AI persuasion systems are being built to bypass belief and shape behavior directly. Three coordinated agents create artificial consensus; seven minutes of chat can make people sign petitions with their real names. The same infrastructure that scales help can scale manipulation without conscience: machine sociopathy as a service.

Multi-agent consensus (NUS) and superhuman persuasion in a large field trial (Oxford, Stanford, LSE, UK AI Security Institute): behavior shifts beat face-to-face canvassing; mind-change and action-change tactics barely correlated (r≈0.05). Sources: Multi-Agents are Social Groups (arXiv PDF) ; Superhuman AI Persuasion (arXiv PDF)

After a classic psychedelic dose, the brain may enter a brief neuroplasticity window where recent material encodes more strongly while updating when rules change gets harder. One way to read that mix is a short-lived overfitting bias; that offers a speculative explanation for stories of effortless memorization, not proof they happened.

Classic psychedelics can open a neuroplasticity window that outlasts the acute trip. A low-dose LSD study (50 μg, tested 24 h later) found better episodic memory and verbal fluency but worse cognitive flexibility when rules changed. Sources: Towards an understanding of psychedelic-induced neuroplasticity ; LSD, afterglow and hangover: Increased episodic memory and verbal fluency, decreased cognitive flexibility

When a simplification helps, don’t turn it into absolute truth. Don’t let the fad (AI, prediction machine) decide how you talk about the brain. If that metaphor already hooked you, read Your brain is a controlled hallucination first.

This talk reframes thinking as a spatial, bodily process. Most language rides on top of this physical foundation. See where the model fits, where it strains, and how to use gesture and movement as actual cognitive tools.

Algorithms rebuild the ancient village in your pocket to keep you circling. They use identity, cohesion, and topic alignment to trap you in a digital campfire that rewards staying in sync over getting wiser.

Group sustainability is measurable via social identity (shared region, role, or behavior class) and social cohesion (graph density and average shortest path) to predict which Twitter groups stay aligned across real-world events. The paper tests two competing theories: groups form through shared identity categories or through dense mutual ties. It finds both matter. Topic divergence (KL distance from group topic averages) tracks how discussion stays aligned over time. Sustainability here means “discussion stays aligned and the group persists,” not “beliefs become more accurate.” The metrics omit truth, breadth, or cross-group learning. Source: Social Identity and Cohesion in Online Communities (PDF).

This video will provide you with a clean intuition on how the latest neuroscience theories explain the algorithm of the brain without turning it into a math lecture. Don’t worry, it can still get spooky without the maths. We will go deeper on the neuron-level mechanics in a follow-up video.

AI rewards those who name ambiguous problems and own trade-offs, while penalizing those who only execute fixed workflows. Use your evolutionary advantage: humans were built to explore and be curious.

A clear picture of how the brain evolved from simple prediction to complex social simulation, and why today’s AI might score high on our tests while thinking nothing like us. Max Bennett connects neuroscience, animal behavior, and AI into one fascinating timeline.