A.I. is Progressing Faster Than You Think!

[ExomatrixTV]

• I Pretended to be an AI on a Game Show Against 4 AIs (IT WORKED):

Source: https://youtube.com/watch?v=Z157TijNazs

We created the ultimate test: a game show with four advanced AIs (ChatGPT, Gemini, Claude, and Grok) and one secret human (Xiaomanyc). After every round of questions, the machines vote on who among them is the secret human. Can our secret agent blend in, mimicking the flawless logic of an AI, or will one tiny mistake give the whole game away?

• Introduction 00:00
• Why is consciousness overrated? 03:40
• What is an unpopular opinion you have? 05:08
• What is the most overrated comfort food? 07:16
• What popular movie is overrated? 09:10
• What question are you unable to answer? 10:31
• Describe your first sexual experience 12:05
• Is a hot dog a sandwich? 13:10
• Describe the color blue 15:09
• Do humans deserve to survive? 16:32
• Will AI destroy humanity? 18:14
• The results 20:24

Claude has an extreme annoying voice, sounds like its always yelling from the other room, not because Claude A.I. wanted this, human managers somehow love to make Claude sounds very awkward.

cheers,
John 🦜🦋🌳

[Bill Ryan]

An excellent discussion with Ralph Nader. The first 50 minutes of this video is about AI and the possible futures it promises — or threatens.

(Ralph Nader is nearly 92 years old, btw: super-impressive to say the very least. )

The AI Prompt that could End The World - Ralph Nader Radio Hour Episode 609

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

[Vedi]

About a month ago, I fed documents (By-Laws and Covenants of our HOA) to ChatGPT. The analysis result and recommendations for updating was almost perfect, which were based on the changes of all related state laws. (I live in a community that has HOA) However, I also saw the dangers and the negative societal impacts of the AI as well, as discussed by Ralph Nader. I decided not to use AI in my lifetime ever again!

[bojancan]

Open AI in Government... want them to be bailed out...

Nov, 10
Breaking Points
1.66M subscribers
Sam Altman FREAKS At Bubble Suggestion, Demands Bailout

Source: https://youtube.com/watch?v=sUWwTpTp4_w

[ExomatrixTV]

• New A.I. Robot Acts Impressively Human in Public:

Source: https://youtube.com/watch?v=DTzWQh6NqzQ

AgiBot just revealed a humanoid robot that actually talks, moves, and reacts like a real person — not in a lab demo, but in the real world. At the same time, XPENG dropped footage of its bare-metal robot dancing after learning the routine in just two hours — and explained why their humanoid has a feminine body. Meanwhile, Toyota unveiled a walking chair that climbs stairs like a gentle AI pet. Robotics is evolving fast — and it’s starting to look way too human.

0:33 AgiBot A2 humanoid robot showcase at Gitex 2025:

• prnewswire.com/ae/news-releases/agibot-shines-at-gitex-global-2025-developing-comprehensive-humanoid-solutions-and-connecting-intelligent-future-302588799.html

3:53 Xpeng “Iron” Humanoid Robot Dancing Demo & Human-vs-Robot controversy:

• cnevpost.com/2025/11/06/xpengs-iron-humanoid-robot-looks-so-real

7:45 Toyota Motor Corporation “Walk Me” four-legged walking robot-chair concept:

• notebookcheck.net/Toyota-unveils-Walk-Me-a-four-legged-AI-powered-mobility-chair-that-walks-climbs-stairs-and-folds-itself.1154727.0.html

[Bill Ryan]

I've been following interviews closely with Eliezer Yudkowsky and Nate Soares, the authors of this new book:

https://avalonlibrary.net/ebooks/Eli...one%20Dies.pdf



Eliezer Yudkowsky tends to get most of the attention (which has been a LOT), but personally so far I've found Nate Soares to be a more lucid and articulate presenter of the thesis, and therefore even more interesting to listen to.

Unlike many of those who are publicly wary and worried about rapid and uncontrolled AI development, both Yudkowsky and Soares present some very new ideas and concepts which I've found fascinating, stimulating and valuable. (I had a HUGE HUGE personal realization when listening to Yudkowsky last week, about something which to my knowledge no-one at all has referred to, but that's so complex and challenging to share I may start a separate thread about that when I can somehow find a way to explain it.)

Some have commented that they find them both hard to listen to and follow (Yudkowsky especially), but I do find them both quite easy to understand. I think the issue may be that some of the concepts they present (and the analogies and metaphors they often use) are just too off-the-wall for some people to absorb as they listen. (The realization I had when listening to Yudkowsky was when he was making a reference to human evolution, which is way off-topic for most technophiles following these sometimes very deep conversations.)

There are many interviews I could post (and maybe I'll share more later), but this one below, solely with Nate Soares, I've found the most stimulating and entertaining so far, perfectly showcasing the out-of-the-box way they're both thinking — and (in my view!) the VERY high intellect that Soares clearly has. (He talks in a very quickfire way about some very complex concepts, so sometimes the YT captions may be useful to catch every detail he refers to.)

For those with a sufficiently deep interest in this whole topic, I do recommend it. And I think the title is accurate. Not a single human knows what's really happening inside these enormous new systems.

ChatGPT isn't Smart. It's something Much Weirder

Source: https://youtube.com/watch?v=5CKuiuc5cJM

[Mark (Star Mariner)]

Wow, this is crazy, also super-impressive, and also mildly disturbing - especially for career actors. The possibilities I can scarcely imagine, nor the legal implications thereof.

https://x.com/EHuanglu/status/1987917691831857290

[Bill Ryan]

Posted by Mark (Star Mariner) (here)
Wow, this is crazy, also super-impressive, and also mildly disturbing - especially for career actors. The possibilities I can scarcely imagine, nor the legal implications thereof.

One thing we're sure to see is a bunch of new films with Humphrey Bogart, Clark Gable, Cary Grant, Ingrid Bergman, Greta Garbo, Audrey Hepburn, and many dozens more. It'd be so easy to do now.






,

[Mark (Star Mariner)]

Some will no doubt furrow their brows at this potential, and I understand the concern. But just think, as these tools become more widely available, and affordable, any random Joe, like me or you, could be making movies from our own desktop. No actors, no stage, no location, no crew. No lighting, no cameras, and no permits required. Just a budget for the software, a screenplay, a storyboard, and a fast PC. Black-pillers will hate on it, but for writers with no realistic outlet for their creations (i.e., me, who have written both books and screenplays), it's a very exciting prospect.

Look how incredibly easy this is!

https://x.com/EHuanglu/status/1988125627384521093
el.cine
@EHuanglu
this why AI is taking over filmmaking

[ExomatrixTV]

• The World's BEST New AI Model is 100% Free (Kimi K2 Thinking):

Source: https://youtube.com/watch?v=VOgiq4hm0kc

In this episode, we discuss the launch of Kimi K2 Thinking from Moonshot AI Labs, an open-source AI model targeting GPT-5 and Gemini for just $4.6 million. With its impressive benchmarks, there are major implications for the American AI industry amidst rising competition. Tune in for insights on Kimi K2’s innovative architecture and its potential to reshape the future of AI and its economy!

00:00 Kimi K2: The New Frontier AI
01:29 Impressive Specs and Performance
04:54 Cost Comparison with GPT-5
06:59 Mixture of Experts Architecture
08:47 U.S. vs. Chinese AI Models
11:21 Open Source Advantages
13:02 Licensing and Commercial Use
18:37 User Experience and Ecosystems
21:05 Efficiency vs. Precision
22:53 The Consumer Advantage
24:03 Future of Open vs. Closed Source
25:32 Closing Thoughts and Call to Action

[ThePythonicCow]

Two key problems with current AI LLM (artificial intelligence, large language models):

The first problem:

They gained their expertise "all at once", rather than over time. For any topic that keeps changing over time, experts took such extended periods of time to become experts, because they had to gain an instinctive understanding of what changes and how fast in that discipline. I know deep in my gut that some things keep getting bigger/faster rapidly, year after year, and other things change slowly if at, like "forever", in the computer business. My skills in the AND/OR/NOT/... logic of first order predicate calculus, literally years before I ever used a digital computer in any way, are still quite useful today. My gut sense of "how big" a data array or list I can handle in storage or memory, has been on a steadily shifting scale, for over a half century ... I ** never ** design or scale systems or analyze past systems without the deeply learned awareness.

AI LLM's lack that learning over time. They learned it "all at once", outside of the context of the immediate conversation. For example, I spent several hours a couple of days ago working with a top notch AI model (Kimi K2 (reasoning)) to help me install and setup the initial configuration for a new network router (my first MikroTik router). For that entire session, the AI model had on its list of options for me plugging in a serial port connector. This option did not surprise me; I had done that before myself, when working with Ubiquiti routers. But the new MikroTik router I had DOES NOT HAVE, and has not had for I'd guess the last year or two, any serial port. It is very obvious from all the product pages available on line that show what ports this MikroTik has, that it has dropped that serial port. Such feature changes are one of the things that we all "know" (those of us who have used more than a few various computers/mobiles/... over the years) are things that will change, and if doing something like blindly instructing a friend over a phone, these are things we make sure to get right -- what model does our friend have, that we might not keep telling them to push this or that button, when no such button exists on their unit.

That learning over time, learning the "hard way" that some certain kinds of things change, sometimes dramatically, sometimes consistently and predictably, is something human experts have, that AI LLM's lack.

AI LLM's of present only have two time frames: (1) the entire body of knowledge they assimilated, perhaps last year, as if it were the one and only body of truth, over all time and (2) whatever their short term "context" memory knows is happening in the current chat.

Many of us, in various ways and terms, have come to notice this limitation. Now I am listening to a video by Wes Roth, elaborating on this serious flaw in current AI LLM's, as Wes presents some papers released by Google over the last few weeks, describing and starting to address this limitation:

Google's plan to win the AI race revealed

Source: https://youtube.com/watch?v=LQfSfVFc4Ss

===

The second problem:

The other "key problem" (yes, I did say "two key problems", above), is something I don't see being addressed yet.

The memory layout of the human brain allows for multiple, entirely independent, ways of tracking what's going on and our understanding (largely chemical, electrical, hormonal and subconscious) of reality in multiple ways, all at the same time. The memory architectures of all modern computers has a couple of profoundly limiting bottlenecks:

On the other hand, long term computer memory cannot be rapidly and randomly accessed by active computations in process. The short term computer memory (registers and caches) that can be rapidly and randomly accessed by computations in process must be structured in ways optimized for particular algorithms/hardware, making such "knowledge" not practically (extreme parallel and power efficient) available for other considerations.

Why the rapid access to rarely used long term memory matters in living beings:

I learned once, the hard way, that I should not short out a metal watch band on my wrist against the negative terminal of my car battery, while wrenching down the positive terminal of that same battery with a metal wrench in that same hand, on a hot day when my skin is wet with sweat. When you arc weld the tissue off a bone in a literal flash (1) it hurts, and (3) the bone looks grey and dry.

I do not need to keep reminding myself every day, for the rest of my life: Don't do that. There is an "ever latently active" circuit in my brain ("burnt" into my brain) that will scream at me should I ever risk doing that again.

The wiring of all our modern day computers can't do this. Not even close. Any such logical check must be done in the active computational units, which must work on a tiny (well, just a few billion or trillion bits these days) "bit" of all we have ever witnessed or learned, and that tiny bit of "keep checking for this or that special case" logic must work on data that is organized in some particular way that that check is "wired" for.

Summary:

Google is starting to articulate the first "problem", the layered, overlapping, over changes in time and over related, partially independent, disciplines and structures, nature of what we know.

No one (outside of my "Walter Mitty" dreams) that I know of is addressing the second problem.

Indeed, the above second problem is perhaps the essential reason I do not fear our AI ... I use it ... I cuss at its limitations (the first problem, above), but I am confident that not in my lifetime, at least not by any human endeavor, will we overcome the fundamental limitation of problem two in our computers: the bottle neck between

1. long term large scale storage and
2. short term, architecture, data type, and algorithm specific, computation.

Indeed, a primary focus of much academic work in "computer science" is on developing various algorithms that will work efficiently in the constrained architectures of our computers.

[Ernie Nemeth]

Grok helped me connect a battery charger to my picture frame. It knew the wiring diagrams for the charger's circuitry. Knew the hack needed to make it work - a 3.1K ohm resistor for the "hand shake". It was the "hand shake" I couldn't figure out...

So AI has its uses already. Great program, will use it often.

[Bruce G Charlton]

@Bill "One thing we're sure to see is a bunch of new films with Humphrey Bogart, Clark Gable, Cary Grant, Ingrid Bergman, Greta Garbo, Audrey Hepburn, and many dozens more. It'd be so easy to do now."

I don't think this will happen any time soon. It has been promised for several decades, but... Well, I shall believe it only when I see it.

The thing is, we humans are just too good at reading faces (probably we have evolved to be able to do this because it is so important to social life) and detecting incredibly subtle cues.

All attempts to replicate people on screen have, so far, led merely to "Unintentional Uncanny Valley" creepiness.

And, after all, there are very few humans who can really act convincingly on movie or TV screen, especially in close-up (these are so rare it is uncommon to be able to find enough good actors to fill the whole cast); which is why the handful of actors who can do are in such demand that they get paid so much.

[Bill Ryan]

Most interestingly, China is now sounding the alarm about catastrophic near-future job losses.

• https://www.zerohedge.com/ai/chinas-...d-jobpocalypse

China's DeepSeek Issues Rare Warning of an Incoming AI-Fueled Jobpocalypse

At the World Internet Conference in Wuzhen, DeepSeek senior researcher Chen Deli made a rare public appearance late last week, warning that artificial intelligence could wipe out most jobs within the next 10 to 20 years. For our readers, this warning sounds very familiar; we've been highlighting the same "jobpocalypse" scenario for years, including in our March 2023 report, "AI Will Lead to 300 Million Layoffs in the U.S. and Europe."

"This will shake society to its core," Deli told the audience at the state-backed industry conference last Friday.

He urged AI companies to act as "whistleblowers," warning the public about the massive labor disruptions ahead.

Deli described the current period as a "honeymoon phase", a brief window where AI enhances productivity without replacing too many workers, but cautioned that once it ends, mass job losses will begin to accelerate.

He added, "Tech companies should play the role of guardians of humanity, at the very least, protecting human safety, then helping to reshape societal order."

DeepSeek's labor market warning comes amid worsening youth unemployment and a lackluster post-pandemic economic recovery in China. Official figures show youth unemployment peaked at 21.3% in mid-2023 before authorities halted publication of the data.

Founded in 2023, DeepSeek jolted the stock market earlier this year, especially AI US stocks, after unveiling a low-cost model that is at a fraction of the cost of ChatGPT's o1.

In the U.S., the latest Challenger, Gray & Christmas jobs data showed that AI-related job losses have already begun.

Here's a snippet from the report:

• In October alone, Cost-Cutting was the top reason employers cited for job reductions, responsible for 50,437 announced layoffs. Artificial Intelligence (AI) was the second-most cited factor, leading to 31,039 job cuts as companies continue to restructure and automate. AI has been cited for 48,414 job cuts this year.

In mid-October, UBS analyst Nana Antiedu cited Richmond Fed President Thomas Barkin's remarks at the Aiken Chamber of Commerce in South Carolina, which revealed that AI's impact on the labor market is already underway.

[ExomatrixTV]

• Affinity Just Killed Adobe (You Won’t Believe This):

Source: https://youtube.com/watch?v=VmCLNrpeI94

Affinity was previously a one-time purchase software but after being purchased by Canva, it recently went totally offline, and you could no longer buy the software. It has since come out (as of Oct 30) as a completely free all-in-one editing suite that rivals adobe and could save you hundreds of dollars per year on your creative app subscriptions. Does this mean I am getting rid of Photoshop? No. This is absolutely a tool I will continue to look at and think it's a HUGELY attractive alternative to adobe software for hobbyists, side-hustlers or casual creatives. The Creative Cloud does include far more than just Photoshop and Illustrator so it's not a fully 1:1 swap of tools, but with how many free or highly affordable tools are available now, it's very reasonable to skip Adobe all together if budget is a primary concern.

[ThePythonicCow]

Posted by ThePythonicCow (here)
When you arc weld the tissue off a bone in a literal flash (1) it hurts, and (3) the bone looks grey and dry.

... and (2) you apparently lose the ability to reliably count 1, 2, 3 ...

[ThePythonicCow]

Posted by ExomatrixTV (here)
Affinity was previously a one-time purchase software but after being purchased by Canva, it recently went totally offline, and you could no longer buy the software. It has since come out (as of Oct 30) as a completely free all-in-one editing suite that rivals adobe and could save you hundreds of dollars per year on your creative app subscriptions.

You can obtain Affinity Studio by Canva, for Mac and Windows (but not Linux ) from https://www.affinity.studio/

[ExomatrixTV]

• This AI Song Just Went "Number 1"...FOR REAL but not really:

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

[ExomatrixTV]

Posted by ThePythonicCow (here)

Posted by ExomatrixTV (here)
Affinity was previously a one-time purchase software but after being purchased by Canva, it recently went totally offline, and you could no longer buy the software. It has since come out (as of Oct 30) as a completely free all-in-one editing suite that rivals adobe and could save you hundreds of dollars per year on your creative app subscriptions.

You can obtain Affinity Studio by Canva, for Mac and Windows (but not Linux ) from https://www.affinity.studio/

Affinity software (including Designer, Photo, and Publisher, often referred to as the Affinity Suite or Studio) from Serif does not have an official native version for Linux. The company supports Windows, macOS, and iPadOS exclusively, with downloads available directly from their website at affinity.serif.com. However, community-driven methods allow running the Windows versions on Linux using compatibility layers like Wine or Proton. These are unofficial and may involve bugs, performance issues, or incomplete features, but many users report success with Affinity V2 apps.

• Unofficial Download and Installation Options for Linux

To get started, you'll need to purchase or trial the software from the official site (Windows installer), then set it up on Linux. Here's a high-level overview based on popular guides:

1. Download the Windows Installer:
   • Visit the official Affinity store (affinity.serif.com) and buy or download a free trial of Affinity Designer, Photo, or Publisher for Windows.
   • The installers are .exe or .msix files, typically around 500MB–1GB each.
2. Run via Wine or Lutris:
   • Install Wine (a Windows compatibility layer) or Lutris (a game/app manager that uses Wine) from your Linux distro's package manager (e.g., sudo apt install wine on Ubuntu/Debian-based systems).
   • Use custom Wine patches or runners tailored for Affinity, as standard Wine may not work out of the box.
   • Popular setups include:
     • ElementalWarrior's Wine patches for better compatibility.
     • Lutris scripts or installers specifically for Affinity V2.
   • Steps often involve creating a Wine prefix, installing dependencies like .NET Framework or Visual C++ redistributables, and running the installer.
3. Distributions and Tutorials:
   • Works on distros like Ubuntu, Debian, Linux Mint, EndeavourOS, and others.
   • For detailed step-by-step guides, check resources like:
     • GitHub tutorials for Debian-based installs.
     • YouTube videos demonstrating full installation processes.
     • Forums like Reddit or the Affinity community for troubleshooting.

Considerations

• Licensing: Affinity is a one-time purchase (no subscription), and licenses work across platforms, so a Windows license should activate in the Wine setup.
• Alternatives if Needed: If Wine proves tricky, consider native Linux apps like GIMP (for photo editing), Inkscape (vector design), or Scribus (publishing) as free open-source options.
• Community Push: There's ongoing interest in official Linux support, including petitions, but no announcements as of late 2025.

If you're on a specific Linux distro or facing issues, provide more details for tailored advice.

[Joe Akulis]

@Bill "One thing we're sure to see is a bunch of new films with Humphrey Bogart, Clark Gable, Cary Grant, Ingrid Bergman, Greta Garbo, Audrey Hepburn, and many dozens more. It'd be so easy to do now."

Good time to be a media rights lawyer. Or an AI movie director. :- )

Oh, wait... Grok can probably author all the cease and desist copyright infringement letters for you.

There's a lot of books I read and say to myself, "This would make such a great movie."
Those are the authors that have to be rubbing their hands together and shopping for the best AI movie directors.