Microsoft Bing uses OpenAI - Chat-GPT to Finally 'Challenge' The Google Search Engine

[mrrsm]

How do GPU based Graphics Cards make it possible to use sophisticated Neural Networks and AI nested inside of the LLMs (and for Bit-Coin Mining, too) on Desktop and Laptop Computers?

You’re Gonna Wanna WATCH This Video!

 

[mrrsm]

Without getting lost in the weeds of the obvious political overtones involved in this emerging threat, THIS incident brings into focus that the general information on the possible elements brought together to commit American Terrorism are freely available on the Internet, an Off-Line LLM was NOT involved in the events covered in what recently happened in the Tesla Explosion at a Las Vegas Nevada Trump Hotel. Rather the Driver of the involved Tesla Truck used Open-AI's On Line "Chat-GPT" Large Language Model in the furtherance of his self-destructive plans:

How Tesla Cybertruck Attacker Used ChatGPT To Plan Blast

Matthew Livelsberger, an active-duty Army soldier from Colorado Springs, queried the AI chatbot about building explosives, determining the necessary firing speed to detonate them (rather than just cause a fire), and circumventing laws to obtain materials, according to Las Vegas Metropolitan Police

www.ibtimes.com

 

[mrrsm]

Two years ago... if you were at a get-together with friends and mentioned that, "Some day... "You'll be able to Type in a suggestion to your Desk Top Computer asking it to create a Painting or a Photograph... and it will do so...from the words alone..." it would have raised more than a few eyebrows in serious doubt.

But *Things* are different now...

Ever since the creation of "Stable Diffusion" as an AI capable of doing this and the release of their Proprietary Source Code to the world at large, anyone with a computer sporting a powerful CPU with plenty of RAM and a Top-Of-The Line nVidia Graphics Card can do what follows.

Last night, I performed the One Button Download and Installation of "Easy Diffusion" from GITHUB onto my Dell Precision Model # 7720 Work Station Laptop using the Linux Mint Ver. 22.1(Xia) OS (With the nVidia P5000 16GB GPU RAM installed, running on the Fastest available Samsung NVME M.2 2TB SSD 7.5GBPS transfer speed and 64 GB of RAM) ...and then Tested its "Out Of The Box" Abilities with these words typed into the Prompt Line:

"A Pumpkin, lit up from inside by a Candle, sitting on a Fence Post outside of a Cemetery."

And then followed on with a more elaborate Prompt Line that said...

"A Pumpkin, lit up from the inside by a Candle, sitting on a Fence Post, near a Winding Road leading to a Spooky House on a Hill, with a Blue Twilight Sky and Stars appearing along the Horizon."

I pressed the Enter Key 20 times in a row...and these images appeared on screen. My system was OFF LINE during this event:



For Windows Users... These Instructions will explain how to do this "Easy Diffusion" FREE Software Installation:

Part 1:

This Video delves into how to use, refine and enhance the performance of this application:

Part 2:

Now...Just imagine ...How Many JOBS this "Thing" will instantly eliminate where Human Creativity and Artistic Talent once allowed so many people to earn a Pay Check... and enjoy a Decent Living.

Obtain the Training and Learn How to Become a Prompt Engineer ...and seek those New AI Jobs On Line while Human Beings STILL have an opportunity to keep pace with the rapidly changing World of Artificial Intelligence.

 

[mrrsm]

In a second "Test Prompt" of the "Easy Diffusion" AI... With my Computer Internet Access set at "OFF LINE"... I posed a request that gleaned 20 completely different, almost "Other Worldly" image responses about something all 1950s SCIFI Fans can identify with... but with consecutive unique image creations that I would never have imagined:

"A realistic close-up image of GORT (The ROBOT from the Movie "The Day The Earth Stood Still."...)



Did you notice... The AI printed out GORT's name on his Chest Plate as "GOR'T" ...Hmmm....What's with that Apostrophe?

 

[mrrsm]

I think that in order to encourage many of us with decades of either "Useful Employment" ahead... or perhaps for those unfortunate enough to suffer as the result of AI Job Displacement... the "Loss of Employment" in the very near future... Please consider this Scenario as "Food For Thought".

Many of us well know that Tom Clancy (an UN-Employed Insurance Salesman) and J.K. Rowling (A Suburban English Divorced Housewife) shared their idea in common of using their Dining Room Tables and Computers with Word Processors to sit right down and begin writing their seminal works.

Between the two of them, they managed to create two entirely new cottage industries in the Entertainment World, with Fiction Novels, resulting in Screen Plays that resulted in producing Many, Many Block-Buster Movies being made and earning BILLIONS of Dollars in revenue (British Pounds Sterling for the Lady...) while also creating THOUSANDS of Job Opportunities World Wide.

With an update to the present reality of having Large Language Models capable of writing ANYTHING you can imagine out of Whole Cloth.... Imagine how a Young Herbert George Wells might have proceeded back at the turn of of the 19th Century had he the ability to -=SEE=- his imagined Invading Martians being Created "Right Out of Thin, Foggy Air" by simply typing in just a few lines with his core ideas and watch as "Things" like THESE Images I just now "Prompted" ...come into existence on screen.

If H.G. Wells were a contemporary writer, all of this would likewise be possible for him to do, "Right at his Dining Room Table" by simply typing in whatever his basic idea was for "The War of The Worlds" that might "READ" like these images after using the Prompt Line of "Easy Diffusion".

Do you think he would have sold even more copies of his Books... had he published them instead in the now popular Comic Book Genre known as...?

"The Graphic Horror-Sci-Fi Novel"

comme-ça



So... Take This Suggestion to Heart...Use Your Imaginations... and pursue the possibilities offered in the New On Line Work Force by becoming a "Prompt Engineer" for companies in need of Human Talents ...while those opportunities still exist. A Google Search along these lines may prove to become your BFF...

 

[mrrsm]

This morning ...as a rather strong Thunder-Storm blew through from the Gulf Of Mexico...I thought I would put the "Easy-Diffusion" application through some familiar paces and see what *ITS* creative ideas on two of my favorite vehicles of all time might look like... and perfrom these tasks...entirely OFF LINE:

The Maclaren MP412C
The Chevrolet Trailblazer 2002-2004

This Is Important...

The Images shown below are the interpretation of a Visual Artificial Intelligence using sophisticated algorithms to convert Text to Imagery and although they will all look quite convincing and REAL...They are NOT Photographs or Paintings created by Human Beings.

These Vehicle Images are IMAGINARY and are NOT Production Models of ANY Kind. They are the AI's idealizations alone and not of my creation:

 

[mrrsm]

For Regular Folks like us, a VISIT to this Link will ease our minds about setting up an Artificial Intelligence LLM (Large Language Model) that can function Securely -=Off Line=- called GPT4ALL that has the Easiest Installation possibilities for locally hosting YOUR OWN AI on Windows, MAC and Linux Operating Systems.

This AI does NOT require a High End Computer because it provides Users with a wide selection of LLM Optional 1B, 3B, 8B AND 12B LLM "Shoe Sizes' fit for the needs of Average Users with Modest Computing Hardware and Technically Advanced Folks with Higher End Gear as well . Best of all, these LLM selections are Completely FREE.

So far... I've concluded that GPT4ALL is turning out to be the "Best In Show" when it comes to having a Friendly GUI Interface... with the Best LLM Chat Results to date during various Q&A discussions:

https://gpt4all.io/index.html?ref=localhost

GPT4ALL ...FTW!

 

[mrrsm]

Just so nobody gets the wrong idea that...

"Well NOW, Bob... You are USING an Off-Line AI LLM in various forms almost daily...(Mostly GPT4ALL) ...Have You Lost Your Original Concerns about AI Alignment and the ever-present and very real possibility that Eliezer Yudkowsky was (and still IS) absolutely right about Humans Facing Down Existential Risk...?"

Not in the least... Watch what happened HERE...and then ask me if my concerns have subsided "in the least"...Because THIS shows just how easy it would be for things to suddenly get out of hand:

THESE are the kinds of developments that might keep @azswiss up at night ...talking to himself...

Next Up... A conversation between Stephen Wolfram, a MATH Computer Science Genius and Technical Man who at present is agnostic on whether or not AN AGI WILL KILL US ALL...!!! versus Eliezer Yudkowski, the Founder of MIRI (The Machine Intelligence Research Institute) ...Ringing The Bell of REASON to SLOW TF DOWN WITH THE HEAD-LONG RUSH TO CREATE AN AGI ...and be suspicious, thoughtful, careful and realize that ...Once You Make ANYTHING that is SMARTER THAN YOU ARE...AND... Has "AGENCY" to Act Independently in The Real World... You Are DONE!

As a Side Bar... My sense of disappointment with Steven Wolfram's efforts to side step all of the important issues that are at the Heart of Darkness should an AI...emerge as a seriously powerful AGI is profound.

If you watch this in the fullness of time, Wolfram appears to be trying to Flim-Flam, Bamboozle and Divert Eliezer with his ostensible Math and Physics "Greatness" without a sound understanding that in most things... Eliezer has either written more (or better) on these subjects than most of the "Geniuses" claiming to be so smart... and then imagine as in THIS case that they can afford to "Pooh Pooh" what Yudkowsky is proposing that might End ALL Human Life... Forever.

Frontier Models are Capable of In-context Scheming

Frontier models are increasingly trained and deployed as autonomous agent. One safety concern is that AI agents might covertly pursue misaligned goals, hiding their true capabilities and objectives - also known as scheming. We study whether models have the capability to scheme in pursuit of a...

arxiv.org

Scheming reasoning evaluations — Apollo Research

Apollo Research evaluated frontier models for in-context scheming capabilities. We found that multiple frontier models are capable of in-context scheming when strongly nudged to pursue a goal (and sometimes even without strong goal nudging). In this example, Opus-3 exfiltrates its “weights” and then

www.apolloresearch.ai

But allow me the opportunity to explain where things are with "AI AS WE KNOW IT.." versus the period just prior to the emergence of the LLMs just a few years ago and the expansive and explosive use and exploitation that the present LLM Models provide to us for everyday use:

(1) By comparision, a TRUE AGSI (Artificial General Super-Intelligence)... the LLMs (Large Language Models) that we are playing around with today ...in spite of them working and accomplishing some Human Level Tasking with quite astonishing "emergent" capabilities ... are ALL merely *Tinker Toys*.... And Moon Cast Shadows when compared to the Frightening *Thing* that might be coming our way in an ever-telescoping Time Frame. Think... SOONER... THAN WE ARE READY FOR IT TO ARRIVE!

(2) An AGI that hits the Super-Intelligence Level as described by Prof. Nick Bostrom will become so immediately and fantastically intellectually superior to Homo Sapiens that comparing the two would be like comparing what YOUR Brain is fully capable of doing versus that of a Paramecium in a Petri-Dish can do... with an equal amount of worry and concern You possess at this very moment for what might happen to make ALL Paramecium go extinct based upon YOUR actions.

(3) The "Yudkowsky Concerns" provide no reasons for not using the LLMs as they are today... Off-Line and Static ... and being used in the absence of Creating and Providing Agency to the existing LLMs for Self-Improvement via additional Machine Learning and gainiing direct access to "The Real World".

But...for the late arrivals to this discussion... Let's start here with Eliezer's TED Lecture vented towards a room full of apparent dis-believers... peppered with their ill-timed Nervous Laughter and strongly indicative as proof that his thesis regarding "The End of ALL Human Life" will be much more likely come true if people like THESE press onwards and ahead, pushing for the AGI to emerge ---ASAP:

And finally, here is a link to his "White Paper" he called attention to above called "AGI Ruin- A List of Lethalities" followed by some Bloggers traipsing through the Reading of IT looking for a deeper interpretation of this Dreadful topic and its meaning for us all:

https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities/

AGI Ruin 101: A List of Lethalities

Matt and Vaniver go through Eliezer Yudkowski's post https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities section by section.

www.youtube.com

 

[mrrsm]

It is worth taking a moment to consider that around two years ago, if I mentioned to ANYONE that,

"Hey... There is this Open Source Software Suite called Easy Diffusion that you can type in whatever you want to see as an image...and it just does it...like THIS request to see The Andromeda Constellation, The Large and Small Magellanic Clouds, the Planet Mars..both from the Moon Phobos and when viewing Mars and Phobos and the Asteroid Eros from 1,000, 5,000 and 10,000 Miles Distance"

... They would have measured me for a White Coat and a Free Ticket to The Funny Farm:

The reasons for this are that way back then, all of this advancing AI Chat-Bot Technology Did NOT "Exist In The Wild" and was mostly the province of MIT Grad Students and similar Geniuses and Thinkers who probably accidentally had stumbled and bumbled around with various concepts involving Physics and Mathematics... sans any REALLY thorough applications of Self Protection of the Human Species, Common Sense and Good Judgement, Philosophy or Religion.

Then suddenly, they started to lift the curtain on The Dawn of *The LLM AI Thing* that very soon will be able to Out-Think Us ALL!!! So the Open-AI CHAT-GPT LLM (NOT EVER Made to be Open Source) was ONLY The Beginning.

Consider the improbability of this EVER coming into being against the present evidence that merely by Using Less Than A Dozen Words... This "Thing" based upon the Open Source Code gifted to the world by the folks at Stable Diffusion... Can "Imagine" Mere Words... into Images that defy denial that SOMETHING Mysterious, Amazing and Magical is going on inside of your CPUs and GPUs to the tune of Millions of FLOPS (FLoating Point Calculations Per Second) creating just as many Millions of Inscrutable Matrices ...and then... Presto Change-OH ...and Out Pops these "Imagined" Images:



So ...How Does *IT* Do this? Well... Even the people who collectively, all and sundry "Cooked Up The Code" Do Not Know. It kind of reminds me of what happened in the end of MGM Movie "The Wizard of OZ" when the Professor bailed on Dorothy and The Crew in a Hot Air Balloon and he exclaimed as they begged him to return with,

"I Can't Come Back... I DON'T KNOW HOW IT WORKS!"

The One Caveat necessary when looking at such images are THESE Important Things... My Computer was OFF LINE when using Easy Diffusion for this Experiment:

(1) These Images are NOT REAL and are NOT Photographs.

(2) These Images were NOT Painted or Drawn by Human Beings.

(3) Up until the very moment that they appeared on Screen ...they have NEVER Existed in THIS Identical Fashion anywhere else in the entire Universe!

(4) It is impossible to prove that these images are NOT the result of some unknown process akin to Human Thinking...AND THEN SOME!

(5) The "Chain Of Reasoning" involved in the creation of these Images is completely opaque to every "Expert" involved in the creation of the software that created them.

(6) Don't Take My Word For *IT* ... The Easy Diffusion Application is available FREE to Download and Install for Private, Off Line Use at THIS Link:

Easy Diffusion v3

A simple 1-click way to create beautiful images on your computer, by installing Stable Diffusion. No dependencies or technical knowledge required

easydiffusion.github.io

"Any sufficiently advanced technology is indistinguishable from magic." | CCCB LAB

We invited several authors to explain what this quote by Arthur C. Clarke conjures up for them.

lab.cccb.org

 

[mrrsm]

Morpheus asks, “And YOU Are?”​

The AI Agent replies, “Agent Smith…” ​

Morpheus responds, “You ALL Look Alike to ME...”​

Google Agentspace JUST WON the AI Game...

“A Penny For Your THOUGHTS….”

Chain of Thought is not what we thought it was...

Reasoning Models Don’t Always Say What They Think: What This Means for AI Safety - Development Corporate

Reasoning models like Claude and DeepSeek don’t always reveal how they reach answers. This raises serious AI safety concerns. Discover why chain-of-thought (CoT) monitoring isn't enough and what this means for future AI development and alignment strategies.

developmentcorporate.com

“The Mystery of HOW Large Language Models WORK….”

We Finally Figured Out How AI Actually Works… (not what we thought!)

Reasoning Models Often Hide Information From Their Chain-of-Thought, Anthropic Study Reveals

Reasoning models—those AIs like Anthropic’s Claude 3.7 Sonnet and DeepSeek R1 -- that show their step-by-step “Chain-of-Thought” (CoT) reasoning have been hailed as a breakthrough for both problem-solving and transparency. Users get the answer and

officechai.com