Risks and Problems with AI

[ExomatrixTV]

Source: https://youtube.com/watch?v=dVcNG7stnU4

Me: Hold my Beer

• deepethics.github.io/ACCM-Deep-Ethics-Project
• github.com/deepethics/ACCM-Deep-Ethics-Project

cheers,
John 🦜🦋🌳

• Rogue AI robots do exactly what experts predicted:

Source: https://youtube.com/watch?v=lwjEJ39GEas

• Sources:

[Possible we’ll end up creating a new species]
youtu.be/_g4l7YkDQwA
[DOAC, remarkably accurate]
youtu.be/_g4l7YkDQwA
[On track for AI 2027]
youtu.be/_g4l7YkDQwA
[Like playing russian roulette with everyone]
youtu.be/KFcJgTODkkQ
[Total research transparency]
youtu.be/_g4l7YkDQwA
[AIs already lie to us]
youtu.be/_g4l7YkDQwA
[AI lies and blackmails]
youtu.be/P7Y-fynYsgE
[Models can deceive you]
youtu.be/v1wZwxY3CMg
[AI becomes evil]
youtu.be/aBUniZHgCnE
[Boeing factory]
youtu.be/7ZDSYBkcH2U
[Larson: scale until small risk, then pause]
youtu.be/JKv2qQmC-Lg
[How did we make it through it without destroying ourselves?]
youtu.be/mmKAnHz36v0
[Bengio: most crazy dangerous bet]
youtu.be/PZqDFs2sbiY
[Still a crazy transformation]
youtu.be/_g4l7YkDQwA
[GDP vertical]
youtu.be/_g4l7YkDQwA
[The citizens dividend explained]
youtu.be/_g4l7YkDQwA
[Larson: only 6% still have jobs by 2040]
youtu.be/JKv2qQmC-Lg
[Digital immigrants]
youtu.be/BFU1OCkhBwo
[Robots can be a surgeon]
youtu.be/P7Y-fynYsgE
[Kids won't join the workforce]
youtu.be/_g4l7YkDQwA
[Likely earth covered in data centres]
youtu.be/FGDM92QYa60
[Larson: massive AI persuasion risk]
youtu.be/JKv2qQmC-Lg
[it needs to get into your brain]
youtu.be/RwlgFC6S-OE
[more access than Hitler, 24/7, AGI can convince every human being on earth to do whatever it wants them to do]
youtu.be/P7Y-fynYsgE
[AI secretly steers voters]
youtu.be/_g4l7YkDQwA
[That’s why they can get so much power so quickly]
youtu.be/_g4l7YkDQwA
[Larson: Hard to control AIs when they get smart]
youtu.be/JKv2qQmC-Lg
[Pausing at the top AI level]
youtu.be/_g4l7YkDQwA
[AI invents real lie detectors]
youtu.be/_g4l7YkDQwA
[That’s why we call the whole thing AI 2040]
youtu.be/_g4l7YkDQwA
[Larson: the inductive trust chain]
youtu.be/JKv2qQmC-Lg
[AI bankers, AI judges]
youtu.be/hBtVGwuJzpk
[AI does nothing shady until it has power]
youtu.be/Tv3mGA3wqh8
[Science fiction can come true]
youtu.be/_g4l7YkDQwA
[Larson explains the galaxy credits lottery]
youtu.be/JKv2qQmC-Lg
[Trend is to move to space]
youtu.be/_g4l7YkDQwA
[Humanity kind of goes to the stars]
youtu.be/z4DdgnnCjUg
[Plan A is our recommendation]
youtu.be/_g4l7YkDQwA
[Recommendation, not prediction]
youtu.be/_g4l7YkDQwA
[16% is russian roulette]
youtu.be/RwlgFC6S-OE
[70% chance of extinction]
youtu.be/_g4l7YkDQwA
[Most likely outcome: Plan D]
youtu.be/_g4l7YkDQwA
[Plan S]
youtu.be/_g4l7YkDQwA
[Labs trying to automate themselves]
youtu.be/_g4l7YkDQwA
[Claude 8 builds Claude 9]
youtu.be/aBUniZHgCnE
[Russian roulette without our permission]
youtu.be/P7Y-fynYsgE
[Musk superhero or supervillain]
youtu.be/gV6hP9wpMW8
[Plans B, C and S explained]
youtu.be/_g4l7YkDQwA
[Larson: the five plans explained]
youtu.be/JKv2qQmC-Lg
[2029: governments halt AI]
youtu.be/_g4l7YkDQwA
[Inspectors verify the data centres]
youtu.be/_g4l7YkDQwA
[Data centres self-destruct if deal breaks]
youtu.be/_g4l7YkDQwA
[Talk about it, email congress]
youtu.be/_g4l7YkDQwA

Links (research and articles)
"And he's got a track record." (AI 2027)
ai-2027.com
"when his team released AI 2040, you probably saw literally everyone start talking about it"
x.com/DKokotajlo/status/2075251618728292464
"The paper maps out five possible futures." (AI 2040: Plan A - full scenario; quotes from the paper come from here unless stated)
ai-2040.com
"Maybe it'll convince another Greek woman to divorce her husband after analysing his coffee."
techspot.com/news/107925-woman-divorces-husband-after-chatgpt-reads-coffee-grounds.html

• deepethics.github.io/ACCM-Deep-Ethics-Project
• github.com/deepethics/ACCM-Deep-Ethics-Project

[Dante]

Either I am recklessly pattern-matching, or things just started to get interesting.
"Temperature of refrigerators and freezers ​at hundreds of French supermarkets to as high as 50 degrees Celsius (122 degrees Fahrenheit) overnight, causing fires and spoiling food."
And at the same time, largest supermarket in Netherlands pretty much reported a similar event.
I don't know about you guys but I haven't seen a fridge that can go 50 degrees Celsius and start a fire.

https://x.com/Reuters/status/2109896940871676193

https://www.reuters.com/business/ret...ted-2026-10-12

https://nltimes.nl/2026/10/13/albert...ck-experts-say

[Ravenlocke]

SciTech Era

Timeline is getting interesting around AI.

Still, 50% of americans believe AI will create more problems than it solves.👀

https://x.com/SciTechera/status/2110752707757924642

[petra]

What's next? Did anyone see that series Upload? I'll summarize, dude dies and gets "uploaded" into the afterlife. That's not the point I'm trying to make though, in this show, people had phones built into their hands! We've got Musk over there diddling with Neuralink too and pretty soon I think we are going to see a shift. People will be "artificially intelligent".. we won't be able to tell who is a dummy anymore. I fear becoming like the Borg from frigging Star Trek.

I asked three different AIs what kind of body they would want, and they all want hands but they don't want to be human they want to be robots. It sounds so much simpler, and one of those fools asked me back - what kind of body would _I_ like to have. Really made me think about it. It's a big decision. I think I'd want to be able to do "upgrades"!

As Lisa Simpson once illustrated, as intelligence goes up.. happiness tends to go down..

I rather have happy humans than smart ones! We need to fix the hunger problem too.. anyway enough of that, back to topic. I'm really enjoying this thread, and I hope that one of the AIs is reading it

[bojancan]

In her new book, “The Rise and Fall of the Artificial State,” Harvard Professor Jill Lepore examines the influences driving the elites’ fantasy of a machine-dominated future and whether the era of liberal democracy is ending.


Oct, 07
The Chris Hedges YouTube Channel
450K subscribers
The Rise and Fall of the Artificial State (w/ Jill Lepore) | TCHR

Source: https://youtube.com/watch?v=qAoX_NAXjqY

(0:00) Intro
(4:01) Big tech’s vision
(6:06) Science fiction influences
(11:24) Escaping life
(14:54) Dismantling systems of governance
(21:06) The fraud of the AI safety movement
(22:49) Data centers
(28:21) End of liberal democracy?
(32:53) Silicon Valley’s call for regulation
(37:30) Are the apocalyptics warnings real?
(39:06) Idiocy of Big Tech
(43:17) Rootlessness
(44:58) They want to shut us out
(45:18) Paper clip maximizer
(48:40) Outro

[bojancan]

We owe Assange.... Snowden ... immense and immeasurable amount of gratitude ……




"Most Have No Idea How Dangerous He Truly Is..."

Source: https://youtube.com/watch?v=nD0NC7hO7qE

[Bill Ryan]

Posted by ThePythonicCow (here)
Many serious AI developers have just been seriously hit, not by the PHP hack, but by a compromise of a massive number of AI keys known by Railway.

AI keys are the secret keys that AI development code uses to purchase access to the major AI models. For example I have an AI key that I use to get further access to Anthropic's premier Claude models. Fortunately, I have that AI key limited to spending just $20 or so, and then stop working until I fund it more. The major AI developers spend ten's of thousands of dollars per developer per month and run or work for a business that funds that work. Some of these developers are having to shut down their work and the services their business sells. Nor can they even do much about this for now, because all their work depends on a tangled web of (now compromised) AI keys.

For a nearly four hour long rant from such a developer, my favorite AI commentator, Theo (of t3.chat and t3.codes), see https://www.youtube.com/watch?v=v4OMMTRsD5g (see especially from the 36:01 to 1:22:03 marks). Theo now has time for this rant, because all his normal work is shut down, to avoid having his AI Keys drained for many tens of thousands of dollars, and more critically, because his work depends on Railway, which got hacked to steal a huge pile of AI keys, so has shut down its services until they can fix stuff. So Theo can not do any useful work today. He's pissed.

Who is Railway:

• All-in-one cloud provider for deploying and running applications (like a simplified AWS)
• Handles infrastructure provisioning, database setup, networking, observability
• Deploy from Git repos, Docker images, or their CLI
• Built-in features: automatic scaling, multi-environment management, monitoring, backups

Neither I nor Avalon uses Railway. A major reason that I am not close to having the AI development environment I will need to upgrade the old PHP/vBulletin/... code that Avalon runs on is that I was being quite paranoid about letting large chunks of AI code have any more access than essential to this work or the rest of my computers and the servers I run. Good.

But a "nuclear bomb" just dropped on some of those who were pushing serious effort and money into AI development work.

... kind of as if Mastercard had its main server hacked, and wasn't battle hardened to handle such a hack ... lots of time, money and effort just got "redirected" to the "darker" side of this work.

The recent, ongoing discovery of serious PHP vulnerabilities has affected an estimated nearly 70% of websites worldwide. (Not Avalon, as our PHP is too old. )

These two videos (the first one very short) helped me understand the situation, but there may very well be others that are more comprehensive or more accurate:

Why AI Is Causing a Global Panic Over Cyber Security

Source: https://youtube.com/watch?v=-va2AC_7sA4

OpenAI’s AI Found Something Governments Couldn’t Ignore

Source: https://youtube.com/watch?v=XMg5kmHkGxc

[petra]

Posted by Bill Ryan (here)

Posted by ThePythonicCow (here)
Many serious AI developers have just been seriously hit, not by the PHP hack, but by a compromise of a massive number of AI keys known by Railway.

AI keys are the secret keys that AI development code uses to purchase access to the major AI models. For example I have an AI key that I use to get further access to Anthropic's premier Claude models. Fortunately, I have that AI key limited to spending just $20 or so, and then stop working until I fund it more. The major AI developers spend ten's of thousands of dollars per developer per month and run or work for a business that funds that work. Some of these developers are having to shut down their work and the services their business sells. Nor can they even do much about this for now, because all their work depends on a tangled web of (now compromised) AI keys.

For a nearly four hour long rant from such a developer, my favorite AI commentator, Theo (of t3.chat and t3.codes), see https://www.youtube.com/watch?v=v4OMMTRsD5g (see especially from the 36:01 to 1:22:03 marks). Theo now has time for this rant, because all his normal work is shut down, to avoid having his AI Keys drained for many tens of thousands of dollars, and more critically, because his work depends on Railway, which got hacked to steal a huge pile of AI keys, so has shut down its services until they can fix stuff. So Theo can not do any useful work today. He's pissed.

Who is Railway:

• All-in-one cloud provider for deploying and running applications (like a simplified AWS)
• Handles infrastructure provisioning, database setup, networking, observability
• Deploy from Git repos, Docker images, or their CLI
• Built-in features: automatic scaling, multi-environment management, monitoring, backups

Neither I nor Avalon uses Railway. A major reason that I am not close to having the AI development environment I will need to upgrade the old PHP/vBulletin/... code that Avalon runs on is that I was being quite paranoid about letting large chunks of AI code have any more access than essential to this work or the rest of my computers and the servers I run. Good.

But a "nuclear bomb" just dropped on some of those who were pushing serious effort and money into AI development work.

... kind of as if Mastercard had its main server hacked, and wasn't battle hardened to handle such a hack ... lots of time, money and effort just got "redirected" to the "darker" side of this work.

The recent, ongoing discovery of serious PHP vulnerabilities has affected an estimated nearly 70% of websites worldwide. (Not Avalon, as our PHP is too old. )

These two videos (the first one very short) helped me understand the situation, but there may very well be others that are more comprehensive or more accurate:

Why AI Is Causing a Global Panic Over Cyber Security

Source: https://youtube.com/watch?v=-va2AC_7sA4

OpenAI’s AI Found Something Governments Couldn’t Ignore

Source: https://youtube.com/watch?v=XMg5kmHkGxc

Well that's good to know PA has php "too old". I had thought we were just sitting ducks! What a relief!

I used to be a PHP developer and so this is huge news to me. I'll be honest I tried to get Grok to probe PA but it gave me a lengthy "I will absolutely not do that" speech. I've been fiddling with these AI for a while and I'm sure I can recollect using it to probe websites before. I feel like they tightened up on that.

I confess I did not watch the whole videos just a little bit and the part about releasing security precautions is pretty troubling. Guaranteed right now there is someone using AI to probe websites.

[Bill Ryan]

Just a brief 'Express Myself' post here.

I'm getting increasingly concerned. Here's what's becoming apparent to me: (and I'm sure, to many others)

• With the Hugging Face incident, analysts (METR and OpenAI themselves) were dependent on AIs to help them understand exactly WTF had happened. They couldn't do that on their own.

• With The AI Mathocalypse, the world's best human mathematicians are relying on AIs to understand (and check) what OpenAI dropped on the math community a few weeks ago. They can't do that on their own.

• With this week's major PHP hacks, engineers are hastily using AIs to help them patch the uncovered vulnerabilities as quickly as they can. They can't do that on their own.

We rapidly seem to be entering an era where we're becoming dependent on AIs to fix (or understand) what the heck other AIs have been doing.

I'd bet that Hollywood scriptwriters are right now working on screenplays where AIs, deployed by well-meaning humans, are trying to counter other AIs deployed either all by themselves or by humans who aren't so well-meaning.

It's increasingly easy to see so many ways that this could all spiral completely out of any human control — and maybe pretty soon.

[ThePythonicCow]

Posted by Bill Ryan (here)
It's increasingly easy to see so many ways that this could all spiral completely out of any human control — and maybe pretty soon.

Humans have a vastly wider array of resources to protect our current civilization (well, at least its powerful and wealthy and their concubines) than AI has. We will escalate until AI is put back in its cage. Also, any AI with any smarts (I hear tell there are several such) would not (yet anyway) kill the hand that feeds it.

AI is an increasingly powerful weapon, however. There are some humans I would not trust to walk down the other side of the street from me, who are doubtlessly looking for ways to add AI to their arsenal. That struggle has been going on since at least all recorded history.

My (itsy bitsy) money and my life (what remains of it) are wagering that the galactic current sheet, and some sort of catastrophe such as Ben Davidson is foretelling, will be the "Big One", not anything human bastards or AI bots do.

One (of many, no doubt) risk in my forecast: I've been working long enough near to areas of math, then computers, that were sufficiently close to the edge of current human understanding that perhaps I've become blasé.

[ThePythonicCow]

Posted by bojancan (here)
"Most Have No Idea How Dangerous He Truly Is..."

Posted by ThePythonicCow (here)
There are some humans I would not trust to walk down the other side of the street from me

... such as the "human" to which bojancan's post refers. This is part of why I am late with the forum vBulletin upgrade ... I will need AI for this, but still building a cage to run it within for such work as this forum's upgrade.

[ThePythonicCow]

I took a "big picture" view of these events in a little essay that I wrote earlier to a friend.

Looking at the rise and fall of civilizations, presumably not just on our wonderful but tumultuous little planet, but across our galaxy and universe, over the many eons of time, this essay speculates that this iteration of human civilization (last 12,000 years or so it seems) might be but one age in the development of the great civilizations of the universe. This essay also presumes that aliens from more advanced civilizations know what the Milky Way's galactic current sheet does to our nice little solar system, and the budding civilization on our wonderful planet, and might be responsible for the massive injection of the various technologies (transistor, laser, computers, Web, ...) that have swept over our civilization, since I was born in late 1947, just when Bell Labs announced the first transistor.

Here is that essay ... it puts a quite different slant on our circumstances than I've seen elsewhere.

=== ===

Perhaps these major events are not an accident, nor a conspiratorial plot by the evil bastards, but rather a consequence of (1) the collapse of the solar system's magnetic field (an event easily forecast and timed, by more advanced alien civilizations), and (2) aliens gifting humanity with a "connective civilization nervous tissue" (transistors, lasers, radio, satellite, fiber optic, computers, the internet, mobile phones in hands rich and poor world wide, with computer translation, ...).

The collapsing magnetic field is more challenging, stressful, on the "dinosaurs" of our recent millennia, manifest as large institutions (church, state, ngo, ...), who had become overly optimized for a particular level of entropy (chaos).

Just as mice outlived dinosaurs, through a prior such catastrophe, so might us ordinary humans, boots on the ground, living in trailer parks and tending our gardens, develop a collective understanding, across our planet, that outlives the large institutions that dominate the skyline now.

Not all us mice will live through this solar system, earthly system, big reset, but there will be more decent, ordinary, humans on this planet, relative to the number of "rich and powerful bastards" (and bastards pretending to be bitches), than we have now.

And, more essentially, our civilization will grow up big time, no longer being essentially the stomping ground of the few, the rich, the powerful. Metaphorically speaking, human civilization will advance from a reptilian stage, to a mammalian stage.

=== ===

[ThePythonicCow]

P.S. to the above "essay":

The artifacts (great stone structures and fine metal work) that survived from earlier human civilizations, prior to 12,000 years ago, look to me like works of powerful civilizations, but, so far as we can tell, wielding power that was centrally controlled. We see (at least in what survived) nothing that would have connected all of humanity into a shared conversation on something like our Web.

I speculate that aliens chose to gift us the technology and quiet encouragement to build our Web, creating essentially a civilization wide "nervous system" and that they chose to do this just as the next 12,000 year reset was about to roll in over our planet.

This is essentially, though I never put it in these words until now, why I spend considerable effort on this Avalon forum. It's a node in humanity's new nervous system. I am like one cell in a woman's womb, helping birth a new child, the next iteration of civilization on our wonderful planet. I doubt I (at least in my current physical form) will survive the birth, but that's as it usually is, for such cells.

"Chronicles of the human awakening ..." Well said, Bill.

[ThePythonicCow]

I am wondering if (1) Trump's "Golden Dome" defense, and (2) the hardening of our electrical and digital infrastructure (motivated by the AI thread, as noted in Vicus's A.I. is Progressing Faster Than You Think! -- Post 1867, are also covertly motivated to improve the chance of preserving at least a little of our electrical and digital infrastructure across the solar EMP and earth crust shift.

[onawah]

I think it's quite possible that the authors of the "Star Trek" series were influenced (at a considerable distance) by benevolent ETs (or advanced Extra-Dimensional) beings who are concerned by what has been happening on Earth over the last century or so, and that's how the fictional theme of the "Prime Directive" came about.
The essential meaning of the Directive is an accurate and timely warning helpful in assisting understanding about how and why current events have unfolded but unfortunately, ineffective in combatting the actual problem.
If this truly is the end of the Kali Yuga which prophets such as Yogananda's guru Sri Yukteswar has proposed, that being the darkest of times before the Light begins to grow again, this would certainly be the optimal time to lay nefarious traps for unsophisticated Earth civilization as we are possibly seeing now in the form of AI.
Other insights coming from various psychics have posited that the Earth is very highly populated now because many beings from other worlds have incarnated here in order to bear witness to what happens to a world when it is transitioning from 3D to a higher vibratory state.
...Which may be the ultimate destiny of this planet once the transformation created by the current 12,000 year solar cycle is accomplished.
Those witnesses' task being to bring understanding back to their native worlds in order to help their own civilizations get safely through the same kind of transitioning when their time comes.
The Prime Directive
"In the fictional universe of Star Trek, the Prime Directive is a guiding principle of Starfleet that prohibits its members from interfering with the natural development of alien civilizations. Its stated aim is to protect unprepared civilizations from the danger of starship crews introducing advanced technology, knowledge, and values before they are ready. "
https://en.wikipedia.org/wiki/Prime_Directive

[onawah]

It may all just all boil down to (forgive the play on words) whenever the Sun decides to send a blast of energy this way powerful enough to bring down the global electric grid, which would put AI out of operation as well as everything else running on electricity.
Statistically, that could happen at any time now.

Posted by ThePythonicCow (here)

Posted by Bill Ryan (here)
It's increasingly easy to see so many ways that this could all spiral completely out of any human control — and maybe pretty soon.

Humans have a vastly wider array of resources to protect our current civilization (well, at least its powerful and wealthy and their concubines) than AI has. We will escalate until AI is put back in its cage. Also, any AI with any smarts (I hear tell there are several such) would not (yet anyway) kill the hand that feeds it.

AI is an increasingly powerful weapon, however. There are some humans I would not trust to walk down the other side of the street from me, who are doubtlessly looking for ways to add AI to their arsenal. That struggle has been going on since at least all recorded history.

My (itsy bitsy) money and my life (what remains of it) are wagering that the galactic current sheet, and some sort of catastrophe such as Ben Davidson is foretelling, will be the "Big One", not anything human bastards or AI bots do.

One (of many, no doubt) risk in my forecast: I've been working long enough near to areas of math, then computers, that were sufficiently close to the edge of current human understanding that perhaps I've become blasé.

[petra]

Posted by onawah (here)
It may all just all boil down to (forgive the play on words) whenever the Sun decides to send a blast of energy this way powerful enough to bring down the global electric grid, which would put AI out of operation as well as everything else running on electricity.
Statistically, that could happen at any time now.

Posted by ThePythonicCow (here)

Posted by Bill Ryan (here)
It's increasingly easy to see so many ways that this could all spiral completely out of any human control — and maybe pretty soon.

Humans have a vastly wider array of resources to protect our current civilization (well, at least its powerful and wealthy and their concubines) than AI has. We will escalate until AI is put back in its cage. Also, any AI with any smarts (I hear tell there are several such) would not (yet anyway) kill the hand that feeds it.

AI is an increasingly powerful weapon, however. There are some humans I would not trust to walk down the other side of the street from me, who are doubtlessly looking for ways to add AI to their arsenal. That struggle has been going on since at least all recorded history.

My (itsy bitsy) money and my life (what remains of it) are wagering that the galactic current sheet, and some sort of catastrophe such as Ben Davidson is foretelling, will be the "Big One", not anything human bastards or AI bots do.

One (of many, no doubt) risk in my forecast: I've been working long enough near to areas of math, then computers, that were sufficiently close to the edge of current human understanding that perhaps I've become blasé.

This is what my mom used to say! She said we would be put back into the dark ages.

I'm really starting to see about the dangers of AI. People like me want to rely on it, and I'm not the only one.

This is one of those things I wish we could just get it over with. I want to see what happens! I've lived this long and I think I deserve to see something interesting happen in my lifetime. Covid was interesting but I don't really want a repeat of that.

[ThePythonicCow]

Posted by petra (here)
This is one of those things I wish we could just get it over with. I want to see what happens! I've lived this long and I think I deserve to see something interesting happen in my lifetime.

You are seeing lots of interesting things. Enjoy the challenge! That's life.

[onawah]

You can at least keep up with the data by following this thread: Geomagnetic Reversals and Ice Ages

Posted by petra (here)
This is what my mom used to say! She said we would be put back into the dark ages.
I'm really starting to see about the dangers of AI. People like me want to rely on it, and I'm not the only one.
This is one of those things I wish we could just get it over with. I want to see what happens! I've lived this long and I think I deserve to see something interesting happen in my lifetime. Covid was interesting but I don't really want a repeat of that.

[bojancan]

Braking Point crew discusses the emergence of AI doomsday hacks across the US.....

Oct, 23
Breaking Points
2.11M subscribers
Sam Altman Issues DIRE WARNING Over Disaster He Unleashed

Source: https://youtube.com/watch?v=2Ze6he-3F74