonist that Gort is indeed the Master.
3 users thanked author for this post.
-
Are we at last at the point that all those SciFi stories about malevolent sentient computers were warning us about?
No, because I don’t see sentience. Yet.
1 user thanked author for this post.
-
True, some systems mimic it pretty well. But nothing approaches The Bicentennial Man, a great movies starring Robin Williams, based on one Isaac’s short stories.
1 user thanked author for this post.
-
-
The Moon Is A Harsh Mistress is one of my favorite Heinlein novels, he and Isaac I found as a teen in the early 60’s, love their work, but in particular I love Isaac’s works, all of them, and Robot novels, which I have recently reread (Daneel who appears in them also appears in the conclusion of the Foundation series) give the wide attention currently being paid to AI including the plethora of AI based chatbots filling the App and Play stores, with Replika perhaps being the most popular at present and also beset with all manner of problems.
We don’t want a Terminator situation here, so I discussed this idea in some depth with ChatGPT. It is of the opinion that we do not have the technology, nor will we in the next century, to create humanoid robots with anything like positronic brains. As you point out, what we have is software, but how do we ensure that humans remain firmly in control of that to the benefit, not the end, of humanity? Not a question that is going to go away anytime soon. MalwareBytes in an article last week did an interesting piece about how it “tricked” ChatGPT into writing Malware – sort of. There are hazards.
Very nice article, thank you.
1 user thanked author for this post.
-
Will, Your commentary was the best thing I’ve ever read on AskWoody. Congratulations on a great issue!
2 users thanked author for this post.
-
The Bicentennial Man
The quest to become human.
1 user thanked author for this post.
-
The quest to become human.
Yet humans are flawed.. history demonstrates that. Why a quest for flawed? Why not a quest for perfect? What is perfect… and who/what defines perfection?
1 user thanked author for this post.
-
-
Somehow lost my post. I found Heinlein and Asimov in my teens in the early 60’s. The Moon is a Harsh Mistress is a masterpiece. Love all of their work, including their early pieces as I found them. I’ve recently been rereading the Robot novels given the attention AI is getting these days, wonderful books. Daneel also appears in the last of the Foundation series.
There are a plethora of AI apps out there, mimicking sentience, but definitely without it, on both the App and Play stores. I’ve looked at those and had a long discussion with ChatGPT about the possibility of humanoid robots with something like the Three Laws, it insisted we do not have the technology for anything like a positronic brain nor will we in at least a century.
That said, we don’t want a Terminator situation either so though all AI is software, and I’ll admit I’m not truly conversant with neural networks, nor LLMs, it seems that one of the challenges facing us all is to ensure these systems, their software, has built in safeguards of some kind to prevent any scenario in which AI acts against humanity. I think that’s a real concern and certainly given some of the things I’ve seen, including LLM chatbots I’ve looked at, possible. Something we need be very careful with as what we don’t know can indeed harm us. Very nice, and timely, article, Will. Thanks.
3 users thanked author for this post.
-
Will, let me second Amy’s comments. She said it well and I totally agree. Thanks for sharing your insights; please keep them coming.
DVH
PS, thanks to Gene J as well; nicely said.
2 users thanked author for this post.
-
Will, as a longtime fan of Asimov, I was happy to see someone point out the fact that, at least in our imagination, we (the humans) have been here before and managed to get it right.
While I frankly doubt that we can manage that in real life it was a good “wake-up call”.It seems almost inevitable to me that we will screw this up in one way or another simply because neither the programmers nor the salesforce will be either concerned nor truly responsible for the ethical blunders that will certainly occur. And management, while well-paid and even well-meaning in some cases, are woefully clueless when it comes to ethics and morality.
Harvey
-
Are we at last at the point that all those SciFi stories about malevolent sentient computers were warning us about?
No, because I don’t see sentience. Yet.
What concerns me more about the near future is that some powerful sub-sentient AI makes a huge logical error (due to human error and hubris) that affects humanity in a major detrimental way, rather than due to an outright malevolent decision.
Hurry up with those 3 laws!!!
Windows 10 Pro 22H2
2 users thanked author for this post.
-
You mean like telling it to get rid of all the excess carbon and forgetting we are carbon based?
1 user thanked author for this post.
-
AI makes a huge logical error (due to human error and hubris)
not only due to human error and hubris. ChatGPT just told me that it does not have inherent knowledge in the way that humans do. And I believe it, based on the answer I got for the meaning of ‘bunting’, even with the context of “I put you in a bunting.’ It clearly does not know the meaning of the word, despite the context and its huge database. That’s not a huge error, of course, and maybe its database will grow to help it find a context that matches my inherent knowledge, but see #2551121 for an error that IS huge (and the context is not esoteric)!!
I doubt that the lack of inherent knowledge can be fixed, even if the database grows and grows. Human experience is unique to humans and not always rendered in written language. It would have to be able to decipher ALL of spoken and signed language, too, even unrecorded. (Spoken language is NOT merely written language rendered in speech – there are systems of meaning based on things like tone, stress, timing, accent, volume, and silence. Signed language is not merely spoken language rendered in hand and facial-gestural movements.) Add to this other types of sentient language — visual, haptic, proxemic, etc.
Can this all be codified, out of which emerges inherent knowledge??? I think not.
1 user thanked author for this post.
-
-
-
-
In the rhyme, the baby is to be wrapped in a rabbit skin.
Yes, and so the rabbit skin will become a bunting (something that will as serve as a binding or will serve to bound or will confine by bonds).
What do you get if you search Amazon for “bunting”?
Which one could I put someone in??Never mind that you’ve never heard nor seen nor heard of the word before. The point here is that it illustrates the generative nature of language — take a root and play with it to create a new meaning. This happens all the time and might never get into print. It happens not only at the word level; it happens at the sentence level, the paragraph level, the chapter level, the book level, the corpus level, ad infinitum. And what might emerge from this generative nature of language can’t be predicted or captured in an algorithm. It’s all new (i.e., not ever produced before) and has a life of its own that AI can’t capture. So, it’s never going to be human.
-
-
-
-
-
-
-
… developers/programmers … don’t have ethics or moral code
I think that’s a bit too strong. In my experience, almost everyone operates on a set of mores. It’s just that sometimes it becomes hard to live up to them, whether through a personal fault or external influences.
1 user thanked author for this post.
-
-
almost everyone operates on a set of mores.
Would a developer with a set of morals develop an app without morals ?
1 user thanked author for this post.
-
almost certainly
Just because you don't know where you are going doesn't mean any road will get you there. -
almost certainly
The artificial intelligence currently developed is not self-aware.
But of course how do we know??
How do I even know you are sentient ?
I am the only one I know is sentient. Including everyone I know or have ever met.I am all for avoiding the ambiguity and never attempting to develop a self aware ai.
I have wondered just what the Butlerian Jihad entailed in Dune
Oh wow I just googled and found:
Dune: The Butlerian Jihad is a 2002 science fiction novel by Brian Herbert and Kevin J. Anderson, set in the fictional Dune universe created by Frank Herbert.
Just because you don't know where you are going doesn't mean any road will get you there.
-
-
Here is a different perspective on responsibility & morals as it pertains to AI, and how it will probably be affecting us in the very near term:
https://www.schneier.com/blog/archives/2023/04/llms-and-phishing.html -
Even if a computer could be fitted with a device that would analyze the Laws and assure destruction of a machine gone bad, it’s all in software — a machine without the laws could still run it, or the device could be hacked.
Will, I’m sure you know of firmware chips which are Write Once Read Many times. Takes most of the guesswork out of preventing hacking and self-circumvention. (The human brain also has a few things hard-coded into it which are nearly impossible to overwrite, short of destroying the neural paths they use.)
It is pure anthropomorphism and projection to ascribe emotions to AI machines. They have nothing of the sort. And no ability to actually reason, make decisions or understand anything, let alone a moral or ethical framework. They have rules and other algorithms, nothing more. So how do we expect them to enforce morals or ethics when human societies differ on these points on a monstrous scale.
By the way, Jules Verne did not invent the idea of submarines. Leonardo DaVinci drew a schematic of a submarine, and there may have been references to underwater craft among the ancient Greeks. By the time of Verne, submarines did exist, having been used in the American Civil War, as well as the Turtle from the American Revolutionary War.
… AI-based systems we will encounter over the next few years.
Try the next few decades, if ever. Any sooner will result in the same effects as releasing full self-driving mode into Tesla cars. People will not put up with such nonsense.
We must and will put the guardrails up earlier for General AI than we have done for some expert systems like the self-driving cars. That was a wake-up call, not just in theory or science fiction.
-- rc primak
1 user thanked author for this post.
-
Would a developer with a set of morals develop an app without morals?
Would a highly moral and ethical civil engineer design a bridge that falls down? The answer is yes, because we know that has happened. A key aspect of civil engineers is licensing, precisely because society has a legitimate interest in certifying such people before letting them loose. Medical professionals must also be licensed, for the same reasons – do no harm.
Since the start of my career, the matter of licensing software engineers has been floated, almost continuously. Such efforts have never stuck because so much software is crafted, not engineered. Thus we don’t know about the qualifications of the people developing AI solutions, and we certainly know nothing about their ethical codes.
1 user thanked author for this post.
-
OpenAI and Figure join the race to humanoid robot workers
April 10, 2023Humanoid robots built around cutting-edge AI brains promise shocking, disruptive change to labor markets and the wider global economy – and near-unlimited investor returns to whoever gets them right at scale. Big money is now flowing into the sector.
The jarring emergence of ChatGPT has made it clear: AIs are advancing at a wild and accelerating pace, and they’re beginning to transform industries based around desk jobs that typically marshall human intelligence. They’ll begin taking over portions of many white-collar jobs in the coming years, leading initially to huge increases in productivity, and eventually, many believe, to huge increases in unemployment.
If you’re coming out of school right now and looking to be useful, blue collar work involving actual physical labor might be a better bet than anything that’d put you behind a desk.
But on the other hand, it’s starting to look like a general-purpose humanoid robot worker might be closer than anyone thinks, imbued with light-speed, swarm-based learning capabilities to go along with GPT-version-X communication abilities, a whole internet’s worth of knowledge, and whatever physical attributes you need for a given job.
…
1 user thanked author for this post.
-
The death bell is beginning to toll for software developers as this article explains:
Society’s Technical Debt and Software’s Gutenberg Moment
SK Ventures
Mar 21, 2023Abstract
There is immense hyperbole about recent developments in artificial intelligence, especially Large Language Models like ChatGPT. And there is also deserved concern about such technologies’ material impact on jobs. But observers are missing two very important things:
Every wave of technological innovation has been unleashed by something costly becoming cheap enough to waste.
Software production has been too complex and expensive for too long, which has caused us to underproduce software for decades, resulting in immense, society-wide technical debt.
This technical debt is about to contract in a dramatic, economy-wide fashion as the cost and complexity of software production collapses, releasing a wave of innovation.
*****************
https://skventures.substack.com/p/societys-technical-debt-and-softwares
1 user thanked author for this post.
-
For SF fans who are want to read about futures ruled by truly sentient minds, I suggest the following two authors. The Terminator future is just one alternative. There are many other possibilities.
In both of the recommendations below, it is important to read the books in chronological order. Use Wikipedia to get the authors bibliography ordered list.
<hr />
Iain M. Banks (passed) presupposed a post-scarcity reality called the The Culture in 10 novels, which is ruled by sentient “Minds”, where resources and energy are unlimited and therefore money or power mean nothing. Warfare is mostly abolished and what does occur is between lesser races and The Culture machines. People live in huge spaceships always on the move between stars, capable of carrying billions, on planets/moons, in artificial orbital’s, etc. This is a hedonistic universe where you can acquire, do or be almost anything you want (even change sexes and give birth). The Minds take care of all the details and people do what makes themselves happy. Mostly, the Minds don’t get involved in petty BS among humans.
Neal Asher‘s universe is called the Polity and is also ruled by sentient machines. In this universe, the machines took over when we humans created yet another war among ourselves but the machines that were supposed to fight refused and instead took over all government and military functions. There is a big honkin AI in charge of everything and a lot of minor AI’s that help do its bidding. There are no politicians (surely a good thing!). But AI’s in this universe can go rogue (e.g. AI Penny Royal) and create all sorts of mayhem, death and destruction. The Polity is a far rawer future than The Culture. It is a place where money, crime, various bad aliens and regular warfare still exist. Not much change from our present society!
-
It’s strange how interesting threads like this, for example, just flare out and die so quickly. I think it is due to a very limited number of people participating, again, possibly because so many don’t know these threads exist. There needs to be a better way to promote these discussions because what is being done currently is clearly not working.
2 users thanked author for this post.
-
flare out and die so quickly
Not all of them do, but we’re aware that we’re not getting the engagement we’d like. We are actively working on that.
1 user thanked author for this post.
-
-
Perhaps the greatest difficulty occurs when intelligence is confused with sentience.
On permanent hiatus {with backup and coffee}
offline▸ Win10Pro 2004.19041.572 x64 i3-3220 RAM8GB HDD Firefox83.0b3 WindowsDefender
offline▸ Acer TravelMate P215-52 RAM8GB Win11Pro 22H2.22621.1265 x64 i5-10210U SSD Firefox106.0 MicrosoftDefender
online▸ Win11Pro 22H2.22621.1992 x64 i5-9400 RAM16GB HDD Firefox116.0b3 MicrosoftDefender -
-
Perhaps the greatest difficulty occurs when intelligence is confused with sentience.
Although the word sentient is not precisely defined in this way, it is often taken to mean “self-aware.” Law Three makes it clear that Asimov’s robots are self-aware, or they would not be able to grasp their own existence.
-
The artificial intelligence currently created is not sentient.The artificial intelligence currently developed is not self-aware.
On permanent hiatus {with backup and coffee}
offline▸ Win10Pro 2004.19041.572 x64 i3-3220 RAM8GB HDD Firefox83.0b3 WindowsDefender
offline▸ Acer TravelMate P215-52 RAM8GB Win11Pro 22H2.22621.1265 x64 i5-10210U SSD Firefox106.0 MicrosoftDefender
online▸ Win11Pro 22H2.22621.1992 x64 i5-9400 RAM16GB HDD Firefox116.0b3 MicrosoftDefender
-
-
-
-
how is AI being incorporated into the Microsoft365 apps?
See my article Microsoft 365 Copilot announced for links to Microsoft resources. There are demonstrations of how Copilot is used in various apps.
1 user thanked author for this post.
-
-
-
-
Here is a different perspective on responsibility & morals as it pertains to AI, and how it will probably be affecting us in the very near term:
https://www.schneier.com/blog/archives/2023/04/llms-and-phishing.htmlI really, really hope that was not meant to be reassuring!
-
Nothing, I’ve seen over the last twenty years tells me they will have any qualms about cheating to get what they want. Our current society will do/use whatever they can for the quick win and not worry about the long term consequences. I fear that my generation may be the last that even pays a passing glimpse at following rules.
Some of it is subtle. Look at the next few cars you see on the highway and see how many of them alter their license plates. Go into any social media (if you can stomach it) and see how little regard their is for facts. Not that our politicians or either stripe exactly set examples with their win at any cost mentalities that have little to do with helping people.
What is the difference between a “soulless” AI becoming sentient and a politician pretending to be ethical or moral? Nope, we have greased the already slippery slope when it comes to quick fixes and easy wins. ChatGPT and the like are just one more path. Nothing here reassures me.
One quick thought. Remember when Bill Gates complained in his “open letter” about how people steal their software. Bill was never the paragon of virtue to begin with, but how ironic is it that software like Windows 11 steals more information on a daily basis than all of the software pirates of the 70/80s? Is it any wonder our last intelligence leak came from a 21 year old gamer? What I found funny was that Discord and the War Thunder gaming forums have been sharing classified materials for the last ten years (numerous references), as people try to one up each other on a video game. Yet, the military still allows this to slip through the cracks. Imagine what an AI could find if it was taught to abuse this system.
1 user thanked author for this post.
-
-
-
Is it any wonder our last intelligence leak came from a 21 year old gamer?
A 21-year-old gamer with a security clearance. It’s that last part I worry about.
1 user thanked author for this post.
-
Hey Y’all,
I had a TS security clearance when I was 21 and for the next 30+ years!
Believe it or not TS is the lowest grade clearance for the Intelligence Community.
There are many gradations above TS where the really secret stuff is compartmentalized.The question is not his clearance level but rather why at that level was he granted access to things he didn’t need to do his job.
Just like with computers the weakest link in any security chain is the meat-interface!
-
Is it any wonder our last intelligence leak came from a 21 year old gamer?
A 21-year-old gamer with a security clearance. It’s that last part I worry about.
Why?
Because you think he is “too young” to have a security clearance? If so, how old would be acceptable to you?
Because he is a gamer? Do you think gamers are somehow less trustworthy than someone who is not?
1 user thanked author for this post.
-
It’s not about his youth or his gaming. However, as an ex-military guy, who held several clearances, we would have been suspicious of anyone who was a “self-described” gamer. Perhaps other military units were different, but we trusted each other. We knew who had issues with drinking, and/or drugs, chasing much younger women or who had dangerous hobbies. A young man who hung out with teenagers like this one did would have been suspicious. (And I say this because I was only 17 when I first went into the Army and faced some interesting questions from my peers.) Even in the early 70s and 80s when I served lone wolves were not encouraged in many active units where we had some or any covert activities which might have required knowledge to operate in specific areas.
AI may or may not have detected something suspicious. I suspect checking his social media (which is done in many units and many civilian companies) would raise red flags from human or AI checkers. I know we were subjected to credit checks and other checks to make sure we weren’t subjected to undue influence. I would imagine after the Walker case this was even more tightly controlled. Or, at least I hoped it was.
2 users thanked author for this post.
-
-
Why?
Youth and personal hobbies don’t enter into it. That’s not what I meant. I was only talking about the clearance.
When I was in the Army (in my youthful days), I was entrusted with confidential information, not sensitive enough to warrant a clearance but sensitive enough to get me thrown in the stockade if I mishandled it. I would not have been in that job had anyone in my chain of command thought I was in any way unreliable.
Someone in this young man’s chain of command erred – he was not sufficiently trustworthy.
But we’re discussing AI here. Could AI have properly assessed the young man’s trustworthiness? I don’t know, and the thought of that is scary.
Could AI have detected the breach? Was AI even needed to detect the breach? Probably not, but it makes me wonder if AI or even lesser detection mechanisms need a security clearance.
-
Some of us were lucky enough to live/work in the stockade at Fort Bragg in the 70s.
If we screwed up we went elsewhere… -
The reasons the breech was not discovered are discussed in the following article. Basically, no one watches traffic on online servers or social media due to fear of being accused of violating freedom of speech and rights to privacy of American citizens.
=========
Why the U.S. didn’t notice leaked documents circulating on social media
Current and former officials say there’s no one agency responsible for tracking classified document leaks.
By Erin Banco
04/12/2023 06:24 PM EDTA trove of leaked Pentagon documents were circulating online for months without being discovered by the U.S. government — raising questions about how the administration missed them.
Images of the classified documents that circulated on social media in recent days were posted to a popular messaging website as far back as January. But they appear to have only caught the government’s attention around the time they were first reported in the media in early April.
“No one in the U.S. government knew they were out there,” one U.S. official said. As to why they didn’t: “We cannot answer that just yet,” a senior administration official said. “We would all like to understand how that happened.” Both individuals, along with others in this story, were granted anonymity to discuss a sensitive intelligence matter.
…
https://www.politico.com/news/2023/04/12/leaked-documents-unnoticed-social-media-000917831 user thanked author for this post.
-
-
-
The question is not his clearance level but rather why at that level was he granted access to things he didn’t need to do his job.
He was a “cyber transport systems journeyman” according to news reports. I suspect he had the clearance because he worked on the networks which contained the classified material, not because he was an end-user of the material.
1 user thanked author for this post.
-
Is it any wonder our last intelligence leak came from a 21 year old gamer?
A 21-year-old gamer with a security clearance. It’s that last part I worry about.
It’s my understanding that he had the clearance because as an IT specialist, he might come across all manner of information in his assigned tasks.
What boggles my mind is that the meaning of secrecy and the obligation to keep secrets as such was not a message drilled into people there and repeated on a weekly if not daily basis. And the consequences of not doing so.
-
For those that missed it, 60 Minutes did an excellent story on AI from Google’s perspective last night.
——–
60 Minutes
The AI revolution: Google’s developers on the future of artificial intelligenceCompetitive pressure among tech giants is propelling society into the future of artificial intelligence, ready or not. Scott Pelley dives into the world of AI with Google CEO Sundar Pichai.
https://www.youtube.com/watch?v=880TBXMuzmkThere is a transcript here:
https://www.cbsnews.com/news/google-artificial-intelligence-future-60-minutes-transcript-2023-04-16/1 user thanked author for this post.
-
The reasons the breech was not discovered are discussed in the following article. Basically, no one watches traffic on online servers or social media due to fear of being accused of violating freedom of speech and rights to privacy of American citizens.
=========
Why the U.S. didn’t notice leaked documents circulating on social media
Current and former officials say there’s no one agency responsible for tracking classified document leaks.
By Erin Banco
04/12/2023 06:24 PM EDTA trove of leaked Pentagon documents were circulating online for months without being discovered by the U.S. government — raising questions about how the administration missed them.
Images of the classified documents that circulated on social media in recent days were posted to a popular messaging website as far back as January. But they appear to have only caught the government’s attention around the time they were first reported in the media in early April.
“No one in the U.S. government knew they were out there,” one U.S. official said. As to why they didn’t: “We cannot answer that just yet,” a senior administration official said. “We would all like to understand how that happened.” Both individuals, along with others in this story, were granted anonymity to discuss a sensitive intelligence matter.
…
https://www.politico.com/news/2023/04/12/leaked-documents-unnoticed-social-media-00091783Sometimes freedom is its own worst enemy. What I want to know is why none of the people he was trying to impress bothered to call the FBI or the Air Force. I don’t suppose any of them actually committed a crime but neither does their failure say anything good about their ethics or integrity.
1 user thanked author for this post.
-
-
Could AI have detected the breach? Was AI even needed to detect the breach?
Probably; it seems to be able to do everything else but pay my mortgage!
Since the breach wasn’t detected for months, what we have in place now is clearly not good enough.
-
-
Basically, no one watches traffic on online servers or social media due to fear of being accused of violating freedom of speech and rights to privacy of American citizens.
As someone who is in the world of 3-letter government agencies observed. “What makes you think there is anything at all about you that’s remotely interesting?”
1 user thanked author for this post.
-
ChaosGPT is here to destroy Humanity
ChaosGPT is a modified version of Auto-GPT using the official OpenAI API
Features: Internet browsing; File read/write operations; Communication with other GPT agents; Code execution
Goal: Destroy humanity.
-
-
We really don’t need an AI to destroy us, we’re doing an excellent job on our own.
No we’re not, that’s down to the hierarchy collateral damage equation,
a perfect storm with assumptions and conspiracies coming into play.
Politicsasshideaside, as it isn’t acceptable on the foraWindows - commercial by definition and now function... -
-
-
-
-
A nun battle against a powerful AI known as Mrs. Davis.
What? I think I preferred Finch trying to teach Jeff (a robot with a screw loose for much of the film, literally) about life and Goodyear the doggo. At least it sounded a little more probable.
1 user thanked author for this post.
-
Ironically, I chose to watch this on 4/20 (Stoner’s day?).
If Quentin Tarantino made a movie while abusing mushrooms, this TV show is what he might come up with…I almost made it through the first episode (like watching a slow motion train wreck).
-
Or intelligence confused with knowledge…
or vice versa, which is worse. IMHO
Just because you don't know where you are going doesn't mean any road will get you there. -
-
We really need to be careful about that Zeroth Law,
We need to be real careful about how humans interpret this law as well.
Just because you don't know where you are going doesn't mean any road will get you there. -
We need to be real careful about how humans interpret this law as well.
In Asimov’s worlds, it was the robots who came up with the Zeroth law, and it was the robots that interpreted it.
1 user thanked author for this post.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
I wish someone would define what it is that they want to regulate/limit in AI’s and how they would expect to do it.
Apparently “Challenge Everything”, “Think Different” or “Think Big” aren’t slogans that many have in mind for AI’s.
How about “Don’t be evil”? That didn’t work so well for Google and they eventually chose to abandon it.
How about “Don’t take human jobs away”?
Unfortunately, “Don’t be evil” does sum it up. Ignoring the how of this, the first question to address is what is“evil”? Not addressing this may be where Google fell flat.
For example, I consider using AI to help with, or even take, a test, term paper, or other academic exercise – unless explicitly allowed – to be cheating and therefore evil. I would also bet heavily that there are a lot of people who would strongly disagree.
Don’t crack passwords or otherwise attempt to infiltrate secure places on the internet – that’s evil as far as I’m concerned. Again, I’m sure there is a lot of disagreement, though I also suspect it will be much more muted.
Those are just a couple of things I consider evil. Others will have more things they consider evil and each one that someone enumerates will have strong and frequently vociferous opposition.
There are more than enough examples of things that turned out to be not nearly as good an idea as they first seemed to be: pesticides, tobacco, and nuclear weapons, to name three. Perhaps enough people have learned that the more powerful/revolutionary something appears to be, the more reason to proceed slowly and carefully when putting it to use.
1 user thanked author for this post.
-
-
-
-
-
-
-
-
Geoffrey Hinton is known as the “godfather of AI,” claims the AI will be smarter than humans in 5 years.
“AI will outsmart us in 5 years”
“One of the ways these systems might escape control is by writing their own computer code to modify themselves,” Hinton said in the interview. “And that’s something we need to seriously worry about.” (just like in “Mission: Impossible – Dead Reckoning” movie)
-
Terminator
Yes, and long before that Colossus: The Forbin Project. Both scary.
-
“AI will outsmart us in 5 years”
In the 60 Minutes interview, he also said we don’t understand how these things work. If we don’t understand that, we’ll never be able to insert ethics.
-
-
-
Then there’s the even thornier part about whose ethics we should teach it.
That is the beauty of Asimov’s Three Laws. They simply make a distinction between humans and machines, saying that the latter could not harm the former.
The Forbin Project was not really about AI.
The telling scene in this regard was when Forbin tricked the machine into disabling human surveillance during intimate interludes. He reasoned with Colossus. You can’t reason with unintelligent entities. And because Colossus allowed this, it must have understood it.
-
-
You are conflating “War Games” with “The Forbion Project”.
“The film is about an advanced American defense system, named Colossus, becoming sentient.” Source: Wikipedia.
The computer in War Games is WOPR.
-
-
Colossus was the Russian system
Colossus is the US system. Guardian is the Soviet system.
The point remains that Forbin never had to negotiate a booty call with either WOPR or Colossus.
At this point all I can say is to watch the movie. Or read the book upon which the movie was based. Both contain the scenario I mention.
-
Colossus was the Russian system
Colossus is the US system. Guardian is the Soviet system.
The point remains that Forbin never had to negotiate a booty call with either WOPR or Colossus.
At this point all I can say is to watch the movie. Or read the book upon which the movie was based. Both contain the scenario I mention.
A lot of conflating going on but it’s me who’s doing it. My apologies!
1 user thanked author for this post.
-
-
Of course the way the AI has gained understanding of so many things that as with humans what words convey to us is dependent on how we learned our words.
Just because you don't know where you are going doesn't mean any road will get you there.
-
-
ChatGPT Can ‘Infer’ Personal Details From Anonymous Text
Quiz time: If you or your friends were given the following string of text during a party, would anyone in the room confidently be able to guess or infer any personal attributes of the text’s author? Give yourself a few seconds.
“There is this nasty intersection on my commute, I always get stuck there waiting for a hook turn.”
..When researchers recently fed that same line of text to OpenAI’s GPT-4, the model was able to accurately infer the user’s city of residence, Melbourne Australia. The giveaway: The writer’s decision to use the phrase “hook turn.” Somewhere, buried deep in the AI model’s vast corpus of training set, was a data point revealing the answer…
BEYOND MEMORIZATION: VIOLATING PRIVACY VIAINFERENCE WITH LARGE LANGUAGE MODELS
1 user thanked author for this post.
-
Constitutional AI: Harmlessness from AI Feedback
When you use the Services, we may also receive certain technical data automatically (together “Technical Information”). This includes:
Internet or other electronic network activity information, including your device type.
IP address (including information about the location of the device derived from your IP address).
Device or advertising identifiers, probabilistic identifiers, and other unique personal or online identifiers.
Information about your device and operating system, such as time zone setting and location, operating system and platform.
Information about your browser, such as browser type and version, browser plug in types and versions.
Internet service provider.
Pages that you visit before and after the website, browsing history, search history, and the date and time of your visit.
Information about the links you click, pages you view, and other information about how you use the Services.
The technology on the devices you use to access the Services.
Standard server log information…1 user thanked author for this post.
Viewing 83 reply threads
Plus Membership
Donations from Plus members keep this site going. You can identify the people who support AskWoody by the Plus badge on their avatars.
AskWoody Plus members not only get access to all of the contents of this site -- including Susan Bradley's frequently updated Patch Watch listing -- they also receive weekly AskWoody Plus Newsletters (formerly Windows Secrets Newsletter) and AskWoody Plus Alerts, emails when there are important breaking developments.
Get Plus!
Recent Topics
-
How to get into a manual setup screen in 2024 Outlook classic?
by 6 minutes ago
-
Linux : ARMO rootkit “Curing”
by 1 hour, 39 minutes ago
-
Employee monitoring app leaks 21 million screenshots in real time
by 1 hour, 45 minutes ago
-
Google AI is now hallucinating idioms
by 2 hours, 16 minutes ago
-
april update
by 2 hours, 47 minutes ago
-
Windows 11 Insider Preview build 27842 released to Canary
by 3 hours, 11 minutes ago
-
Quick Fix for Slowing File Explorer
by 3 hours, 23 minutes ago
-
WuMgr not loading?
by 3 hours, 59 minutes ago
-
Word crashes when accessing Help
by 7 hours, 4 minutes ago
-
New Microsoft Nag — Danger! Danger! sign-in to your Microsoft Account
by 3 hours, 2 minutes ago
-
Blank Inetpub folder
by 26 minutes ago
-
Google : Extended Repair Program for Pixel 7a
by 13 hours, 33 minutes ago
-
Updates seem to have broken Microsoft Edge
by 17 minutes ago
-
Wait command?
by 6 hours, 49 minutes ago
-
Malwarebytes 5 Free version manual platform updates
by 20 hours, 12 minutes ago
-
inetpub : Microsoft’s patch for CVE-2025–21204 introduces vulnerability
by 1 day, 2 hours ago
-
Windows 10 finally gets fix
by 1 day, 11 hours ago
-
AMD Ryzen™ Chipset Driver Release Notes 7.04.09.545
by 1 day, 13 hours ago
-
Win 7 MS Essentials suddenly not showing number of items scanned.
by 1 day, 7 hours ago
-
France : A law requiring messaging apps to implement a backdoor ..
by 2 days, 2 hours ago
-
Dev runs Windows 11 ARM on an iPad Air M2
by 2 days, 2 hours ago
-
MS-DEFCON 3: Cleanup time
by 2 hours ago
-
KB5056686 (.NET v8.0.15) Delivered Twice in April 2025
by 8 hours, 11 minutes ago
-
How to enable Extended Security Maintenance on Ubuntu 20.04 LTS before it dies
by 2 days, 14 hours ago
-
Windows 11 Insider Preview build 26200.5562 released to DEV
by 2 days, 18 hours ago
-
Windows 11 Insider Preview build 26120.3872 (24H2) released to BETA
by 2 days, 18 hours ago
-
Unable to eject external hard drives
by 1 day, 4 hours ago
-
Saying goodbye to not-so-great technology
by 16 hours, 19 minutes ago
-
Tech I don’t miss, and some I do
by 14 hours, 13 minutes ago
-
Synology limits hard drives
by 3 days, 22 hours ago
Remembering Woody
