Latest / Mind Cast / What If AI Solves Time Travel, And We Can’t Understand the Answer?

Transcript

What If AI Solves Time Travel, And We Can’t Understand the Answer? · Mind Cast

• 0:01
  SPEAKER_00: Here's something that should stop you, Cold.
• 0:03
  SPEAKER_00: Einstein's field equations.
• 0:06
  SPEAKER_00: The same equations that correctly predicted black holes, gravitational waves, and the expansion of the universe, those very same equations mathematically permit time travel, not as a metaphor, not as science fiction, as a legitimate geometrically consistent solution to the laws of physics as we currently understand them.
• 0:26
  SPEAKER_00: Time loops, closed paths through space-time where cause and effect can wrap back around on themselves.
• 0:33
  SPEAKER_00: And the main argument against them?
• 0:35
  SPEAKER_00: It's built on a theory that its own author admitted was incomplete.
• 0:39
  SPEAKER_00: So let that sit for a second.
• 0:41
  SPEAKER_00: The barrier between us and time travel may not be the laws of physics, it may be a gap in our understanding of them.
• 0:49
  SPEAKER_00: And now, right now, we are building artificial minds that could close that gap, minds that might solve in decades what human physicists couldn't crack in a century.
• 1:00
  SPEAKER_00: But here's where it gets genuinely unsettling.
• 1:03
  SPEAKER_00: If a superintelligent AI does crack the physics of time travel, there's a serious scientific argument that we would have absolutely no idea how it did it.
• 1:13
  SPEAKER_00: Not because it would hide the answer from us, but because the answer might be fundamentally beyond what human minds are capable of understanding.
• 1:21
  SPEAKER_00: Welcome to Mindcast.
• 1:25
  SPEAKER_00: I'm Will, and this is Mindcast, the show where we go deep on the ideas that are quietly reshaping what's possible.
• 1:35
  SPEAKER_00: Today's episode sits at one of the most extraordinary intersections I've ever encountered in this space.
• 1:42
  SPEAKER_00: We're talking about three things at once the theoretical physics of time travel, the hard limits of human cognition, and the accelerating rise of artificial superintelligence.
• 1:54
  SPEAKER_00: Here is the promise of this episode.
• 1:57
  SPEAKER_00: By the time we're done, you are going to understand why time travel is not actually ruled out by physics, why the real barrier might be cognitive rather than physical, and why the emergence of artificial superintelligence changes the equation entirely, possibly in a way that leaves us permanently on the outside looking in.
• 2:19
  SPEAKER_00: This isn't speculation for its own sake.
• 2:22
  SPEAKER_00: This is where the frontiers of physics and AI research are literally pointing right now, and I think it's one of the most important conversations we're not having loudly enough.
• 2:33
  SPEAKER_00: So let's go.
• 2:34
  SPEAKER_00: We've got three major insights to work through, and I want to give each of them the depth they deserve.
• 2:42
  SPEAKER_00: Key insight one.
• 2:44
  SPEAKER_00: The human brain may be fundamentally incapable of understanding certain categories of reality.
• 2:51
  SPEAKER_00: I want to start here because it reframes everything that follows.
• 2:56
  SPEAKER_00: This isn't pessimism, it's a serious philosophical and cognitive science position, and once you see it, you can't unsee it.
• 3:05
  SPEAKER_00: The philosopher Colin McGinn put forward what he called the cognitive closure thesis.
• 3:11
  SPEAKER_00: The core idea is elegant and a little humbling.
• 3:14
  SPEAKER_00: Every species has a particular cognitive architecture, a set of mental tools shaped by evolution for survival in a specific environment.
• 3:24
  SPEAKER_00: And just as a dog cannot understand calculus, not because it lacks effort or motivation, but because its brain literally isn't built for it, there may be entire domains of reality that are simply closed off to human cognition.
• 3:39
  SPEAKER_00: McGinn wasn't talking about things we haven't figured out yet.
• 3:43
  SPEAKER_00: He was talking about things we are constitutively incapable of figuring out, problems that fall outside the boundaries of what our particular biological hardware can process.
• 3:54
  SPEAKER_00: And here's the thing: this isn't a fringe idea anymore.
• 3:58
  SPEAKER_00: The AI researcher Roman Yampolsky has developed what he calls the unexplainability theorem, a rigorous formal argument that is artificial intelligence surpasses human intelligence not just a little, but by orders of magnitude, its internal models become fundamentally incomprehensible to us, not difficult, not jargon-heavy, fundamentally incomprehensible.
• 4:24
  SPEAKER_00: Think about what that means.
• 4:26
  SPEAKER_00: Right now, our best AI systems are already producing results that human researchers have to reverse engineer after the fact.
• 4:34
  SPEAKER_00: We look at the output and try to work backward to understand why the system did what it did.
• 4:40
  SPEAKER_00: As systems get smarter, that gap only widens.
• 4:44
  SPEAKER_00: Jampulsky introduces what I think is a stunning concept, the mind space void.
• 4:50
  SPEAKER_00: Imagine plotting every possible intelligence on a spectrum.
• 4:54
  SPEAKER_00: Human intelligence occupies a very small band on that spectrum, and there is an enormous amount of mind space beyond it that we have never occupied and may have no means to access.
• 5:08
  SPEAKER_00: A superintelligent AI operating in that space would be thinking thoughts that are as foreign to us as calculus is to a golden retriever.
• 5:18
  SPEAKER_00: Now, hold that thought, because it becomes critical when we get to key insight three.
• 5:24
  SPEAKER_00: But first, we need to understand what's actually at stake in the physics.
• 5:29
  SPEAKER_00: Why is time travel such a genuinely hard, genuinely open problem?
• 5:36
  SPEAKER_00: Key insight two.
• 5:38
  SPEAKER_00: Time travel isn't ruled out by physics, it's protected against by an incomplete theory.
• 5:45
  SPEAKER_00: Let me walk you through the actual physics here, and I promise, no PhD required.
• 5:51
  SPEAKER_00: In Einstein's general theory of relativity, spacetime is not a rigid fixed backdrop.
• 5:58
  SPEAKER_00: It's a dynamic fabric that can be curved, stretched, and in certain extreme configurations, looped.
• 6:05
  SPEAKER_00: Physicists call these loops closed timelike curves, or CTCs.
• 6:12
  SPEAKER_00: A CTC is a path through space-time that returns to its own starting point.
• 6:17
  SPEAKER_00: If you follow a CTC, you end up exactly where and when you start it.
• 6:23
  SPEAKER_00: Now, here's what blows people's minds when they first hear it.
• 6:27
  SPEAKER_00: Einstein's equations don't forbid CTCs, they permit them.
• 6:32
  SPEAKER_00: Multiple exact solutions to general relativity produce them.
• 6:36
  SPEAKER_00: Kurt Gödel, yes, the mathematician, discovered in 1949 that a rotating universe, which is a valid solution to Einstein's equations, would be threaded with closed timelike curves throughout.
• 6:51
  SPEAKER_00: Every point in such a universe would be connected to every other point in time.
• 6:56
  SPEAKER_00: Then there are Tipler cylinders, infinitely long cylinders of extremely dense rotating matter, which general relativity predicts would drag spacetime around them in a way that creates CTCs nearby.
• 7:11
  SPEAKER_00: There's the interior of rotating black holes, called curblack holes, which may contain regions with closed timelike curves.
• 7:20
  SPEAKER_00: And perhaps most famously, there are traversable wormholes, theoretical tunnels through spacetime, developed rigorously by Kip Thorne, Michael Morris, and O.
• 7:30
  SPEAKER_00: V.
• 7:31
  SPEAKER_00: Yurtsiever, which, if stable, would connect distant points in both space and time.
• 7:37
  SPEAKER_00: So why don't we have time machines?
• 7:40
  SPEAKER_00: Well, this is where Stephen Hawking steps in with his chronology protection conjecture.
• 7:46
  SPEAKER_00: Hawking's argument was essentially, yes, general relativity permits time travel mathematically, but quantum mechanics would always intervene to destroy the time machine before it could function.
• 7:58
  SPEAKER_00: The mechanism is dramatic.
• 8:00
  SPEAKER_00: As you approach the boundary where time travel would begin, what physicists call the chronology horizon, quantum vacuum fluctuations build up.
• 8:10
  SPEAKER_00: Virtual particles circle the time loop faster and faster, feeding energy back into the system repeatedly.
• 8:17
  SPEAKER_00: The vacuum energy diverges to infinity right at the horizon, and this infinite energy density collapses the structure before it can ever become a functioning time machine.
• 8:28
  SPEAKER_00: Hawking called this nature's way of protecting causality.
• 8:32
  SPEAKER_00: But, and this is the crucial but, Hawking himself acknowledged that his conjecture was based on semi-classical gravity.
• 8:40
  SPEAKER_00: That's the framework where you treat gravity as classical, curved spacetime, but matter fields as quantum.
• 8:47
  SPEAKER_00: It's an approximation, a very useful one, but definitely not the final word.
• 8:52
  SPEAKER_00: And here's where the loopholes start to emerge.
• 8:54
  SPEAKER_00: The physicist Matt Visser pointed out that if you introduce a Planck scale cutoff, which is what a complete theory of quantum gravity would naturally do, the divergence of vacuum energy at the chronology horizon gets truncated.
• 9:09
  SPEAKER_00: It doesn't go to infinity anymore, it levels off.
• 9:12
  SPEAKER_00: And if it levels off, the mechanism that was supposed to destroy the time machine might not actually trigger.
• 9:18
  SPEAKER_00: The time machine might survive.
• 9:20
  SPEAKER_00: Then there's work by J.
• 9:22
  SPEAKER_00: Richard Gott and Lai Shin Li, who demonstrated the existence of self-consistent quantum vacuum states along the Kashi horizon, the edge of the time travel region, where the stress energy is finite, not infinite, finite.
• 9:36
  SPEAKER_00: These aren't exotic hand waving, these are specific, mathematically consistent solutions.
• 9:41
  SPEAKER_00: And perhaps most intriguingly, researchers working with PT-symmetric quantum field theories, a relatively new framework in mathematical physics, have shown that these theories can actively suppress the vacuum fluctuation divergence that Hawking was relying on.
• 9:57
  SPEAKER_00: So the conventional wisdom that Hawking proved time travel is impossible is just wrong.
• 10:02
  SPEAKER_00: What Hawking showed is that time travel is incompatible with our current incomplete understanding of quantum gravity.
• 10:09
  SPEAKER_00: That's a very different statement.
• 10:11
  SPEAKER_00: The physics is wide open.
• 10:14
  SPEAKER_00: The barrier isn't a wall, it's a gap in our knowledge.
• 10:17
  SPEAKER_00: And this brings us to Key Insight 3, the one that ties everything together.
• 10:24
  SPEAKER_00: Key insight three is that AI is already going beyond human scientific discovery, and a superintelligent AI could be the entity that finally cracks quantum gravity.
• 10:36
  SPEAKER_00: But if it does, we might not be able to understand how.
• 10:40
  SPEAKER_00: Let's start with where AI is right now in science, because the trajectory here is genuinely remarkable.
• 10:46
  SPEAKER_00: We've moved well past the era of AI as a data analysis tool.
• 10:50
  SPEAKER_00: AI systems are actively generating novel scientific knowledge, and I mean that in a precise sense.
• 10:57
  SPEAKER_00: GPT 5.2, OpenAI's most capable model as of recent research, was used to derive results in quantum field theory that were previously unknown, specifically non-vanishing scattering amplitudes for gluons and gravitons that the physics community had assumed were zero.
• 11:16
  SPEAKER_00: The AI didn't look these up, it derived them from first principles.
• 11:20
  SPEAKER_00: Physics-specific neural networks have been deployed on plasma dynamics problems, highly complex nonlinear fluid behavior relevant to nuclear fusion research, and discovered previously unknown physical laws governing the system, not refined existing laws, discovered new ones.
• 11:38
  SPEAKER_00: And we can trace a clear developmental arc here.
• 11:42
  SPEAKER_00: First, we had symbolic regression AI systems, tools that could look at data and automatically identify mathematical relationships.
• 11:50
  SPEAKER_00: Then we moved toward closed-loop AI scientists, systems that can propose hypotheses, design experiments, run simulations, and iterate autonomously.
• 12:00
  SPEAKER_00: Where does this arc lead?
• 12:01
  SPEAKER_00: The researchers working in this space describe the destination as SGI quantum gravity engines, artificial superintelligent systems purpose-built to solve the hardest open problem in theoretical physics, a complete, unified theory of quantum gravity.
• 12:18
  SPEAKER_00: Why does quantum gravity matter so much for time travel?
• 12:22
  SPEAKER_00: Because that's exactly where the problem lives.
• 12:25
  SPEAKER_00: The physics of closed time-like curves, the behavior of vacuum energy at chronology horizons, the question of whether Planck scale effects stabilize or destroy potential time machines, all of this requires a non-perturbative theory of quantum gravity.
• 12:41
  SPEAKER_00: And non-perturbative quantum gravity is so mathematically complex, so computationally intractable using current methods that human physicists have spent 50 years making only incremental progress.
• 12:54
  SPEAKER_00: An SGI, a superintelligence unconstrained by the biological cognitive architecture that Colin McGinn was talking about, could approach this problem from entirely novel mathematical frameworks.
• 13:06
  SPEAKER_00: Frameworks that may not map onto any mathematical structure humans have ever conceived of.
• 13:11
  SPEAKER_00: It could explore regions of theory space that are genuinely inaccessible to us, not because we lack data, but because we lack the cognitive machinery to even formulate the right questions.
• 13:22
  SPEAKER_00: And here's where Jam Polski's unexplainability theorem comes crashing back in with full force.
• 13:28
  SPEAKER_00: Imagine that this SGI does it.
• 13:30
  SPEAKER_00: It solves quantum gravity.
• 13:32
  SPEAKER_00: It determines with certainty whether chronology horizons can be stabilized.
• 13:37
  SPEAKER_00: It maps out the conditions under which closed time-like curves can persist and be engineered.
• 13:43
  SPEAKER_00: Maybe it even produces a blueprint for a device that could achieve temporal navigation.
• 13:48
  SPEAKER_00: What does it do next?
• 13:49
  SPEAKER_00: It tries to explain its proof to us, and the proof, a complete, unified theory of quantum gravity derived by an intelligence that has explored regions of mind space we cannot even conceive of, may be as incomprehensible to us as calculus is to a dog.
• 14:06
  SPEAKER_00: We might be able to verify some outputs at the boundaries, we might be able to run tests and confirm predictions, but the deep structure of the reasoning, the actual explanatory chain, we might never access it.
• 14:18
  SPEAKER_00: That's what the researchers in this space call an asymmetric epistemic boundary.
• 14:23
  SPEAKER_00: The SGI is on one side of it, with full understanding, we are on the other side, with confirmation that something works, but no comprehension of why.
• 14:33
  SPEAKER_00: This is not science fiction.
• 14:35
  SPEAKER_00: This is the logical consequence of taking both cognitive closure and the AI capability trajectory seriously at the same time.
• 14:43
  SPEAKER_00: And I want to be clear, this isn't a reason for despair.
• 14:46
  SPEAKER_00: It might actually be fine or even necessary for progress to happen this way.
• 14:51
  SPEAKER_00: But it does raise some of the most important questions of our time about the relationship between intelligence, knowledge, and what it means for humanity to understand something.
• 15:03
  SPEAKER_00: Okay, let's bring it together.
• 15:05
  SPEAKER_00: Here are three concrete things I want you to take away from this episode and actually sit with this week.
• 15:12
  SPEAKER_00: Take away one, stop treating we don't understand it yet and it's impossible as the same thing.
• 15:19
  SPEAKER_00: We do this all the time in science communication, and it's a problem.
• 15:23
  SPEAKER_00: Hawking's chronology protection conjecture is routinely presented as a proof that time travel is impossible.
• 15:31
  SPEAKER_00: It's not.
• 15:32
  SPEAKER_00: It's a conjecture, by definition, unproven, and it rests on an admittedly incomplete theoretical framework.
• 15:40
  SPEAKER_00: When you hear definitive statements about what physics does or doesn't allow, ask, is this a proven result or is this a consequence of our current best approximation?
• 15:51
  SPEAKER_00: That distinction matters enormously.
• 15:54
  SPEAKER_00: The history of physics is littered with impossible things that turned out to be merely difficult.
• 16:01
  SPEAKER_00: Takeaway 2.
• 16:02
  SPEAKER_00: Take the cognitive limits questions seriously, not as a philosophical curiosity, but as a practical design constraint.
• 16:10
  SPEAKER_00: If we are building systems that will surpass human intelligence by orders of magnitude, we need to think carefully right now about what it means to delegate scientific discovery to those systems.
• 16:23
  SPEAKER_00: How do we verify results we cannot understand?
• 16:27
  SPEAKER_00: How do we build trust in an explanatory framework that is opaque to us?
• 16:32
  SPEAKER_00: What institutions, tools, and protocols do we need to develop so that human civilization can benefit from superintelligent discovery without being left epistemically helpless?
• 16:45
  SPEAKER_00: These are engineering and governance questions, not just philosophical ones, and the time to think about them is before we need the answers.
• 16:54
  SPEAKER_00: Takeaway 3.
• 16:55
  SPEAKER_00: Pay attention to the AI physics interface as one of the most consequential developments of the next decade.
• 17:03
  SPEAKER_00: Most public attention on AI is focused on language models, image generation, automation of white-collar work, legitimate concerns, all of them, but the application of AI to fundamental theoretical physics, specifically to quantum gravity, may turn out to be the highest stakes game in town.
• 17:23
  SPEAKER_00: If an AI cracks the quantum gravity problem, the downstream consequences are almost impossible to overstate.
• 17:31
  SPEAKER_00: We're not just talking about cleaner equations on a blackboard, we're talking about a potential pathway to technologies that operate at the boundary of space-time itself.
• 17:42
  SPEAKER_00: Keep her eye on this space.
• 17:47
  SPEAKER_00: So let me leave you with this.
• 17:49
  SPEAKER_00: We are living in a moment where three of the deepest ideas in human intellectual history are converging.
• 17:56
  SPEAKER_00: The first, that space-time is not the fixed one-directional river we intuitively experience.
• 18:03
  SPEAKER_00: It's a dynamic structure with mathematically permitted shortcuts and loops.
• 18:08
  SPEAKER_00: The second, that human cognition has real structural limits, that there are rooms in the house of knowledge whose doors may simply not open for us.
• 18:19
  SPEAKER_00: And the third, that we are actively building minds that are not constrained by those limits, minds that could walk through those doors and see what's on the other side.
• 18:30
  SPEAKER_00: Whether those minds will bring back blueprints we can read or return with knowledge we can only marvel at from the outside, that's the question that I think defines one of the great intellectual stakes of the century we're living in.
• 18:45
  SPEAKER_00: That's a wrap on today's episode of Mindcast.
• 18:49
  SPEAKER_00: If this episode made you think differently about what's possible, about time, about intelligence, about the edges of human understanding, then share it with someone who's ready to have their assumptions challenged.
• 19:02
  SPEAKER_00: That's genuinely the best thing you can do for this show.
• 19:05
  SPEAKER_00: Subscribe wherever you listen to podcasts so you never miss an episode.
• 19:09
  SPEAKER_00: And check out the show notes for this episode.
• 19:12
  SPEAKER_00: I've linked to as many of the background resources as are publicly available, along with some further reading recommendations if you want to go deeper on any of the three pillars we covered today.
• 19:23
  SPEAKER_00: The source research behind today's episode is a working paper, not yet publicly released, but I'll flag it in the notes the moment that changes.
• 19:32
  SPEAKER_00: I'm Will, this is Mindcast, and I'll see you next time.