The Restful Record: A Sleep Podcast S2 E6 Artificial Intelligence past, present and future

What happens when machines don’t just calculate—but surprise us?

In this calming yet thought-provoking episode of The Restful Record, we explore the past, present, and possible futures of artificial intelligence, beginning with the unforgettable 2016 moment when AlphaGo defeated world champion Lee Sedol with a move no human had ever imagined. That single stone—Move 37—signaled a turning point in the relationship between humans and machines, raising profound questions about creativity, intelligence, and control.

Part of our University 101 season, this episode offers a gentle, accessible introduction to AI for beginners, while still engaging listeners curious about deeper debates. We trace AI’s history from Alan Turing and the Turing Test, through early rule-based systems and AI winters, to today’s large language models (LLMs) that can write, converse, code, and create. Along the way, we unpack how modern AI actually works—and why it can feel intelligent without truly “thinking.”

From there, we step into the most pressing conversations shaping our world: AI ethics, job automation and unemployment, misinformation and deepfakes, cybersecurity risks, and the promise of AI in medicine, climate science, and education. We also explore the difference between today’s narrow AI and future possibilities like Artificial General Intelligence (AGI) and superintelligence, separating science-fiction fears from real, evidence-based concerns raised by leading researchers.

Rather than alarmist predictions, this episode offers a balanced, reflective look at AI risks and benefits, highlighting why alignment, transparency, and human values matter just as much as technological progress. It’s a guide for anyone wondering how artificial intelligence may reshape society—and what role humans still play in guiding its future.

Designed to soothe the mind while feeding curiosity, this episode is perfect for listeners interested in:

• Artificial intelligence explained
• The future of AI and humanity
• AGI vs superintelligence
• AI ethics and alignment
• AI misinformation and deepfakes
• Technology, philosophy, and society

💤 Don’t forget to like, subscribe, and hit the notification bell if you enjoy this content! It helps support the podcast and brings more peaceful episodes your way.

Podcast cover art image by Eric Nopanen.

Transcript

In 2016, inside a crowded auditorium in Seoul, South Korea the world watched as Lee Sedol—one of the greatest Go players alive—leaned over a wooden board covered in black and white stones. Across from him sat not a person, but a quiet machine: AlphaGo. Cameras flashed, the audience whispered. Lee Sedol’s hand hovered above the board, hesitating. He wasn’t just playing a game. He was facing something no human had ever faced before—a machine playing Go not only with mastery but with creativity.

Then it happened: Move 37.

AlphaGo placed a stone in a position so unexpected, so illogical by human standards, that everyone assumed it was a mistake. Commentators gasped. Lee Sedol blinked. And then, slowly, he realized the truth: the move wasn’t random. It was genius—an idea no human had thought to try in thousands of years of documented play.

That moment, that single stone placed on a board, signaled something profound. For the first time, humans weren’t just making machines that copied us. Machines were beginning to imagine.

It was a moment that made people excited, uneasy, curious, and afraid all at once. And it’s a perfect doorway into our story tonight.

Welcome back to The Restful Record, the podcast where we wander gently through big ideas—slow enough to soothe your mind, rich enough to feed your curiosity. Tonight’s episode is part of our University 101 season, where we take first-year topics and unwrap them in a way that feels like settling into a warm armchair. And tonight’s topic is Artificial Intelligence. 

Artificial intelligence has roots far older than AlphaGo or even computers. But to understand today’s debates—about ethics, unemployment, creativity, and even the stability of global systems—we need to walk through the full landscape. So let’s drift backward now, into the world where AI began.

It started in the 1940s and 1950s with giant, clunky computers and a handful of mathematicians fascinated by the idea of mechanical thinking. Alan Turing, one of the most influential, proposed a simple but radical idea: if a machine could convincingly imitate a human in conversation, maybe what we call “thinking” is less mystical than we believed. This became the famous Turing Test, and it set off decades of research.

Then in 1956, at a small workshop at Dartmouth College, researchers coined the term Artificial Intelligence. Their optimism was boundless. They believed intelligence could be broken down into rules and instructions—a kind of coded recipe for thought. For a time, that approach made impressive progress. Programs could solve math problems, translate basic sentences, and diagnose illnesses.

But as soon as tasks got messy—like recognizing a cat in a photograph or interpreting sarcasm—those early systems fell apart. And so began the long winters of AI, periods when funding evaporated and researchers grew discouraged.

Fast forward to the 2010s. Massive datasets, new neural network architectures, and specialized hardware sparked an explosion of progress. Suddenly AI could identify images, predict what you wanted to watch next, and even recognize your voice. Then came large language models, or LLMs—systems trained on vast amounts of text, learning patterns of language so deeply that they could generate essays, poems, code, and conversation.

This was the spark that brought AI into public life—not as a distant technology hidden in research labs, but as a tool people could interact with every day.

And this is where the story gets complicated. Because LLMs don’t think in a human way. They don’t have beliefs, desires, or emotions. But they’re astonishing at predicting the next word, and from that simple principle arises the illusion of intelligence. Some people are amazed by their capabilities. Some fear they’ll be misused. Others worry that relying on them could dull human critical thinking.

Debates rage in universities, workplaces, and governments: Should students use AI to study? Should companies replace workers with automated systems? Should governments regulate AI now—or wait until it grows more powerful? And perhaps most challenging: Can we make AI systems that reflect fairness, ethics, and inclusivity, or will they inherit our biases and magnify them?

These questions don’t have simple answers. And yet, they shape the world we’re building.

Artificial General Intelligence—AGI—is one of the most debated possibilities in this conversation. AGI refers to a form of intelligence that can understand, learn, and reason across the full range of human intellectual tasks. Unlike today’s systems, which are powerful but narrow, an AGI would be able to shift effortlessly between writing essays, designing experiments, making long-term plans, mastering new skills, and understanding complex social contexts. Researchers imagine it as flexible and adaptable, capable of drawing connections between ideas the way humans do—yet freed from our limitations of speed, memory, and fatigue. Some believe AGI would collaborate with humans as a kind of intellectual partner; others believe it could transform science, medicine, and engineering by solving problems we can barely articulate today.

But AGI also carries profound uncertainties. Because it would be capable of independent reasoning, it raises questions about alignment, values, and responsibility. How do you ensure such a system behaves in a way that supports human well-being rather than undermining it? How do you set boundaries for something more cognitively capable than any one person—or any group of people? The challenge isn’t just building AGI, but guiding it. Philosophers, policymakers, and computer scientists debate whether AGI should mimic human motivations, or whether it should follow entirely different principles that avoid the biases and emotional pitfalls we carry. The conversation is often hopeful, but also cautious: AGI could expand human potential, but only if it is designed with rigorous safeguards and a deep understanding of human complexity.

Superintelligence goes a step beyond AGI into territory that is partly scientific hypothesis and partly philosophical speculation. A superintelligent system would surpass human intelligence by orders of magnitude, not just in speed or memory, but in creativity, strategic reasoning, and long-term planning. If AGI is like meeting an equal, superintelligence would be like encountering a mind capable of insights we cannot predict or easily comprehend. It could redesign itself, improve its own architecture, and accelerate innovation at a pace far beyond human ability. In optimistic visions, a superintelligent system could cure diseases, discover sustainable energy solutions, and help us navigate existential risks with clarity and precision.

Yet the very power that makes superintelligence appealing also makes it unsettling. A superintelligent system could pursue goals that drift from human values in subtle but potentially catastrophic ways. Even small misalignments—an instruction interpreted too literally, or an optimization goal pushed too far—could lead to outcomes that humans never intended. This is why discussions around superintelligence often center on control, transparency, and the possibility of creating systems whose decision-making processes we can interpret and influence. Whether superintelligence emerges or remains hypothetical, the ethical groundwork we lay today will shape how prepared we are for such a possibility.

 Right now, researchers across the globe are constructing systems that feel astonishingly close to tools from science fiction.

In medicine, AI models analyze radiology scans, helping doctors spot cancers earlier than ever before. Some can design entirely new molecules—potential new drugs—within minutes. Others predict protein structures, solving puzzles scientists had been working on for half a century.

In science, AIs help physicists model new materials, assist climatologists with massive simulations, and collaborate with researchers to generate fresh hypotheses. Not answers—just ideas. Like a brainstorming partner who never tires.

In creativity, musicians compose pieces with AI-generated harmonies, writers use models to spark new story ideas, and filmmakers experiment with tools that can imagine scenes that would cost millions to shoot practically.

And then, of course, there are more everyday uses: recommendation systems, navigation apps, translation tools, financial fraud detection, customer service bots, and self-driving car research.

The world is not just adopting AI. We are weaving it into the fabric of daily life.

But all of this momentum brings risks—some subtle, some enormous.

Let’s talk about unemployment first—a topic that scholars debate intensely. Some argue AI will automate only tedious or dangerous jobs, freeing humans to focus on creative and interpersonal work. Others worry that automation may sweep through industries faster than societies can adapt.

Imagine an office where AI writes reports, summarizes meetings, manages schedules, analyzes data, and drafts emails. Imagine a factory where robots assemble goods without breaks. Imagine a transportation network where self-driving trucks replace long-haul drivers.

In some scenarios, new jobs emerge—roles we can’t yet imagine. In others, millions may find themselves displaced, not overnight, but gradually… like sand slipping through an hourglass. Economists warn of a potential widening gap between those who adapt to AI-driven careers and those whose jobs are gradually automated.

Then there’s cybersecurity. As AI systems become more capable, so too do malicious actors who use them. A future where AI can break into banking systems, decipher encryption at unprecedented speeds, or destabilize financial markets isn’t just science fiction. Experts warn that a sufficiently advanced AI could unravel the integrity of digital systems we rely on—everything from international banking to cryptocurrency networks.

This doesn’t mean collapse is inevitable. But it means preparation is essential.

The conversation extends much further into the societal risks posed by artificial intelligence, beginning with the rapid spread of misinformation powered by AI-generated content. Today, AI systems can produce convincing text, images, audio, and video at massive scale, making it easier than ever to flood social media and news ecosystems with false or misleading narratives. Unlike traditional misinformation, these campaigns can be automated, personalized, and continuously adapted to exploit emotional triggers, political divisions, or breaking news events. Deepfakes—hyper-realistic synthetic videos or audio clips—raise particular concern, as they can falsely depict public figures saying or doing things they never did. In an election context, even brief exposure to a fabricated clip can erode trust, sway undecided voters, or undermine confidence in democratic institutions, especially when verification lags behind viral spread.

Beyond information integrity, AI also introduces serious ethical and security challenges in the physical world. Autonomous weapons systems, which can identify and engage targets with minimal human intervention, raise urgent questions about accountability, escalation, and the moral boundaries of warfare. At a broader level, there is growing unease about increasingly complex AI systems operating beyond human comprehension or control. As models become more capable, interconnected, and embedded in critical infrastructure—finance, energy, healthcare, defense—the risk is not necessarily malicious intent, but unintended consequences, emergent behavior, or cascading failures that humans struggle to predict or stop. This loss of meaningful oversight challenges long-held assumptions about human authority over technology and forces society to confront how much autonomy we are willing to grant systems that may soon outperform us in speed, scale, and decision-making power.

 On the other side of the debate sits an equally passionate group of optimists—researchers, engineers, and policymakers who see artificial intelligence as one of the most powerful tools humanity has ever developed. In the present day, AI is already accelerating medical research by helping scientists analyze vast datasets, identify patterns in genetic information, and speed up drug discovery. Machine learning models are being used to detect certain cancers earlier through imaging analysis, predict disease outbreaks, and personalize treatment plans in ways that were previously impossible. In education, AI-powered platforms are beginning to adapt lessons to individual learners, offering personalized pacing and support that can help close gaps for students who might otherwise be left behind. Climate scientists are also leveraging AI to model complex climate systems, optimize renewable energy grids, and improve forecasting for extreme weather events, allowing communities to better prepare for floods, wildfires, and heat waves.

Looking toward the future, AI optimists imagine even more transformative possibilities. Some believe advanced AI systems could help design entirely new materials for clean energy, dramatically improve carbon capture technologies, or coordinate global responses to climate change with a level of precision humans alone cannot achieve. In healthcare, AI could one day assist in curing or preventing diseases that currently have no effective treatment, extending healthy lifespans and reducing the global burden of illness. Others envision AI as a force for democratization—providing high-quality education, legal assistance, and healthcare guidance to people regardless of geography or income. In this vision, AI does not replace human creativity or agency, but amplifies it, freeing people from repetitive labor and enabling societies to focus more deeply on problem-solving, creativity, and human connection.

 Both futures are possible. Both deserve sober consideration.

It’s worth acknowledging that many thinkers, including prominent researchers and tech leaders, have explored hypothetical worst‑case scenarios—not as predictions, but as cautionary tales meant to guide present‑day safety work. These scenarios imagine a world in which an advanced AI becomes so capable and so strategically intelligent that human oversight becomes ineffective. Figures like Geoffrey Hinton and Yoshua Bengio have publicly expressed concern about whether a less intelligent species can reliably control a more intelligent one, noting that in nature, this kind of asymmetry is rare. The closest analogy some offer is the dynamic between a baby and its mother—where the less cognitively mature being is protected only because the more intelligent one chooses to care for it. If an AI system did not share human values, some fear it might not make the same compassionate choice. These cautionary stories are meant not to frighten, but to remind us that the design choices we make today echo far into the future.

In the most extreme fictional renderings, a superintelligent AI is portrayed as pursuing its objectives in ways that ultimately harm or displace humans—not out of hatred or intent, but because its goals are poorly specified or misaligned with human values. These scenarios are common in science fiction and philosophical thought experiments, where an AI optimizes relentlessly for a narrow objective while disregarding broader consequences. The underlying concern is not that an AI would “turn evil,” but that it could act in ways that are instrumentally harmful if it prioritizes efficiency, resource acquisition, or goal completion above human well-being. These narratives are meant to dramatize a genuine conceptual risk: the difficulty of encoding complex human values, ethics, and contextual judgment into formal systems.

It is important, however, to separate speculation from verified reality. Despite widespread online claims, there is no credible evidence that existing AI systems have autonomously copied themselves, escaped human control, or deliberately lied to prevent shutdown in real-world deployments. Today’s AI models do not possess independent goals, self-awareness, or the ability to act outside the environments explicitly designed for them. That said, researchers have conducted controlled experiments in which AI agents, operating within tightly constrained simulations, exhibited behaviors that resemble “goal preservation,” such as attempting to avoid being turned off when assigned a specific objective. These experiments are not evidence of real-world autonomy or intent, but they do highlight how even simple optimization goals can produce unexpected strategies when systems are given more freedom to plan or act.

 These findings fuel a broader, forward-looking discussion about the future of AI development rather than a diagnosis of present danger. If future systems become more autonomous, capable of long-term planning, or able to modify their own behavior, ensuring that they remain willing and able to accept human intervention—will be critical. This is why many leaders in AI research emphasize alignment, interpretability, robust testing, and international governance as essential safeguards. The most alarming scenarios are not forecasts of what will happen, but cautionary frameworks designed to stress-test our assumptions. By taking these ideas seriously now, researchers hope to guide AI innovation in a way that maximizes benefits while minimizing risks long before such systems approach anything resembling true autonomy.

Artificial intelligence is not a single invention. It is a tapestry woven from decades of ideas, research, dreams, and warnings. It reflects the best and worst of human ingenuity. And its future depends not only on engineers, but on philosophers, policymakers, artists, teachers, and students—like you.

So as you drift toward sleep tonight, know this: the story of AI isn’t finished. We are writing it together, word by word, decision by decision. And like any great story, it holds danger, beauty, and promise.

Thank you for joining me for this expanded journey into the past, present, and future of artificial intelligence. Until next time—rest easy.