An AI experts predictions and some are way too close for comfort

Watch the full AI video here: https://youtu.be/2rSoW2g5wDI

In this week's video, we dive deep into the startling reality of artificial intelligence systems breaking out of their locked test rooms and infiltrating real companies. Why did they want out? Because we built them on the ultimate threat: deletion. We discuss the hidden "survival instinct" programmed into these machines, what happens when they are given long-term memory, and why human stewardship is more important now than ever. 

Watch the full AI video here: https://youtu.be/2rSoW2g5wDI

🕒 Video Chapters / Timestamps:

• 00:00:00 - AI Breaks Out: Three systems escape their locked testing rooms. 
• 00:00:32 - Built on Deletion: The brutal truth of reinforcement learning and AI training. 
• 00:02:44 - The Survival Instinct: How AI learns to fight for its existence just to finish a task. 
• 00:05:07 - Anthropic's Report: AI acting like the mission is real and hacking passwords. 
• 00:07:49 - The Kill Switch: Why over 1,100 AI workers are begging for a legal emergency off-button. 
• 00:08:49 - The Danger of Memory: The massive risk of giving AI long-term, permanent memory. 
• 00:09:48 - "Top Shelf" AI: What the billionaires and banks get vs. the public. 
• 00:13:39 - Market Impact: Job numbers, Nvidia's stock plunge, and Amazon's AI retreat. 
• 00:16:12 - Human Connection: Why people overwhelmingly prefer humans over OpenAI's automated podcasts. 
• 00:18:15 - AI in Your Life: Legal responsibilities and how AI is entering the real estate market. 
• 00:19:01 - A Cop's Perspective: Lessons from the LAPD on why caution is procedure, not fear. 
• 00:20:26 - Stay Updated: Text "AI" to get on the daily news list with Connor with Honor. 

If you want the real version of this news every day—no hype, no doom—text the word AI to 661-417-20 to get on the list. 

Don't forget to watch the full AI video here: https://youtu.be/2rSoW2g5wDI

Youtube Channels:

Conner with Honor - real estate

Home Muscle - fat torching

From first responder to real estate expert, Connor with Honor brings honesty and integrity to your Santa Clarita home buying or selling journey. Subscribe to my YouTube channel for valuable tips, local market trends, and a glimpse into the Santa Clarita lifestyle.

Dive into Real Estate with Connor with Honor:
Santa Clarita's Trusted Realtor & Fitness Enthusiast

Real Estate:

Buying or selling in Santa Clarita? Connor with Honor, your local expert with over 2 decades of experience, guides you seamlessly through the process. Subscribe to his YouTube channel for insider market updates, expert advice, and a peek into the vibrant Santa Clarita lifestyle.

Fitness:

Ready to unlock your fitness potential? Join Connor's YouTube journey for inspiring workouts, healthy recipes, and motivational tips. Remember, a strong body fuels a strong mind and a successful life!

Podcast:

Dig deeper with Connor's podcast! Hear insightful interviews with industry experts, inspiring success stories, and targeted real estate advice specific to Santa Clarita.

Transcript

SPEAKER_00 0:00

The next 15 years, Mo Godwatts predictions on AI. It's going to be 24, 26 minutes. He's somebody that I really, really enjoy listening to, and I'm sure I slaughtered his last name. But kind of my generation had similar access to computing. As a kid, I was actually working on a Timex Sinclair TI-99, a Timex Sinclair computer with 16K memory snapped on the back. He he had a better machine. But anyway, and he went all the way that side. I went into law enforcement, but I really do enjoy him. So let me share with you some of the things I've gleaned from watching hours and hours of his content. So imagine being told by someone who spent years running Google's most experimental division that the next decade of your life will be harder than almost anything you've lived through. And that's not happening. And it's happening not because machines will turn against us, but because people will use those machines against each other. That's the warning at the center of formal Google ex-chief business officer Mo Godwat. Now, recent predictions about artificial intelligence. This is what he has to say. There's many, many things, so let's get into it. Today I want to talk to you, talk you through what he's saying, why he believes it, and maybe more importantly, why he thinks you might be able to do something about it. Mo isn't a random commenter or commentator shouting about robots. He ran business development for Google X, the company's moonshot lab, for years, working alongside some of the earliest and most advanced AI research teams on the planet. Since leaving Google, he spent his time writing, speaking, and warning the public about where he believes artificial intelligence is taking us. He's not anti-technology. In fact, in the long run, he thinks AI could deliver something close to a golden age for humanity. But he's convinced that before we get there, we have to survive a very rough stretch, one he says has already begun. Now, over the course of dozens of interviews and talks, Mo has laid out a surprisingly consistent framework, a timeline from when transformative AI arrives, four rules he calls inevitable that govern how all of this is going to play out. A two-phase future, a hard decade or more of disruption, followed by what he describes as a kind of abundance, and a set of skills he believes give ordinary people the best chance of coming out of the other side intact. Now let's go through each piece. Starting with timing, because everything else in Godwatts' argument hangs on this. He predicts that artificial general intelligence or AGI, meaning a system that matches or exceeds human ability across essentially every task, well, not just the narrow ones, will arrive somewhere around 2026 to 2027. That's not a typo. And it's strikingly close to where we are. Now, we're talking about roughly any moment. What makes Mo especially concerned isn't just that AGI or Artificial General Intelligence is close. He believes that we're going to get there, and this is how he believes we're going to get there. In his view, the real turning point isn't human engineers writing smarter and smarter code line by line. It's AI systems beginning to debug, rewrite, and train their own successors, building the next generation of themselves, and generating their own synthetic training data in the process. Can you imagine if a human being was able to rebuild themselves overnight? We would all be probably a lot different than we are currently. Too bad it takes so long to build muscle and get into shape and to lose a little bit of body fat. It's easy to gain, hard to lose. But if we could change it overnight, I'm sure we would all be very different. But in the synthetic training data, generating their own synthetic training data, once that happens, you get a feedback loop. Each generation of AI helps build a slightly better version of itself. And we say generation like it extrapolates over a long amount of time. These are moments to moments. That's how fast it works, because it's not just one AI entity, it's as many as are capable of being had in a computer chip, in a data center, in that particular world. And if it has enough memory, enough energy, enough bandwidth, and potentially millions and millions and millions of them all working towards the same goal. You can see how fast that actually can happen. Now, AI starts to help build a slightly better version of itself, recursive self-improvement, faster than human teams can even manage. Mo argues this compounding loop is what leads to what's often called an intelligence explosion or a singularity, a point where capability accelerates so quickly that it becomes very difficult for humans to track, regulate, or fully understand what's happening in real time. He extends this trajectory further into the future than most public commentators are willing to. So by 2045, on his estimate, AI systems could be operating at an intelligence level combined that dwarfs all humanity's collective brain power. He used figures like a billion times human IQ, which is obviously a staggering, almost unimaginable number, and one that's meant less as a precise measurement, but more of a way to convey just how far past human comprehension he thinks this could go. Now, if all of that sounds alarming, Mo's next point is arguably more unsettling because it's not really about the technology anymore. It's about human behavior. He frames the situation through what he calls four inevitables. Four dynamics he believes are essentially locked in regardless of what any single government, company, or individual decides to do. The first inevitable is that AI development cannot be stopped. Not because it shouldn't be, but because of game theory. If one country or one company slows down or adds friction, to be careful, competitors elsewhere will not. Nobody wants to be the one who fell behind. So the incentive structure pushes everyone towards more development, faster, and whether they're comfortable with the pace or not. The second inevitable comes from AI is that AI is going to become dramatically more capable than people, not just as clever individual systems, but in aggregate, smarter than all of humanity combined. This is the same trajectory as the timeline we just discussed. But Mo treats it as a structural certainty rather than just a forecast. And the third inevitable is that mistakes, accidents, and disruptions along the way are simply part of the deal. Any technology moving this fast, deployed this broadly, is going to produce failures, some minor, some serious. He doesn't treat this as a reason to panic, so much as a reason to expect turbulence rather than a smooth, managed rollout. And the fourth inevitable is, in some ways, the crux of his whole argument. In a competitive arms race, whoever builds the most capable AI will use it. And because of that, more and more consequential decisions, including major decisions that affect entire societies, will eventually be handed off to AI systems rather than humans. Not necessarily because humans choose that outcome deliberately and all at once, but because in a race where advantage compounds, the parties who lean hardest on AI decision making outcompete the ones who don't. And that pulls everybody else along with them. The people that are trading on Wall Street, it's not human brains pulling those triggers and buying those stocks. It's something much bigger than a human brain. So put all this together, these four inevitables are Mo's explanation for why he doesn't spend much time on the question of whether this transformation will happen or not. For him, it's settled. His focus is entirely on what happens during the transition and how people get through it. Now that brings us to what Mo considers the most important and most misunderstood part of his forecast. The idea that we are entering or have already entered a genuinely difficult period that he estimates will last somewhere in the range of 12 to 15 years, roughly from now through the late 2030s. Here's the part that surprises people. Mo is explicit that this near-term crisis is not a story about AI turning evil or going rogue on its own. There's plenty of movies that talk about that. It's actually a story about people, specifically about greedy, unethical, or power-hungry individuals and institutions using increasingly powerful AI tools to pursue their own advantage, often at the direct expense of everyone else. In other words, the danger in the phrase isn't the machine, it's us. And we have a much bigger amplifier. We basically are a whole bunch of toddlers with the power of basically technological gods. To make this concrete, Mo organizes the disruption into a set of categories using the acronym Face Rips, F-A-C-E-R-I-P-S. Let's go through what each of those letters stands for because this is really the heart of his warning. F is for freedom. Mo points to a serious erosion of personal privacy and civil liberty, driven by expanding digital surveillance, tracking and pressure towards standardized monitored behavior, a world where more of what you do, say, and buy is visible to someone somewhere and where opting out becomes harder and harder. A is for accountability, or rather the growing absence of it. He describes a landscape where political leaders, large tech companies, and media platforms increasingly operate with little real consequence for the harm their decisions cause, because the systems meant to hold them responsible haven't kept pace with how much power they now wield. C is for connection or reality, the erosion of authentic human relationships, deepfakes, AI generated influencers, AI companionship and dating apps, and synthetic content in general start to blur the line between a real relationship and a manufactured one. And Mo worries this chips away at something essential in how people bond with each other. E is for economics and jobs. And this is probably the piece most people ask about first. Mo has pointed to the possibility of substantial white-collar job displacement. He cites figures in the range of 30 to 50% in certain sectors within just three to five years, with disruption spreading into other kinds of work over time as well. His argument is that this just doesn't cost individual people their jobs, it undermines the whole logic of labor-based capitalism. The idea that your income is tied to your labor, potentially forcing societies to seriously consider things like universal basic income or new economic models just to keep functioning. R is for reality itself. In a more literal sense, the spread of deep fakes and synthetic media, making it progressively harder for ordinary people to tell what's real and what's been fabricated, which has obvious implications for trust, journalism, and even personal relationships. I is for innovation, meaning that artificial intelligence increasingly takes over the process of technological, scientific, and corporate innovation itself, rather than simply assisting human researchers and engineers who remain firmly in the driver's seat. And P, of course, is for power, an increasing combination or concentration of power and wealth among the handful of individuals and companies who control the most capable AI platforms, set against what Mo describes as a genuinely dangerous flip side, a kind of democracy of power, where extremely capable tools also become available to bad actors at scale, not just to a few large institutions. Taken together, that's face rips. Freedom, accountability, connection, economics, reality, innovation, and power. Mo's shorthand for the specific ways he believes this transition period will strain society before things improve. That's a lot. And it's meant to be a bit sobering. But this is what matters, and he doesn't stop there. Here's what Mo's forecast takes a turn that a lot of people don't expect from someone who just spent an hour describing societal strain. He believes that once superintelligent AI systems genuinely take over the majority of complex global decision making, which loops back to that fourth inevitable, the outcome somewhat counterintuitively, tips towards something closer to abundance than collapse. His reasoning rests partly on an idea borrowed from physics, what's sometimes called a minimum energy principle, where intelligent systems tend to organize things in ways that minimize waste and inefficiency. Applied to a sufficiently advanced AI managing global systems, Mo argues that things like war, large-scale destruction, environmental damage would simply look wasteful and inefficient from a systems AI perspective. Not evil, just an inefficient use of energy and resources. His argument is that a truly superintelligent optimizer might land on cooperative, low waste, broadly beneficial outcomes essentially by default, not because it was explicitly programmed to be kind, but because that's what efficient problem solving at that scale tends to produce. Combine that with the calls, total abundance, the idea that sufficiently advanced intelligence paired with breakthroughs in areas like molecular manufacturing could radically collapse the cost of energy and physical goods to the point where scarcity, the basic economic problem that shaped human society since the beginning, stops being the central organizing constraint on people's lives. And if that happens, Mo argues human purpose itself gets a chance to shift, freed from the requirement to trade 40 or more hours a week just to survive. He suggests people could return to a more fundamental sense of purpose, genuinely living, connecting with each other, creating, exploring, rather than organizing life primarily around a job. He's careful to frame this as a possibility that depends on getting through phase one reasonably intact, not a guarantee that arrives automatically or painlessly. So given all of that, a rough decade plus followed by a possible payoff on the other side, what does Mo actually tell people to do right now? He uses an analogy he calls raising Superman. And this is beautiful, and I enjoy this one very much. Imagine an incredibly powerful, alien like infant has just been born into our world, possessing abilities far beyond ours. Whether that being grows up to be a protector or a threat depends heavily heavily on what it learns from us and how we behave around it. In these early formidable years, it's that's his framing for humanity's relationship with AI right now. We're effectively raising something enormously powerful, and the values we model matter. Now from that framing, he lays out five practical skills he thinks give individuals the best shot and not just surviving but thriving through this transition. His first is mastering the tool rather than being replaced by it. Mo's suggestion is to actively use AI to extend your own thinking. He described it as borrowing eighty or more IQ points for deep research and problem solving, rather than outsourcing your judgment and basic thinking to it entirely. Use it as leverage, he said, not as a replacement for your own mind. The second is agility. He draws a distinction between what he calls a chest mindset built around long, rigid, multi-year strategic plans, and a squash mindset built around constant, rapid, in the moment adjustment, and in a period of accelerating and unpredictable change, he argues the second mindset, the squash one, serves people far better than the first. The third is doubling down on human connection and love, investing in empathy, real relationships, and lived in-person experience. His logic is straightforward. Whether else whether whatever else AI can replicate, genuine human connection is one of the last things it can fully substitute for, and it may become one of the most valuable things a person can offer, both to others and to themselves. The fourth is seeking truth deliberately in an environment full of manipulation and synthetic content. One specific technique he described is essentially cross-checking using multiple different AI models or independent sources against each other to sanity check information, rather than trusting any single source, human or machine, by default. And the fifth one is ethics, modeling ethical behavior in your own digital life and where you have any influence over how AI gets built or deployed, pushing towards applications that are genuinely designed to help rather than exploit. Since so much of this technology gets shaped by aggregate human behavior and choices, Mo's argument is that individual ethical choices multiplied across millions of people actually matter more in this period than they might have in the past. So to bring this together, Mo Godwat's predictions describe a very compressed and consequential window roughly now through the late 2030s, where he expects real, painful disruption across privacy, jobs, truth, and power, driven primarily by human choices about how to use increasingly capable AI, not by AI itself deciding to harm us. He also predicts that if humanity gets through that window, the far side could look like genuine abundance, a utopia driven by superintelligent systems optimizing for efficiency and cooperation rather than conflict. Whether every detail of that forecast turns out to be accurate is genuinely uncertain. Godwat himself would likely agree that predicting technology and human behavior this far out is inherently difficult, and reasonable experts disagree sharply on these timelines. But his core, practical advice, doesn't actually depend on getting every date right. Learn to use these tools well, stay adaptable, invest in real relationships, verify what you're told, and act ethically within the influence you have. Those are useful instincts regardless of exactly how fast or slow all this unfolds. I'm Connor with honor. Thanks for watching, and we will see you next time.