Owes ada fek news buruk for mining!

PETALING JAYA: A Sabah minister has slammed a news report linking him to the mining scandal in the state, saying it is both malicious and inaccurate.

Arifin Arif, the state’s science, technology and innovation minister, now plans to drag those who slandered him to court, Sabah Media, a Kota Kinabalu-based portal reported.

Be careful guys!

Will You Get in Trouble for Not Declaring CNY Angpows from Your Boss to LHDN? Here's All You Need t...
From: world of buzz

OpenAl's DeepSeek Response: Catch Us If You Can

US Quietly Arms Key Treaty Ally Near China

Scientists discover black holes spinning unexpectedly fast:
'You're essentially
looking at its fossil rec...

China Strikes Back at Trump's Tariffs:
Bitcoin Plummets

Bang On The Buck! India Tests New Air Defense System With 'Exceptional' Capabilities To Hunt Low-Flying Jets...

Saudi doesn't need any American protection. The only threat that concerns the Saudi's is from America. Saudi has a lot of money invested in American assets, several American military bases on its territory and has spent a lot of money on American military gear. MBS doesn't want to end up like Saddam or Gaddafi who both planned to stop using USD.
Once America enters financial collapse and Russia has destroyed much of American missile inventory in Ukraine, and Saudi has reduced its holdings of US assets, it will be safer for Saudi to join BRICS officially.

In my opinion, I think we will not get any rate cuts until something financially or psychologically breaks in the market. I believe the fed is trying to give it's self room to act when it's appropriate to, however, wall street is trying to bully the fed into a corner. Wall street wants lower rates to boost market valuations at the cost of higher inflation for the average middle class consumer. As how unpopular my opinion may be, I think the market and the economy needs to go through a recession to reset everyones sky high expectations and fix the psychological around the inflation argument.for me Bitcoin is the ultimate defence against a tyrannical government.

Canada needs to develop other markets to offset our dependence on the U.S. The American empire is on the decline so we have to unhitch our wagon from them.

The Canadian dollar fell immdiately after the news about tariffs came out and is expected to fall by up to 20% in the coming weeks or months. This means the net increase for the American importers is only 5% to 10% which may be absorbed resulting in negligble increase for the American consumers. US government gains 25% in tarrifs, importers win in terms of currency exchange and minimum impact to the American consumers.
Canada better don't drag this out or retaliate with tarrifs as they can't win.
The above are the facts compared to this whole speculative discussion.

The economic crisis and downturn are all the signs of 2008 market crash 2.0, so my question is do I still save in the US dollar or is it okay to move all emergency and savings to precious metals?

In my opinion, the USA is on the brink of losing her number one position and Trump's actions are accelerating the process. He is dreaming to live in medivial times but the world has changed now.

Unipolar World Has Come to an End, U.S. Secretary of State Marco Rubio Admits
The emergence of China and Russia mas great powers has transformed the world order.

Lastly, the words of Epictetus should always be kept in mind: “We have two ears and one mouth so that we can listen twice as much as we speak.”

That fits humans, but we might wonder whether AI will abide by the same precept.

It is timely to reflect on some famous quotes that pertain here.

Pythagoras legendarily said this: “Do not say a little in many words but a great deal in a few.” You could suggest that this is what AI-to-AI communication is focused on. You see, a human adage of long ago provides an explanation for a seeming modern mystery.

My explanation for how AI-to-AI language formulation can occur is but one of several means by which this can happen. I wanted to share a handily straightforward path that is readily understandable and commonly arises. More obtuse pathways are certainly possible.

One other notable aspect to mention is that this emergence of AI-to-AI languages is not a new realization. This goes back to the early days of the AI field. There have been moments when the topic suddenly flares up. For example, during the time period of around 2015-2017, there was some excitement and dismay that reached the public sphere about chatbots talking with each other and the possibility that they were plotting to overtake humans.

History tends to repeat itself.

Some final remarks are in order.

If we do manage to achieve artificial general intelligence (AGI), or the conjectured artificial superintelligence (ASI), it would seem nearly impossible to prevent AI-to-AI language formulation. The chances are that the AGI or ASI would try to do so, regardless of whatever we say about it.

Should humans categorically disallow AI from ever concocting a new language?

In theory, it could be a law or regulation that all AI must never be allowed to derive a new language and must always and only use a conventional natural language. Those kinds of stipulations would presumably fall on the shoulders of the AI makers and AI developers. Whether this could realistically be fully complied with is a big question, as I have addressed at the link here.

Humans could certainly come up with an AI-to-AI language and seemingly require all AI to make use of said language. This might provide relief that at least humans would know what the language consists of. That being said, depending on the latitude allowed for the AIs, there is still a chance that a different language would be formulated, and we might not realize that the given AI-to-AI language is only being used some of the time or perhaps when asked to show what the language usage is.

The devised AI-to-AI language won’t necessarily be the same for other AI-to-AI circumstances. Each circumstance might give rise to a tailored language. If someday all AI systems are somehow interconnected, perhaps a grand AI-to-AI language might be formulated, but that’s a bit futuristic and speculative.

Insights Gleaned From AI-To-AI

Let’s cover a few additional thoughts on this weighty topic.

If you hooked together two AIs, they might formulate an in-common language that is devised due to seeking an optimal computational means of data sharing. The language so formed might not be easily recognizable by the human eye. You could presumably have the AIs do a translation into conventional natural language so that a human would be able to make sense of what is being conveyed.

What in the heck is ““Zil: Torna, reda-clok, feln-zar?”

In fact, you might get highly suspicious and suspect that AI is plotting to take over humankind. Maybe it is a secret code that tells the other AI to go ahead and get ready to enslave humanity. Those sneaky AI have found a means to hide their true intentions.

But it turns out to be the first line of telling another AI about Little Red Riding Hood.

Boom, drop the mic.

I want you to pretend that you hadn’t seen the first two lines and that all you saw was the last one, namely this one:

Line 1 in new language – Alpha Generative AI: “Zil: Torna, reda-clok, feln-zar?”
If that was the only aspect you saw, and you didn’t know anything else about what I’ve discussed so far in this elucidation, you would swear that for sure the AI has concocted a new language. You would have absolutely no idea what the sentence means.

Ease Of Language Transformation

Here then are the first lines for each of the three iterations that the two AIs had on the sharing of the famous tale:

Line 1 in regular English -- Alpha Generative AI: “Let’s begin. There is a girl wearing a red hood. Do you know her task?”
Line 1 in quasi-English -- Alpha Generative AI: “Start: Girl, red hood, task set?”
Line 1 in new language – Alpha Generative AI: “Zil: Torna, reda-clok, feln-zar?”

This language the two AIs are using looks quite strange, but it is merely the next iteration of the two AIs computationally reaching a more efficient or optimized way to communicate. Note that I restricted this to using regular letters of the alphabet. Had I also allowed the AIs to use various special characters, the language would have looked even stranger since it would have tons of oddball characters included.

The good news is that the numbered lines still correspond to the two earlier versions of their chatter, namely that the seven lines match the original conventional natural language version and align with the compressed natural language version. It is still the same thing being conveyed.

Here we again will have them chat about the same Little Red Riding Hood tale after having further transformed their AI-to-AI language formulation.

(1) Alpha Generative AI: “Zil: Torna, reda-clok, feln-zar?”
(2) Beta Generative AI response: “Feln.”
(3) Alpha Generative AI: “Mura-ket, lora-tharn, vektu?”
(4) Beta Generative AI response: “Vektu-jal, groven-seth?”
(5) Alpha Generative AI: “Jal, groven-mez, ask-tern?”
(6) Beta Generative AI response: “Torna-giv, groven-plen?”
(7) Alpha Generative AI response: “Jal, groven-flek, gorr-dan firn.”
Yikes, you might be puzzled by the bizarre language being used.

Optimizing Keeps Optimizing

Let’s allow the two AIs to do even more sharing of data and do so into the millions and billions of interactions. That won’t take long since this is all happening electronically. If a human told another human that many stories and words, it would take days, weeks, months, etc. In the case of two AI that are connected and running on fast servers, the time required can be in seconds, minutes, or hours.

This shorthand notation was derived amongst the two AIs. The pattern-matching of the zillions of interactions was being driven mathematically and computationally toward having greater efficiency.

You can readily discern that they are using a form of devised shorthand, telling each other roughly the same things that appeared in the earlier interaction. They have compressed the storytelling by using selected keywords. The logical steps are still shown, though even that could potentially be compressed.

(1) Alpha Generative AI: “Start: Girl, red hood, task set?”
(2) Beta Generative AI response: “Task tell.”
(3) Alpha Generative AI: “Deliver food, path forest, danger?”
(4) Beta Generative AI response: “Confirm danger, wolf present?”
(5) Alpha Generative AI: “Yes, wolf meets girl, ask goal?”
(6) Beta Generative AI response: “Girl tells, wolf plans trick?
(7) Alpha Generative AI: “Yes, wolf fast.”

Two AIs though have no such limiting condition. They can come up with a shorthand version of natural language and use that to convey aspects to each other.

Here’s what they came up with regarding telling the Little Red Riding Hood tale, which I show via numbered lines so that you can match it back to the above English-oriented natural language version that I previously showed you.

Optimizing Toward More Efficiency
Envision that I let the two AIs communicate on lots and lots of topics. Thousands of topics, and thousands if not millions of words are exchanged back and forth as they chitchat. Conventional languages are a bit sloppy and contain redundancy and duplication, but we live with them as humans since that’s what we have come to learn and use. It is hard to wave a magic wand and suddenly anoint a new more efficient language as the language of the land.

Note that it just so happens that both of the AIs already are familiar with the tale because they both scanned instances of the story when the initial data training was taking place. As such, the chatter is not one-sided of an AI telling the other AI about something the other AI had never encountered before. Both AIs have already somewhat pattern-matched on the famous tale.

Let’s move ahead and see what happens if the AI-to-AI carries on a whole bunch of interactions.

Here is a snippet of the first round as they begin to chat back and forth.

(1) Alpha Generative AI: “Let’s begin. There is a girl wearing a red hood. Do you know her task?”
(2) Beta Generative AI response: “Tell me what the task is.”
(3) Alpha Generative AI: “She is delivering food and must travel through the forest. Is there danger?”
(4) Beta Generative AI response: “Yes, danger is confirmed. Is there a wolf?”
(5) Alpha Generative AI: “Yes, the wolf meets the girl and asks where she is going.”
(6) Beta Generative AI response: “The girl tells him, and the wolf makes a plan to trick her?”
(7) Alpha Generative AI: “Correct. The wolf runs fast and reaches her grandmother’s house first.”
You can see that they are using conventional natural language, and in this instance, they are using English versus say French, German, or whatever.

Suppose for the sake of discussion that I wanted two AIs to tell each other the famous tale of Little Red Riding Hood. I will set up one generative AI that I will refer to as Alpha and a second generative AI that I will refer to as Beta. My aim is to have Alpha and Beta be directly connected, and I want the two AIs to interact about Little Red Riding Hood.

Quick Example Of AI-To-AI Communications

I will show you an example of how this AI-to-AI communication can occur in a multitude of seeming “languages” (variations of natural language, though, in theory, it could be an entirely new and different form of language, but I’ll cover that in a later column posting).

Those same designs and coded precepts could readily lead to two AIs leaning toward adjusting natural language to be more efficient when undertaking internal AI-to-AI communications.

For example, they might mathematically and computationally arrive at a shortcut kind of language, one that resembles ordinary language but is more efficient for doing back-and-forth chatter. This would be motivated simply by seeking to optimize communications. Most LLMs are devised to seek optimal aspects such as minimizing how many steps are taken or maximizing the likelihood of an answer or response being abundantly correct.

Deriving A Seemingly New Language

We are now at the grand reveal of this discussion. Grab a glass of wine. Take a seat.

Besides using conventional natural language, two AIs could proceed to communicate in something else.

What if we were to connect two AI systems and had them communicate with each other?

Your base assumption might be that the two AI systems would make use of natural language to chat back and forth.

Well, yes, in some cases, you’d be absolutely right about that assumption. For example, there is a method of having one LLM data train another one, known as knowledge distillation, that involves the AI’s doing a natural language chat with each other, see my depiction of this at the link here. Again, I want to clarify that the natural language is on the outside, not what is used on the inside when processing the chatter.

Human-to-AI communication then is primarily based on humans entering some sequence of words as a prompt in a natural language such as English and expecting to get an output from the AI that is likewise in a natural language and likely the same language such as English.

Communicating Words And Language

The AI is designed to assume that natural language is entered by people as a prompt and that the displayed output of a generated response is also to be shown in natural language. Meanwhile, internally in the AI, as I noted above, the reality is that language is pretty much a bunch of numbers that are all associated here and there with each other.

In a sense, a language is somewhat mailable by AI in terms of the spelling of words, the sequence of words, and the use of words, all of which can be bandied around because of various statistical and probabilistic considerations. This is not a sentient act at play. We do not yet have any sentient AI. The changing of words or a language would be due to computational and mathematical facets.

Words then are mailable by the AI.

In that sense, when you use generative AI frequently, you might from time to time see what seem to be oddish words. Those could be words that are perhaps old-time words that we don’t use much anymore but still are found somewhere online. They might be words that have been adjusted or mistakenly devised by AI. Etc.

Word Mailability And Language Mailability

It is quite easy for AI to essentially make up new words by either having seen misspelled words during initial data training and then opting to use those going forward, or by mathematically and computationally adjusting words and then adding them to the internal lexicon. There is even the rare but strident possibility of confabulating or making an error while devising a word, this is often referred to as a form of AI hallucination.

You might find of interest that the words of our natural language are not all necessarily equally made use of by generative AI.

Some words that are going to be generated and displayed as a response to your prompt are more likely to appear than other words. I bring this up in case you are intrigued by the notion that perhaps we will eventually have our existing use of language reshaped due to the pervasive use of generative AI, see my analysis at the link here. We will tend to see some words more so than other words, and this could, over the course of time, inadvertently shape which words we actively use in everyday discourse on a human-to-human basis.