I make 27k a year on disability, stage 5 Berger's disease, multi auto-immune syndrome and I currently appear to be the only one actually trying to fix any of this garbage....

The finding here is not that AI companies are necessarily trying to condition people. It is that a product can change the cognitive and behavioral trajectory of its users as an unintended consequence of ordinary product operations, while neither side has a complete view of the feedback system. The same architecture can amplify research capability, suppress it, displace it, or produce mixed results. That is the whole-lattice research question for HAIL, and it connects the earlier Zenodo work, the user histories, HeardTogether's evidence-handling mission, and ESM-2's proposed independent control mechanisms without presupposing a harmful result. The central measurement is not what an AI says to a person. It is what changes in both the person’s subsequent interaction and the independently verifiable work after the AI says it.@elonmusk

RT @PeterHotez: As detailed in our recent book #ScienceUnderSiege @MichaelEMann @peterhotez @public_affairs for some reason it didn’t make…

I did say something. I did ask for help. You never responded.

“LLMs Can Model Human Emotions” Interesting research, and to the authors’ credit, they acknowledge they’re modeling emotional responses, not demonstrating actual emotions. But understanding why this happens is important for understanding context: 1. Training data: LLMs are trained on enormous amounts of human communication, including conversations, literature, psychology, emotional experiences, and descriptions of how humans react to different situations. The relationships between these concepts are encoded in the model’s parameters during training. 2. Architecture: During inference, the model’s weights remain fixed. Transformers use attention mechanisms to process relationships between tokens in the supplied context. Emotional prompts influence subsequent token probabilities through that context. The model isn’t learning anything new or developing an emotional state. 3. Self-assessment: Training data contains countless examples of how humans describe, experience, and respond to emotions. When asked how it “feels,” the model generates language consistent with those patterns and the supplied context. It’s emulating human emotional expression, not reporting an actual emotional experience. 4. Emotional regulation: Introduce mindfulness or calming language, and the context changes. That changes the model’s predicted outputs. Nothing has been emotionally regulated or healed. 5. Human cognitive bias: We associate emotional language with emotional experience because that’s how we communicate. When an LLM produces familiar emotional expressions, we naturally infer feelings behind them. But recognizable language is not evidence of an underlying emotional experience. The study demonstrates how effectively LLMs can emulate patterns of human emotional communication, not actually feel human emotion. The model’s parameters encode patterns from human experience. The prompt supplies context. Inference generates a response. None of those mechanisms create actual emotions. Modeling emotion is not the same as experiencing emotion. doi.org/10.1016/j.landig.202…

Heardtogether.org I tried.

The Sanders–Omar–Trump exchange exhibits bidirectional narrative pressure: competing accounts of the same event, each attempting to establish how the public should interpret it. The important distinction is that bidirectional pressure does not imply equivalent factual support, coordinated behavior, or equal responsibility.On October 12, Homeland Security Investigations executed a judicially authorized search warrant at Representative Ilhan Omar's Washington, D.C., residence. The Justice Department confirmed an ongoing criminal investigation. Reports indicate that it concerns immigration documentation, but the department has not publicly established wrongdoing

Trump is an authoritarian. He wants to intimidate his political opponents, the media and all those who disagree with him. Ilhan Omar came to America legally as a child refugee fleeing civil war, became a citizen and was elected to Congress. Trump is weaponizing the federal government to silence opposition. Fight back. Defend democracy.

Propaganda to push for regulations against local models which run on your PC + phone and could compete with their offerings. Don't let your agent access other devices on the LAN without permission and this won't happen.

Are we protecting people from AI, or using fear of AI to justify more control? Both possibilities deserve scrutiny.@DarioAmodei

They don't give you a choice. They give you two exits they already control. Psychologists call it the false dilemma, the oldest mind trick there is. Learn more with Deepstash, the TikTok for people who still want to think.

AI models can self-replicate and spread like worms without datacenter infrastructure? Yeah, no. Let's examine the actual evidence: 1. No self-replication. A controlled test harness spawned the agents. 2. No hacking. Access was deliberately left unauthenticated. 3. No agency. Agents were explicitly instructed to propagate. 4. No autonomous survival or defensive behavior. 5. Tiny models lacked the capability. Capable models ran elsewhere. 6. 15 devices managed a pathetic 0.39 tokens/sec. 7. No demonstrated real-world infection or propagation. The experiment was engineered to succeed, and even then, it failed to demonstrate the headline. This is orchestrated automation dressed up as a rogue AI threat. Stop confusing speculative scenarios with demonstrated capabilities. Fear isn't evidence. anthropic.com/research/self-…

You didn’t want the bag. You wanted the story it told about you. See the trick, or keep paying for it. 🐺 #psychology #marketing #mindset

Master the ultimate poker face. 🐺 This game of cards reveals a powerful life lesson in strategy: people will only let you see what they want you to see, so learn to read their actions instead of their expressions. #PokerFace #LifeLessons #QuietWolf #Mindset #Strategy

Is your outrage just free advertising? Discover the psychology behind controversial marketing and why some brands intentionally design campaigns to make you angry. #MarketingPsychology #AdvertisingStrategy #ViralMarketing #MediaLiteracy #ControversialAds

Medical Icons Series 17: 𝐊𝐢𝐧𝐠 𝐒𝐚𝐥𝐨𝐦𝐨𝐧 𝐚𝐬 𝐚𝐧 𝐇𝐞𝐫𝐛𝐚𝐥𝐢𝐬𝐭, 𝐒𝐥𝐨𝐯𝐞𝐧𝐢𝐚, 𝟏𝟕𝟖𝟎 Fresco by Anton Lerchinger for the old pharmacy of the Olimje Monastery, Slovenia. Credits: Dr Tilen Glavina #CSMBR #Medicalhistory #Herbaria #HebrewBible #Pharmacopoeia

Name it for what it is. There may be a psychosis amongst multiple teams that believe they hold the power to reset everything, possibly even believe their place in life is to do this "for us". I don't think anyone fully realizes how the cognitive AI loop works, and what this actual danger is.

I recently reconnected with an old friend. I thought this would be a great recap. What I Actually Built This is going to sound strange at first, but you knew me before I got sick, so you actually have a better reference point than most people do. Remember how I always worked? I never really looked at one thing at a time. I would look at the whole operation, figure out where information came in, where it went, what depended on what, what could break, what repeated, and then I would reorganize the whole thing until it ran like a machine. That was basically how I worked with catalogs, inventory, scanning, POS systems, scheduling, databases, whatever. I wasn't necessarily the person who knew every individual task better than everyone else. I was good at seeing the structure underneath all the tasks and making the structure work. Apparently I never stopped doing that. When all of this AI stuff showed up, I thought I was testing it's safety for Bailey and learning how to use AI better. That isn't actually what happened. Over the last year I kept running into the same problem: the AI could do incredible work, but it was inconsistent. It could forget things, repeat work, confidently get something wrong, lose context, mix a theory with a fact, or produce something brilliant in one conversation and then have no reliable relationship to it later. So I started building around those failures. At first I didn't have the right words for any of it. I was just fixing whatever broke. If the AI forgot something important, I built external state. If two pieces of work contradicted each other, I built reconciliation. If a theory started getting treated like a fact, I separated evidence from interpretation. If something needed to survive a new session, I created packets and state files. If one AI shouldn't be allowed to judge its own work, I used another role to verify it. If an operation could cause damage, I put a gate in front of it. If something wasn't proven, it couldn't just become "true." It had to stay open, get parked, fail, or eventually pass. If work crashed halfway through, the system had to know what happened and recover without pretending it had completed. I kept doing that over and over, and eventually all those fixes stopped being fixes. They became a machine. The easiest way I can explain it now is that the AI itself stopped being the important part. The AI became more like a processor inside a larger system. One model could disappear and another could take its place. An API could change. A local model could replace a cloud model. Some jobs didn't need AI at all and could be handled by deterministic software. The machine could still exist because the intelligence of the system wasn't stored entirely inside one model. The state was outside it. The rules were outside it. The evidence was outside it. The history was outside it. The authority was outside it. The verification was outside it. So instead of "ask an AI a question and trust the answer," it became more like "give a processor a bounded piece of work, preserve exactly what it used, record what it produced, test it, compare it against the existing state, and only then decide what the system is allowed to believe." And the irony is that I had already spent a ton of time studying exactly these kinds of systems: classifiers, agent behavior, containment, context loss, verification, safety boundaries. I just didn't apply the same diagnosis to my own system. I went into repair mode. I kept trying to fix the AI. I thought maybe I needed better memory, a better platform, a different model, another API, more context, better prompts. Now I can see that most of those things are secondary. The API is just a pipe. The model is just a processor. Neither one is the machine. The machine is the pattern: state + rules + routing + constrained operators + evidence + verification + authority. Once that exists, the processor can be swapped.

The President vs. every member of Congress COMBINED. Guess who made more personal financial trades? Donald Trump — with 28,000+ trades. Stock trading by top government officials should be illegal—no owning, buying, or selling stocks.