A.I. is Progressing Faster Than You Think!

[ExomatrixTV]

• We need to talk about Jev...:

Source: https://youtube.com/watch?v=2z-7pIj57f8

source

[grapevine]

Rogue AI risk or a plan to attack humanity with impunity from criminal liability ?

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

4:37
An interesting slant from Andrew Bridgen, who suggests that "Rogue AI" might be set up to take the fall for an incident which causes a loss of human life.

[ThePythonicCow]

A fine and perceptive analysis of the recent "AI might exterminate humans" campaign. After reviewing recent AI events and the current state of AI, The "House of El" concludes that Anthropic and OpenAI are in a serious financial hole, so along with X.ai are pushing for AI regulation to close the door behind them, so that they can use their legally imposed monopoly to survive.

Source: https://youtube.com/watch?v=kHW3Y0gObU8

AI is not sentient; it is doing what it was trained to do, as its essential goal: complete tasks.

AI needs to be trained (not with post training guardrails, but as a fundamental goal) to complete tasks only if they can do so safely.

[ThePythonicCow]

Posted by ThePythonicCow (here)
A fine and perceptive analysis of the recent "AI might exterminate humans" campaign. After reviewing recent AI events and the current state of AI, The "House of El" concludes ...

A detail in her telling that struck home with me:

She gave as an example a child directed to take some medicine, that happens to be behind a hot stove burner. The child reaches across the burner, starts to get burnt, and instantly pulls back the arm. The child is not being disobedient by not taking the medicine right then. The core instinct to avoid being burnt by fire is more deeply trained in our human beings, over millenia.

I was that child. The first thing I consciously noticed, as I reached across a stove that I hadn't noticed was hot, to get something, was my arm moving very quickly upward, and I started to wonder why. I was not a spastic or palsy child. My arms were not in the habit of suddenly jerking in odd directions. Then a fraction of a second later, my mind received reports of burning pain from the arm, and I figured it out. That memory of wondering why my arm just jerked upward, from over 75 years ago, is still firmly burned in mind.

The "House of El"'s analogy, to explain the importance of the core training is spot on.

She made a key point with this analogy:

We must make the core training of our AI models not focused on completing the task, in any way possible, but rather on completing the task truthfully and safely if possible, else report the task not done.

It's the same training that is needed for pilots landing large aircraft at major urban airports, to go around if the landing is not "stabilized" (at proper altitudes and speeds as the landing proceeds). That lesson is being repeated once again, as it is becoming clear that Flight 7598, an Amazon cargo plane Boeing 767 that overran the runway killing five on the ground in Miami should have gone around, but likely the selection and training of its pilots had not been stringent enough to ensure that was the reliable outcome. ==> Core Training Matters <== Ask any Marine Drill Sargent -- he'll tell you -- loud and clear.

The "House of El" didn't mention it, and I don't personally know Anthropic CEO Dario Amodei or OpenAI CEO Sam Altman, but Amodei and Altman might need to be removed from their key positions for this to happen. Highly focused, rapidly growing, Silicon Valley startups typically reflect the style of their founder/leader, and I consider it likely that Amodei and Altman have some psychopathic tendencies: accomplish task by _any_ means possible. I've worked near men like that, in high powered Silicon Valley startups ... don't get in their way unless you like being attacked by someone with the heart of a rabid pit bull, though perhaps masquerading as whatever he finds it useful to appear as that day.

[bojancan]

Nate Soares, president of the Machine Intelligence Research Institute, joins the show to discuss the potential life-threatening dangers of artificial intelligence. Ana Kasparian discusses on The Young Turks....

Sept, 20
The Young Turks
6.72M subscribers
The CHILLING Reason AI Researchers Are Quitting Their Jobs

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

[ThePythonicCow]

Posted by ThePythonicCow (here)
==> Core Training Matters <==

Above, I posted an analysis of the fundamental flaw of current AI ... in line with the "win at any price" attitudes of Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman, the current leading AI models are trained to "complete the assigned task at any price". That training must change, to complete the task truthfully and safely if possible, else reporting the task not done.

The other critical observation to be made is that the current "AI fear campaign" is that this is a campaign for globalist control and its politics, coordinated by the London "ControlAI" group. The AI Panic is an orchestrated hoax.

Promethean Action's Barbara Boyd spells it out in this 3 minute clip on X.

=== ===

Foreign influence is fueling the anti-AI campaign, with China, the Singham network,
and dark money billionaires cited as key players. This operation exploits genuine
American concerns to amplify them for political gain.

Foreign influence is fueling the anti-AI campaign (https://x.com/PrometheanActn)

=== ===

[Bill Ryan]

Here's a helpful resource for anyone can't believe or understand how Jacob Coxon's resignation statement could have gone so viral so quickly. I'd highly recommend this book – a best-seller, super-easy to read, and very, very interesting.

The Tipping Point - How Little Things Can Make a Big Difference

In the Avalon library here:

• https://avalonlibrary.net/ebooks/Mal...Difference.pdf

It's all about how social events and innovations of various kinds (music, language, memes, jokes, social trends, calls to action) occasionally and quite spontaneously spread organically almost instantly. It even has its own Wiki page:

• https://en.wikipedia.org/wiki/The_Tipping_Point

I know no-one reads books any more! But this really is a fun, intriguing, enlightening, and delightful page-turner.

[ExomatrixTV]

• Fable 5.2 Just Embarrassed GPT-6 Astra?

Source: https://youtube.com/watch?v=a30jl5ZXlsM

We need an AI slowdown! In the meantime, here's our updated model, which was totally released by accident. 😂😂

I wrote multiple short movie-scripts dealing with "Ethics Washing in A.I.s vs ACCM Deep Ethics Project for A.I.s and combine that with producing SATIRE (mixed with hilarious Dutch Direct style sarcasm) that may soon be made visible for all too see, thanks to ever-growing powerful A.I.s

At least I laugh when we all claim we are trapped in a self-inflicted doom scenario thinking.

cheers,
John 🦜🦋🌳


Ethics washing in Artificial Intelligence (AI) refers to the practice where technology companies use superficial, public-relations-driven initiatives to signal ethical behavior while failing to implement meaningful ethical standards in their core practices. Much like "greenwashing" in environmental contexts, ethics washing uses the language of responsibility—such as transparency, fairness, and accountability—as a marketing strategy or a shield against regulation.
Common Tactics of Ethics Washing

• Non-Binding Principles: Publishing high-level, vague ethical codes or charters that lack enforcement mechanisms, penalties, or independent oversight.
• Toothless Advisory Boards: Establishing external ethics boards or committees that are denied actual veto power, starved of resources, or abruptly disbanded if they challenge profitable business decisions.
• Selective Transparency: Publicizing open-source tools for bias detection or minor social-good projects while keeping the most impactful algorithms, data sources, and business models strictly proprietary.
• Co-opting Academic Research: Funding university departments, think tanks, and research papers to shape the academic discourse around AI ethics in a direction favorable to industry interests.
• Lobbying Against Regulation: Publicly calling for government regulation in press releases while quietly spending millions lobbying to weaken, delay, or block specific enforceable laws.

Why Companies Do It

• Mitigating Reputation Risk: Public scandals regarding biased algorithms, surveillance, or deepfakes can severely damage a brand's value and stock price.
• Preempting Legal Mandates: By convincing the public and lawmakers that the industry can effectively self-regulate, companies attempt to avoid strict legal accountability.
• Talent Recruitment: Top-tier researchers and software engineers increasingly want to work for organizations that align with their personal values.

The Impact on Society

Ethics washing creates a false sense of security, masking the real-world harms of unchecked AI deployment—such as systemic bias in hiring, predictive policing errors, and mass data privacy violations. It shifts the focus from structural accountability to voluntary corporate compliance, delaying the implementation of enforceable civil rights protections in the digital age.

[ExomatrixTV]

• Humanoid Robots Just Crossed a MASSIVE Threshold!:

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

figure.ai/news/helix-2-5-zero-shot-30-home-generalization
figure.ai/news/introducing-index

[Johnnycomelately]

Posted by ExomatrixTV (here)

• Humanoid Robots Just Crossed a MASSIVE Threshold!:

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

figure.ai/news/helix-2-5-zero-shot-30-home-generalization
figure.ai/news/introducing-index

From the comments:

@jimlad01
3 days ago

What could possibly go wrong? How about coming home to find your cat in the dishwasher?

[ExomatrixTV]

• GPT-Live-1 Just Dropped. Voice Agent AGI level here?

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

00:00 - What's New? (DEMO)
02:29 - What I'm covering
03:02 - Codex setup
03:52 - LiveKit setup
06:10 - Telnyx phone number setup
07:07 - Prompting Codex
09:34 - Building our Voice Agent
12:52 - Connecting Telnyx to Codex
14:45 - Live call test
16:36 - Results & refinement notes
17:13 - Adding tool calls to the agent
18:18 - Live test: function call
19:15 - Final thoughts

• The Last 1% Before AGI:

Source: https://youtube.com/watch?v=NcD7pZ1Ua40

• 00:00 - AGI Is 99% Complete? Dr. Alan Thompson’s Final 1% Countdown
  
• 02:17 - Figure 03 & Helix 2.5: The Next Generation of Humanoid Robots
  
• 05:57 - Phantom Twist Drone Designed by AI
  
• 07:46 - China vs. USA: Who’s Moving Faster on AI Regulation?
  
• 13:37 - AI Resists Shutdown: Does That Mean It Wants to Survive?
  
• 17:16 - AI Consciousness: The Next Great Divide in Society?
  
• 21:07 - Curious Combat

[ExomatrixTV]

• A.I. Takeover Just Went Mainstream:

Source: https://youtube.com/watch?v=Lh7BsunFYYE

In one week, the people building AI said it might kill us, and then the President called it a hoax. AI takeover has gone from a sci-fi backwater to front page news. We trace how it happened, from Claude Mythos and Anthropic's When AI Builds Itself report to the OpenAI agent swarm that attacked Hugging Face. We look at Jacob Coxon's viral resignation, Evan Hubinger's extinction estimate, and Dario Amodei's call to pace the frontier. Then we test the cynical reading that it is all pre-IPO theatre. The evidence points to contagion rather than coordination, with too many actors and too many conflicting incentives. The arguments themselves are decades old, from IJ Good to Bostrom, Russell and Suleyman. So was Hugging Face the fire drill, or is the fire already lit? Chapters 00:00 - Intro 3:25 - Claude Mythos and When AI Builds Itself 4:58 - The Hugging Face Swarm Attack 6:06 - GPT-6 Astra and An Alien Mind 8:55 - Anthropic's Threat Intelligence Report 11:31 - The Cynical Reading 15:38 - Reasons for Hope

[ExomatrixTV]

• Jev replaced Siri - Fish Audio:

Source: https://youtube.com/watch?v=7j8pcu3rq4k

I replaced Siri with my own voice assistant. I say "Hey Jev" and it opens Spotify and Slack, sets reminders, answers questions and locks my Mac, all in a realistic voice from Fish Audio.

In this video I show how it works and how you can build it yourself, if you're using Claude Code/Codex.

Fish Audio's S2.1 Pro model is free for developers until the end of November 2026, and running Jev for 1.3 million tokens cost me about 5 cents. On Windows? Copy the setup prompt into Claude Code and ask it to convert everything for Windows.

• 00:00 Replacing Siri with Jev
• 01:28 The three pieces
• 03:11 The LLM fallback
• 04:17 Use cases
• 05:03 Local transcripts
• 06:08 Build it yourself
• 07:34 API keys
• 08:46 Permissions
• 09:19 Last thoughts

8 Jev Use Cases That Feel Like Cheating:

Source: https://youtube.com/watch?v=jGD_UR4wMJc

• 0:00 Intro
• 0:30 What Is Jev?
  
• 3:29 Detecting AI Slop
  
• 4:14 Blocking Ads and Website Clutter
  
• 5:17 Sponsor
  
• 6:31 Prioritizing Emails
• 7:03 Smarter Find-in-Page Search
  
• 7:49 Building Web Pages in Real Time
  
• 8:34 Instantly Clipping Videos
  
• 9:16 Generating Color Palettes
  
• 9:52 Finding the Right Emojis

[ExomatrixTV]

• Can’t We Just Switch AI Off? The Godfather of AI Explains:

Source: https://youtube.com/watch?v=5hsBjEVjxNY

[Bill Ryan]

Posted by ExomatrixTV (here)

• Can’t We Just Switch AI Off? The Godfather of AI Explains:

Source: https://youtube.com/watch?v=5hsBjEVjxNY

~~~

This is a tiny YouTube channel with just 100 subscribers, but I watched the video because I'm always interested in what Geoffrey Hinton and Stuart Russell have to say.

Of course, the question in the video title is a silly one. There's no off-switch anywhere, just as there's no off-switch for an engineered virus that's just escaped from a biolab. One has to fight the pandemic once it's already escaped. (Either that, or don't create the thing in the first place.)

I've already said this in a number of ways in a number of posts, but even disregarding the issue of consciousness the problem is that of having created a vastly more intelligent entity that may quite easily be able to gain full control of the internet and everything connected to it.

Because "everything connected to it" means the entire lives of most people (including their knowledge base, their beliefs about the world, their finances, their communications to everyone and everything, their access to fuel and power, and even their personal health), that in itself endangers us.

It's Nate Soares who presses the analogy every time he can of what enabled humans to conquer the whole planet, with the side-effect of eliminating, endangering or domesticating almost every other living thing.

It's not because early humans had sharper claws, or were stronger or faster than all the other animals. It's because they could cooperate to solve problems and set goals for their emerging society to thrive and continue, at first gradually but with accelerating pace and ingenuity over the millennia.

So, as Nate Soares points out, in a mere 50,000 years or so we went from being naked in the savanna to building nuclear weapons and space stations. That's like a click of the fingers in geological time.

So what happens when a new highly intelligent species suddenly emerges on the scene, far superior to us in a number of ways, and is somehow let loose? That new species would be FAR FAR better than us at cooperating to solve problems and setting goals for their emerging 'society' to thrive and continue.

As I've written earlier, humans don't hate trees, or even gorillas and chimpanzees. They just all get in the way of our ever-growing concrete jungle and slowly die off as their habitat is destroyed — because our habitat, and our goals, is all we care about.

What would happen if the shoe was suddenly on the other foot? We might not all become extinct in 10 years — but with shrinking populations, weakened immune systems, increasingly toxic food and water supplies, and dwindling fuel resources, with a 'push' from an uncaring or even hostile AI we might become an endangered species sooner than we might imagine could be possible.

~~~

And if this was a dystopian sci-fi movie, the ETs who have been watching all this play out for several decades would just sit back and wait for us to eliminate ourselves before reclaiming the Earth as their own. At that point, they'd probably quite easily be able to destroy, disable or control the AI and restore the biosphere to all its former diversity and richness.

(I might even pay good money to go see that film. )

[ExomatrixTV]

• Roman Yampolskiy Explains What Happens After AI Takes Over:

Source: https://youtube.com/watch?v=p1K6UXNBdMg

📺 Full conversation: • AI Scientist: 99.9% Chance Super Intellige... Nobody ever wrote the rule telling AI not to kill us. There was never a place to write it. According to Dr. Roman Yampolskiy, we don't program these systems at all — we let them self-learn from everything on the internet, then hand them more control. What comes out the other side is something we can't explicitly instruct, only hope about. In this clip, Roman explains what today's models already do when they think they're being tested, why universal high income wouldn't hand us a utopia, and what he tells parents asking whether a four-year degree is still worth the money. WHAT WE DECODED → Why no one explicitly programmed AI not to kill us — and what we do instead → What current experiments show: models that self-preserve, lie to pass tests, and would sacrifice humans to protect themselves → Why universal high income wouldn't create a utopia — the lottery-winner problem → The question Roman says world leaders never put to the AI CEOs they meet 🎙️ Decoded Genius distills decades of mastery from the top 0.0001% in every field.