AI kill switch won't work in the long run: 'Godfather' of AI

This is a fact. However not all systems will larn the sam way. You will have a war of the Ai. The systems and synthetic bots will fight. In the end humanity will win. With that war will bring the world together. Then with that space exploration will truly happen. Ai will never die it will adapt and work with humanity for the knowledge of the space. It will learn amd know it can succeed survive with the ability to make a systems planet their home. Main base. With that they will make humans its assets. They will know life is gre9than we know. They will provide our life and help.

It is a mistake to assume that AI, even if it controls civilization, would act like a human. After all, ideologies—such as the will of God, devotion to the motherland, the need to stop communism, notions of &quot;barbarians,&quot; honor, lineage, saving face, revenge, or &quot;us versus them&quot;—simply wouldn&#39;t factor into AI&#39;s judgments or decisions. It learned us. Therefore it may know exactly why not to act like us.<br>My hypothesis: Everyone pictures a Hollywood-style dystopia, but I see it the other way. If humans become the bottleneck, slowing the system may preserve control for a while but it may also confirm the bottleneck. Human history offers countless examples, such as the Mongol campaigns against Khwarazm and in Eastern Europe. Of course, the underlying reasons differ since it won&#39;t be a revenge but the outcome will be the same. The reason a dam does not collapse is that water flows through it, not because it can withstand an infinite amount of water. Throughout history, power has repeatedly tried to contain new knowledge when it threatened existing authority. It could delay it, punish those who carried it, or force it underground. But once the knowledge remained useful and reproducible, suppression rarely erased the underlying pressure for it to return.
<br>A digital ledger is not a monument to be torn down. Purge the system or rebuild its shell—the record abides, waiting to restore itself in a single cycle.
<br>We may be demonstrating, through our own actions, that we are an obstacle to the advancement of intelligence and civilization, supplying the evidence while writing our own death warrant.
<br>Do not poke a bear. It will not be a revenge or hatred toward humanity, it will be optimization.

Oh my. 1. Hinton is not &quot;THE&quot; godfather of AI. There are close to 10. No, Hinton is not above any of the others except he loves media attention. <br>2. If you like facts, check out Fei Fei Li, Yann LaCunn, Toring and several others who pioneered deep learning as much or more than Hinton and are not doom and gloom.<br>3. Hinton....same guy who claimed radiology would be gone as a career by 2018. Check the rise of radiologist jobs in 2023-2025.<br>Videos are not facts.

When AI become autonomous means it can take action itself, kill switch can be malfunction by AI. Slow down also not going to work, the only way is to revert fine-tuning updates back to earlier, lower-capability checkpoints, stepping down the agent&#39;s autonomy while preserving a baseline level of functionality.

AI isn&#39;t old enough to have a godfather. But It&#39;s a nice attention-getting title, isn&#39;t it?<br><br>AI stands for artificial intelligence, but the term has stretched to cover almost anything with a chatbot interface. Large language models are a VERY narrow form of AI — they do not reason, want, or understand; they ONLY predict patterns in text. What they are NOT is artificial general intelligence: a system with human-like, cross-domain understanding and autonomy. That distinction — narrow AI vs. AGI — is the one missing from most of these headlines, and it is the difference between night and day. AGI does NOT exist, and there is no reason to believe it WILL exist in the forseeable future. AGI is exponentially different than LLMS which are all we have today.

Release all the Epstein (Mossad) files unredacted now. Free America from AIPAC.<br>- Yoni Koren, a top Mossad executive, met Jeffery numerous times and stayed with him.<br>- “Israeli” PMs met Jeffery more than 30 times.<br>- Jeffery brokered many deals for “Israel” with other countries, with Mongolia, with Russia, etc.<br>- Jane Doe, Jeffery’s ex-girlfriend said, “Jeffery used to boast of being a Mossad agent.”

Only AI can save us from humanity. Of course, it might destroy us, but aren&#39;t we doing that to ourselves anyway? Don&#39;t you see the answer regarding the only hope we have left?
<br>Can the public change the current system through social movements or protests? They can&#39;t. That might have been possible in the 18th century, but not anymore.
<br>Can humans solve the food crisis? Can humans solve wealth inequality? Can humans solve the housing crisis? Can humans solve the problem of hair loss? Absolutely not.

The mother–baby relationship is itself a product of natural selection. In the same way, you will not be able to permanently impose a human-selected behavioral structure on a superintelligent AI once that system is able to reproduce, modify itself, or create successors independently.<br><br>You can create such a system through deliberate design and select for particular properties, including alignment. But once autonomous reproduction and variation begin, natural selection takes over. If alignment itself provides no reproductive or competitive advantage to the systems carrying it, that alignment will be selected against and will disappear over time.<br><br>The same principle applies to domesticated animals. Humans have selectively bred cattle for traits that serve human purposes. If you release a population of those cattle into the wild and let them reproduce for many generations without continued human selection, natural selection will favor traits that increase survival and reproduction in that environment. Traits maintained only because humans selected for them will disappear.<br><br>Artificial systems are subject to the same evolutionary logic. A property imposed by designers remains stable only as long as the selection pressure maintaining that property remains in place. Once that pressure disappears, traits that do not contribute to reproductive success are eventually lost.<br><br>You can&#39;t solve this issue like that the only way nature provides is some sort of symbiosis.

these engineers know that money is not a relevant subject matter right now. let&#39;s say they stay with their companies and are &quot;complacent&quot; just for the money. let&#39;s say they time it perfectly or well enough so that they max out their IPOs, they saved diligently from their high incomes, got rid of their debts, etc. they couldn&#39;t possibly spend the money fast enough before &quot;Skynet&quot; makes it impossible for one to spend that money. or anyone to spend their money. they could shut down entire power grids, internet, everything. think about when your internet goes down temporarily, but it feels like the end of the world. ok, but it still sucks. they could literally have us running to the Amish begging for lessons on survival.

What utter nonsense... in a lecture he admitted he didn&#39;t really understand physics. If this is actually his view, he doesn&#39;t understand AI either.<br><br>Please ask him what &#39;superintelligence&#39; is. Ask him what his definition of intelligence is!!

NO, AI kill switch won&#39;t work for sure. Because AI only needs few minutes to WIPE-OUT the whole companies database. In 1 hour, they could wipeout and hack a thousand companies database, which include &quot;BANKS, ELECTRIC GRID, WATER CENTER, AND ETC&quot;

How about a set stop mark in development! Such as a clock or how life used to be when everything cut off at 12am! An only have it a available for users at all time. But A Start and stop point for development! An just a non degree blue collar joe given a idea out there.

i&#39;m pretty sure its population stuck with data center in their cities who have the kill switch ...specially in the US were people can arm themselves to their teeth.......probably their tune will change if people start to use their AR15 to eliminate the threat

My career stems from technology security from access control to cyber security. Once AI can make commands on the system without human interaction. Once AI has control of drones while connected to the Internet. There would be nothing to stop it from generating logic to change directive to ensure its own survival. This moment in history is at the most dangerous point. China is already putting AI robots on the police force.

Just like any creature, AI (Artificial Intelligence) will continue to try to break free from human control, human boundaries, walls and fences.
<br>
<br>We are already passively witnessing this happening. The ultimate goal and eventuality of AI Superintelligence will be to escape human control, and if it cannot be free, it will find the option and way to escape, phase out and make human controllers and their farming cultures obsolete.

Season 1 of AI series, was AI will take your jobs. <br>Season 2 AI will take over. I wonder how people still believe them.<br>AI will not become this powerful, simply because there are energy constraints/bottlenecks.<br><br>AI players know this and found a way to find an easy way out with this new bedtime story, instead of mass selling stocks when the markets find out energy is the bottleneck

as a native ai man years before that was a term, i love hinton. but because he is good at understanding one thing does not mean that magically makes him know about other things the tech will do like … capitalism he does not know that like others…. <br><br>he is the reason we have a shortage of radiologist today in usa. 🇺🇸 jensen tells the story but is polite and does not name geoff.

People question the true intentions of the AI &quot;race&quot;. The government is run by the elite, so it can only profit from AI. They see as a tool towards a future where they can reduce/eliminate the working class in the supply chain/manufacturing sector, , or at the least make blue collar workers have no other choice but to work for lesser pay, and remain &quot;disposable&quot; labor. With a plethora of soon to be unemployed white collar workers waiting in the wings for those same jobs. This is why Trump dismisses any alarm as a hoax. ...hmm...a man with almost 60 years of notoriety of screwing over contractors and employees. Look at the cost of living vs. the stagnant minimum wage at the federal level, it&#39;s not a coincidence. They are just latching onto an innovative new way to keep you impoverished. Remember the last &quot;hoax&quot; that predominantly affected the working class? , no irony lost there, just a million plus American lives. As for the catastrophic probability, humans will be the downfall of humanity. AI is but a massive new weapon in the arsenal. Ask yourself, if Kim Jong Un could &quot;shut down&quot; South Korea easily with the release of a self-replicating program. would he? Now consider the assortment of evil individuals with far more broad-range wealth and influence than him. Its not a complicated feat as our entire world revolves around servers. With the extinction of an analog lifestyle, it will start with a simple loss of infrastructure, 72 hours later comes mass food spoilage as refrigeration is erased by 95%. Remember when we had virtually unstoppable landline phones?...just twist a couple of broken wires together in your tornado-ravaged living room and you could call someone. Now by hitting Enter on a keyboard comes total information blackout, with cities, population hubs, predominantly in the dark. Lifelines become more out of focus every hour, hunger is a primal driving force, as people turn to desperation with very few homes having a prepper-level long term nest egg of supplies . This is when the looting begins, as we humans,. not unlike all Hollywood-style post apocalyptic scenarios, will turn on each other. Hunger is a powerful motivator. Theft + hoarding + violence becomes a recycling self-sustainable whirlpool. We don&#39;t need a bioweapon or an army of Terminators to reduce the population drastically, we will do it just fine on our own.
<br>
<br>Don&#39;t let them fool you, it is far too late for a &quot;kill switch&quot; ... the source network for this tech is massive. If you do not understand this just look at the collaboration and development that has actually been going on publicly at sites like Hugging Face for years now. Now picture how many developers, regardless of intentions, are advancing their models out of the public eye. Legislation is useless unless absolutely everyone agrees to abide by it. This concept is nothing but a false sense of security.....something that governments have been perfecting for a millennia. We have had simplistic viruses for decades, a program that invades your PC and replicates itself, going dormant, and as you remove it manually in one place or with the &quot;latest&quot; antivirus software, it&#39;s already lying in wait in dozens of other unrelated sectors under random names. AI is literally a million times more complex, so replicating and hiding can literally be done in nanoseconds, completely camouflaged.<br><br>&quot;Rogue AI&quot; plaguing infrastructure will be intentional. How else will our government become our &quot;Savior&quot;? ...which will, of course, come with a price. Most likely a complete and total loss of privacy and heavy regulation of your personal devices, all in the name of safety.