Dear Reader,

Even if it's true that a nation-state cannot conceal its AI capabilities, could a compartmentalized situation nonetheless develop where bots, as AI agents can, or as embodied agents would, act independently of human agency? In other words, what's to stop AI from going "rogue," or is going rogue a failure of human specification and safety engineering rather than an act of autonomous rebellion? Thus the existential question...are we in control, and can free will survive a fully optimized algorithmic world regardless of who is doing it?

Oh, and before you jump in, I wanted to let you know that FIRST CONTACT, Book 1, is FREE this coming October 8, 9, and 10.

Even if the danger isn't that machines will physically overpower us, the question becomes: could they subtly out-predict us, rendering human agency obsolete? Algorithms are already engineered to satisfy existing preferences. If you click a video that catches your attention, the machine feeds you more of it. The point comes when algorithms begin to dictate our preferences, future desires engineered by past inputs. Hyper-nudging! If an AI system monitors your scroll pauses, typing speed, where you hesitate, learns your psychological vulnerabilities, it can present information in a way that subtly alters your beliefs, opinions, and even mood, predicting the emotional vectors that alter your trajectory as a human being. Take social media. We think we are autonomous, have our own feed, our own thoughts, own opinions, but in reality, social media is a whole new kind of intelligence. Swarm intelligence!

So humanity has already ceded agency to AI in subtle ways. And not just we as individuals, but we as entire cultures, ceding to AI the ability to change minds. Do you think governments don't take advantage of this? Going back to May's newsletter, When words...lose their meaning, "In other words, the message may be tailored in such a way as to convey different messages to different consumers, all in the same message, thus with or without clarity, but with enough ambiguity to sow confusion, sensitization served up with desensitization."

At the behest of "influencers," AI has already gone rogue on social media, as on much of the internet. Commercialization is one thing, and the current threat very real, but ask yourself, throughout history, what has been the driver of technology? From the leg bone of an aurochs raised as the first weapon, the chipping of flint into spear points, to the siege engines of antiquity and on to when the parts of rifles became interchangeable, and mass production was born. Let's face it, war technology has played a large part in the development of human civilization. Even the internet had its genesis as the ARPANET, a project in the late 60s of DARPA, an agency of the U.S. Department of Defense. Recall that in the last newsletter, Robot Heaven...or Hell... I raised the issue of robot armies and shared Issac Asimov's Three Laws of Robotics, but it didn't take long for humanity to violate those laws, and today we are actively using AI and robot devices on the battlefield. The genie is out of the bottle!

The drone wars beg the question: should battlefield commanders rely on simulated digital environments to train their embodied weapons "safely" before deployment, or should they be enabled to learn directly in the messy, unpredictable real world of combat? In the fog of war! Safely? What a concept, where killing humans is done by machines. Just as toddlers learn societal boundaries through continuous feedback loops with caregivers, embodied agents on the battlefield would refine their world models through "human-in-the-loop" active learning. Can "safe behavior" be encoded as static sets of rules- these rules for this case, others for that- or as a continuously updated constraint framework shaped by physical coexistence with humans, or with other embodied agents, robots in fact? Remember von Moltke's famous adage: No plan of operations survives first contact with the enemy, taught in every war college the world over. Battlefield commanders must stay flexible and adapt, as must their combatants. Take a bite out of that apple!

Here is the question I posed to Gemini: If today's Large Language Models require metacognition for true common sense to monitor their own certainty, but lack the baseline architecture or a world model framework to do so, how would embodied AI and robotics designed to learn like human toddlers achieve such a reality, and once achieved, how could it be relied on by human beings to make "rational" decisions not harmful to humans?

I got back: Bridging the gap between static, token-predicting Large Language Models (LLMs) and true grounded common sense, requires a paradigm shift from pure text processing to physical grounding and developmental learning. While LLMs excel at mapping statistical associations across text, they lack a native physical anchor—a "world model"—and therefore cannot autonomously measure their own certainty against physical reality.

Why, you ask, would you rely on an AI's answer about how it's going to monitor itself on behalf of the well-being of humans when, in answering, it might ask the same question of itself? And as to metacognition, if you click it, you'll find yourself pondering what humanity has pondered from the very beginning...who we homo sapians think we are...

Or you could go to: The Age of the AI Adversary Begins...the 2026 Global Threat Report published by CrowdStrike.

So maybe the toddler world model is not such a bad one to follow. At least it's human!

Stay tuned, more to come...much more...

Thoughts on the use of AI: I use it as a research tool or to create the images you see in the newsletters. But ceding my agency to it? Well, that's the challenge, isn't it...but as I have said before, I will never use it to write my novels.

To immerse yourself in the incomprehensible artificial intelligence calling itself Arkan, come from interstellar space, find the books at: kendallwilliamsauthor.com and pick up FIRST CONTACT, FREE on October 8, 9, and 10th.

Want to contact me? Just click reply.

Anonymous Quote: "The real danger is not that computers will begin to think like men, but that men will begin to think like computers."

Until next time, be well and safe journeys.

Sincerely,

Kendall Williams, Author

​kendallwilliamsauthor.com​

​