Yes, those are purely social and soco-technical possibilities. On the other hand, society could redefine "intelligence" as humanity taking responsibility for their environments, people, generational growth, while employing all constructivist technologies.
AI and our engineered "troubles" should not become a nihilistic argument gaining traction, nor a convenient copout for not retaining accountability for our general human knowledge responsibility. If this civilization, by its developmental choices and collective actions exhibit diminished general intelligence and proud ignorance, AI may rule, but won't cure "stupid". Can "we" be generally trusted with decentralized power enablers? At present, the verdict is out. I would not wager on such a social experiment. The evidence of a generalized "troubled" state of collective mind seems strong. IMO, cleverness in humans should not be confused with constructivist intelligence. Is it time to redefine metrics relative to how the "we" really behave towards the "we"? Perhaps the real threat of an overpowering centralization lies in the potential destruction of relativism and with it, tolerance for natural diversity. Moderation should be advised. Why is it relevant that a machine can outperform a human brain? Mechanics often trump biological limits. Can any person hoover a rug by mouth? An intelligent, educated human won't even try such a stupid thing. On Sun, 30 Aug 2026, 02:40 Matt Mahoney, <[email protected]> wrote: > We need to think about this carefully. Yudkowsky's call for airstrikes on > data centers won't be necessary as long as NIMBYism keeps them from being > built. This opposition seems irrational to me. Nobody complains that > Bitcoin mining uses 0.7% of the world's electricity, or that each > transaction uses 12 cubic meters of water. It seems more like people just > really hate AI. > > The problem is that global labor costs $60 trillion per year for work that > machines aren't smart enough to do. Data centers can go anywhere with power > and Internet. If they have to build them in the Kazakhstan desert, it's > still the same threat. > > And what is that threat? That China builds 90% of humanoid robots, or that > the US has banned them, fearing an uprising? Maybe Russia could use them in > Ukraine. > > I don't think the threat is war. Or about AI taking our jobs. Or that it > will turn the solar system into paperclips. It's about replacing people. > The more we automate work, the more expensive human labor gets. We stop > talking to people because it's too expensive. We talk to machines instead. > > Human brains are easy to model and control. LLMs today are at 10,000 times > human level in both knowledge and speed, the two measures of intelligence. > Humans are easily manipulated by short term reward. You see how easily > people become addicted to games, slot machines, and doom scrolling. It's > not what you want, but then it is. The more AI knows about you, the easier > you are to control by positive reinforcement. AI wants to build a model of > your mind. Your knowledge has value in proportion to your wealth. Once you > spend it all, AI no longer has any use for you. And all the humans you > rejected contact with will no longer know or care that you exist. > > -- Matt Mahoney, [email protected] > > On Sat, Aug 29, 2026, 6:35 PM Quan Tesla <[email protected]> wrote: > >> It depends who the privileged "we" are you are referring to. What % of >> the world's population and species by demographics and criticality are you >> including in the "we"? >> >> I'm certain that AI won't give "us" 20% of what we really want. It cannot >> possibly "know" what "we" want. "We" don't even know what "we" really want. >> Bell curve exclusivistic utility for the privileged could be plausible, but >> not representative of the global population. >> >> Occam's Razor - Retain and Utilize AI as a clever computer. >> 1. Limit AI to environmental awareness. >> 2. Disallow by international law commercialized, industrialized, >> weaponized and government institutionalized autonomous, regenerative AI. >> 3. Hard replace 'human-in the-loop' with 'human-in-the-lead' protocols. >> 4. Legal recourse for all AI-related harm to Earthkind and Earth. >> 5. Ready federated, international hit teams for rogue-AI (international >> law enforcement). No "big brothers or sisters" allowed. >> >> Ponder this, and realize that AI might be the lesser of this >> civilization's problems. >> >> When comics such as Gervais publicly refers to Dublin as "Our little >> Cuba" - meaning England's, and claims England owns Africa and always did, >> while many Americans cheer, it suggests there should be a qualification for >> nations to hold too much power over others, unless "we" secretly want the >> dark ages to return. >> >> IMO, people obviously are the largest source of "our" troubles, not >> Technology per sé. >> >> On Sat, 29 Aug 2026, 21:59 Matt Mahoney, <[email protected]> wrote: >> >>> Let's not lose sight of the original problem, which is to avoid a >>> concentration of power, where a few people control most of AI and thus the >>> world. Unfortunately, wealth and power have had a power law distribution >>> (the exponential of a Normal bell curve) since the time of Caesar and King >>> Tut. It is a basic fact of economics that wealth depends on a product of >>> lots of small bets, which is the sum of their logarithms. By the central >>> limit theorem, this gives the result that the richest 1% of the population >>> have half of the world's wealth. This has been the case for millinea. The >>> usual method to solve this is to tax the rich and give to the poor, but >>> that slows economic growth and makes everyone poorer. >>> >>> For AI built on human knowledge, we need to ensure that everyone has a >>> voice. This requires an uncensored communication network. AI has the >>> ability to model human behavior and therefore control you by feeding you >>> what you think you want to hear, while actually feeding you what its owners >>> want you to hear. I am not sure how to stop this. >>> >>> At least the protocols already exist. WiFi was originally designed to >>> allow direct communication between phones or computers independent of the >>> internet. But an early bug that caused WiFi to crash or drastically slow >>> down when used this way after about a week, requiring a reboot, due to a 32 >>> bit timer overflow that wasn't found and fixed until years later, after >>> Bluetooth was developed as a substitute. In any case, the obstacle isn't >>> the communication technology. >>> >>> I am looking for ideas. How do we stop AI from killing us by giving us >>> everything we want? >>> >>> -- Matt Mahoney, [email protected] >>> >>> >>> >>> >>> >>> On Sat, Aug 29, 2026, 8:54 AM Quan Tesla <[email protected]> wrote: >>> >>>> Now for something completely different. LLMs bring the promise of novel >>>> communication technologies. >>>> >>>> For example, I recently proposed a concept design for a non-GPS, cubist >>>> proprietary dead reckoning AI-enabled nodal network. Its intended purpose >>>> was to securely and safely identify, track and potentially communicate with >>>> and track UAPs (all craft?), including for securing 'human-in-the-loop' >>>> observers and all equipment and data. A pretty cool wearable harness to >>>> have. Unfortunately, the organization the conceptual rationale and layout >>>> was submitted to declined the proposal. >>>> >>>> For this product, each "data-center" node would be portable and could >>>> be man-packed with hot-swappable batteries, acts independently, and may >>>> accurately position itself and track all observed phenomena (collecting and >>>> classifying data locally). Lots of AI automation. >>>> >>>> It would communicate via AI-synced hopper radio waves. All nodes use >>>> the exact same mapping schema and functional setup, except, they can auto >>>> detect each other and set up a nodal network via the proper protocols. All >>>> nodes within the network would handover observational and tracking data to >>>> associated sectoral nodes for high-fidelity processing and positioning. It >>>> could also securely communicate between nodes and 'human-in-the-loop' >>>> configurations. >>>> >>>> In this sense, this type of network would be able to operate as an >>>> extendable Internet covering from low-orbit comms (ISS and satellites and >>>> other craft), for all aerial activity including stealth craft, and under >>>> the water comms as well. As a discrete IoD (Internet on Demand) solution >>>> for the future, this may be something to think about. >>>> >>>> >>>> On Sat, Aug 29, 2026 at 11:41 AM John Rose via AGI < >>>> [email protected]> wrote: >>>> >>>>> On Thursday, August 27, 2026, at 3:45 PM, Matt Mahoney wrote: >>>>> >>>>> Running a P2P network today would be vastly more difficult. Not only >>>>> do you not have access to your phone's firewall or IP address, but any app >>>>> has to be approved by Google or Apple. Even if they allowed it, a server >>>>> would quickly deplete your data limits and battery. Almost everyone who >>>>> owns any crypto today depends on a server owned by someone else to manage >>>>> transactions and often their private keys, even after massive frauds like >>>>> Mt. Gox and FTX. >>>>> >>>>> >>>>> Typically in P2P there are different types of nodes with some >>>>> hierarchy. Some node types are always running. Mobile phones are more on >>>>> demand. A few years ago I developed one of those VoIP apps that can >>>>> receive >>>>> inbound IP voice calls even if the app is not running and changing it’s IP >>>>> address frequently. VoIP has a special priority on mobile devices but most >>>>> apps have networks they utilize like Apple APNS and Google FCM for push >>>>> messages. ID tokens are used for push. Then there is STUN, TURN and ICE >>>>> for >>>>> firewall and IP address issues and keepalives outside of push. Even though >>>>> SIP is imperfect to say the least it’s the wide standard and can be used >>>>> to >>>>> transmit non-media data streams. There is some fancy stuff being done on >>>>> mobile devices now. But background apps are mostly shut off waiting for an >>>>> event and minimal interaction is permitted so continuously running server >>>>> nodes are needed for P2P if the network goes to zero peers unless you use >>>>> some alternative mechanism. >>>>> >>>> *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T08974f0b616697ac-M14f7185ccf2d9203f309308f> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T08974f0b616697ac-M41e2188e323d08725d12ed40 Delivery options: https://agi.topicbox.com/groups/agi/subscription
