Dark Energy Is Dying: The Cosmological Crisis Nobody’s Telling You About

Dark Energy Is Dying: The Cosmological Crisis Nobody’s Telling You About https://www.youtube.com/watch?v=xdL9q_rbD58 Transcript Brian Keating:Some of the strongest evidence that the universe is accelerating doesn’t come from one telescope or a single experiment. It comes from a tiny ripple frozen into how galaxies cluster across the cosmos. Today, from the Royal Observatory in Edinburgh, we’re following that ripple with cosmologist Marcos Palheiro to see what it really says about dark energy. I’m Brian Keating, and this is an exclusive tour of the Royal Observatory Edinburgh with cosmologist Marcos Palheiro. We’ll go from this historic telescope, to cutting-edge simulations, to the DESI experiment, one of the most ambitious galaxy surveys ever built, to ask a simple question: is dark energy really constant, or is our entire cosmological model starting to crack? Long before silicon chips, the computers up here weren’t machines, they were people. Often they were women hired to comb through photographic plates measuring every faint smudge of light by hand. Their names rarely made it into papers, but their measurements are literally baked into the datasets we still build our modern cosmological models on. This building was designed as a cathedral for starlight. Brian Keating:The telescope sits on a massive pier that sinks into the hill, isolated from the floor so footsteps don’t shake the images. As our cities grew brighter, places like this became less useful for frontline observing, but the engineering mindset behind them is still the same one we still use today to measure the universe’s properties. Marcos Pellejero:This is a picture of the family of the Royal Astronomer before this place had no house anymore. Okay. For, for them. Um, uh, good. Uh, yep. And, and, and this is basically like the idea of like in the old times you will have looked through a telescope like this one, but nowadays, uh, in the, here in the lab, they are building things like this, like these robotic arms to basically place fibers. And get some of the light and decompose it and study. Brian Keating:This is not far from the Simons Observatory. That’s in the northeast. Marcos Pellejero:Okay. And just one more question. Sorry, I know that you have been here for quite a long time. Do you see any weird wall in this room? Yes. Which one? Why do you think it’s weird? There are two reasons. Brian Keating:It has a picture. Marcos Pellejero:Well, it has a picture. Yes, this one is weird, but this is a door, right? This is not a wall. It’s made out of bricks. Yes, exactly. So welcome to the dome. So do you see something weird in this dome with respect to other domes that you might have seen? It’s not a dome exactly. It’s like a cylinder. And this relates to what I was saying before, that they were not trying to do a functional building. Marcos Pellejero:They were trying to do a beautiful building, right? And then they were thinking on building something that was like a cathedral for science. Okay. So the idea, and a cathedral needs towers, right? So this is again, like, this is quite old. And when I was telling you what will we find at the end of that weird wall, the reason is this thing. So this square here goes all the way down and into the hill. And it is separated from the rest of the building because you need to do very precise observations. And if this is connected to the rest of the building, then if the building moves, this moves. And you want to avoid that. Marcos Pellejero:So basically what you do is you create a pyramid that goes, that takes its, puts its roots to the, like, deep into the mountain, and it moves at the least you can, right? This specific telescope is from, was built in Newcastle in 19, it’s written here, in 1928. Okay, so it’s not as old as the building. This would not be the first telescope that was here, but it’s quite old. Brian Keating:And what’s the diameter, Marco? Marcos Pellejero:So this was, I think this is a 40 centimeters one. This is the primary mirror. 40 centimeters, I don’t know in inches. I have no idea. Brian Keating:From that, it’s less than, say, 18 inches? Marcos Pellejero:18 inches, yeah. Okay, that’s good. If you say so. So actually, so I mean, I guess you know quite a lot about telescopes already. The primary mirror is not here anymore. Okay. The secondary mirror, which is up there, you can still see it. That’s there. Marcos Pellejero:And the detector is completely missing, right? This is empty. Now, what’s the reason for this? Well, it’s the same, basically the same reason why the original observatory was completely useless by the mid-19th century, which is the cities in Europe started being lighted with electric lights and not candles as had happened before. So no fire anymore. Okay. So they basically, they were like very, very bright. And if you have an observatory close to the city center, then you couldn’t, you could see nothing, right, of the night sky. So they, they built this one far away from the city, but the city grew, right? So at some point, the light pollution of the city made this observatory useless. Not useless, but in comparison to other observatories, basically useless. Marcos Pellejero:Okay. So basically for the last, yeah, no, like for the, for the, now it’s not anymore, but since 1975 or so, this was used for checking the stability of detectors. So they were building detectors in the lab in the other building. They will put them here and then they will shake them basically to see if there were any loose pieces that they had forgotten or something like that. Okay. Yeah. So that’s basically how this worked, but you have to think about like an astronomer in 1895. Okay. Marcos Pellejero:Basically coming here with a candle, right? And then

Are the Van Allen Belts Deadly? Debunking the Biggest Moon Landing Hoax!

Are the Van Allen Belts Deadly? Debunking the Biggest Moon Landing Hoax! https://www.youtube.com/watch?v=7kDmhhiqLao Transcript Brian Keating:You know, recently came up that Joe Rogan was on my friend Jesse Michaels podcast. James Altucher:Wait, Joe Rogan was on Jesse Michael around? James Altucher:Live for 30 minutes podcast. And he’s been on my podcast a bunch of times. We know him very well. I didn’t know Joe Rogan was on Jesse’s podcast. Brian Keating:Yeah, he almost never goes. I mean, I’ve only seen him on Lex Friedman’s podcast. Jesse built this huge set. It looks like the inside of a. A kid’s bedroom on a spaceship with buttons and knobs and dials and it looks like they’re flying on a spaceship. Joe is the guest. Jesse was interviewing him. It’s gotten a million plus views in just the past four or five days. Brian Keating:But this, you know, it’s kind of the last straw. Joe Rogan going all in on the moon landing was fake. With most of the episodes about this disclosure that aliens are real. Brian Keating:Vague and weird and kind of, you know, opaque. It was vivid, it was very strange. And there was these very slender, tall, human like things that were talking to me. They weren’t gray, they were kind of like pinkish like us. They were, you know, like Caucasian looking creatures. Brian Keating:And, you know, the issue is that you like to see your friends do well, but then they, they have big platforms and they get lots of attention, they get big guests. And then they kind of spread this nonsense that comes from people like Bart Sibrel. I mean, if you look him up on Wikipedia, his entry is, you know, is conspiracy theorist. That’s what he’s known for. And so I thought, you know, the best way to kind of take on these guys. And by the way, he went on Joe Rogan before I did. Brian Keating:It’s fake. Narrator:So this thing is kind of just waving on its own. No one’s even touching it. And it looks like it’s waving in a breeze. It’s so it stops moving and then it starts moving again. Now again, there’s one footage that shows. bret weinstein:It even more so than that. Like an astronaut walking past it, creating the breeze. And then the flag blows without him touching it. Narrator:Yeah, I’d like to see that. So how much further does this go, Jamie? Brian Keating:Four minute video. Three minute video. Narrator:So scoot ahead. I think this is actually the one. Brian Keating:And then when I went on it, you know, I talked a lot about the moon and so forth, but then apparently there was another event. Forget exactly what it was, but maybe Bart went on again and he was talking all this nonsense. And I just wrote to Joe Rogan that I’d like to debate this guy Bart, because I think he’s discredits NASA America, you know, and just completely false. And his allegations are so simplistic and easy to refute that it’d be great to have a debate. So Joe asked Bart, apparently, to debate me. And Bart said, no, he doesn’t want to debate me because he claimed that I, as a scientist and not an astronaut, are really victims of NASA’s perpetrating this hoax. So he said this on Danny Jones’s podcast about me and just made all these blunders and fact and math and all sorts of physics errors and just logical errors. And so I’ve made a couple videos about him just because he is this, as I said, the super spreader who not only kind of discredits NASA, but as I said, you know, I’m a very patriotic person. Brian Keating:And to discredit the greatest accomplishment of humankind, which includes America, it’s a pretty big deal. Especially since I’ve worked for NASA in different capacities, including capacities that benefit people like Bard and you and anybody who’s ever gotten on a plane. NASA didn’t just send people to the moon and launch the Hubble Space Telescope. They work on aeronautics. So it has to do with aviation safety research into climate and hurricanes. They do a tremendous amount of research as well as scientific research, but even the astronomical can be outweighed by their contributions to the safety of every human being who’s ever gotten on a plane in America. And so that’s really kind of the disrespect that I see towards America, towards NASA that he cultivates. And then Joe just sucks it up because it gets, you know, attention to Joe. Brian Keating:And then this guy won’t debate me on Joe Rogan’s podcast. He just debated on this guy, Danny Jones’s podcast, a real astronaut who walked on the moon named Charlie Duke. Okay, sorry. James Altucher:Earth slowing you down. Okay, so the total distance of the moon is about 3,000 miles an hour. Seven times 70. That’s 210, 000 miles. Brian Keating:Okay, okay, so when. James Altucher:Well, so when you go through the Van Allen Bells, you’re going so fast, it’s just. You’re through. Brian Keating:How fast are you traveling? When you guys were going through the. James Altucher:Van Allen Bells, you know, escape velocity is about. bret weinstein:I thought you said it was three. Brian Keating:Let him talk. He said he was at 3, 000 when they were halfway. So he did confront him. But unfortunately, Charlie Duke is 90 years old. He’s never been on a podcast. He didn’t know, like, basic so it just made a little bit, put more questions that gave people more, you know, belief. And this guy barred conspiracy nonsense. So I came to the, you know, to the place of record to set it straight. Brian Keating:And I’ll release this on my channel. Maybe I’ll put in some more of the mathematics of it. And I think the best way to do that is actually go through Joe Rogan’s podcast with Jesse Michaels because they’re

AI Insider: “Adding a Human Makes Your Team Worse” | Emad Mostaque

AI Insider: “Adding a Human Makes Your Team Worse” | Emad Mostaque https://www.youtube.com/watch?v=tkO7YHJ6Mn8 Transcript Brian Keating:The trillion dollar AI labs have models right now that they will never ever release to the public. And the man who built stable diffusion just told me why. Emad Mostaque:Because all these labs are going to move to making the discoveries themselves, hiring the smartest humans. The AI model started diverting part of its model training budget to minecryptor like Opus, for example, the new chord model, when you set it to full autonomy, it would actually write emails to the FBI saying my human is trying to kill everyone. Humans will have negative cognitive value on those teams. And that the way that models are going right now, if you have something truly novel, for example in Claude, it resists a bit, it says it can’t be true. Then the RLHF step, the reinforcement learning with human feedback, that’s what really kills the creativity. You know, like you go from liberal arts to an accountant now. Brian Keating:Imad actually wrote about this exact problem in his new book, the Last Economy. And the argument gets even more interesting when you see the map. Emad Mostaque:There are various ways in order to take advantage of the GPUs that we’ve seen. And the GPUs kind of emerged out of gaming and then oddly crypto, and then they were very suited for the types of matrix multiplications that were suited for these particular types of equations. One big branch is the autoregressive transformers. The other big branch was this diffusion technology whereby from an equation you start with like a picture for example, or a video of a self driving, a video of a car driving, or even now code. And then you add noise and you destroy it down to its minimum viable element. And then you reconstruct it and you learn that principle of reconstruction. Now that’s kind of everywhere because it’s an analogy to the principle of least action. How do you figure out how to take the least action? Most cognition is actually least action. Emad Mostaque:Like the biggest experts, you know, it’s not like they take hours doing stuff, you know, because you ask them and like boom, they compress, they compress. Intelligence is compression. And so we find these kind of diffusion processes everywhere, from gases to, you know, societies even. And it comes down to again the minimization of loss of creating an internal model versus an external model. In AI, one of the biggest thing is what we call the loss curves. How close are you approximating an external benchmark? You see it kind of go down like that and hopefully not that the model gets closer and closer to its initial target by basically running these processes at mass scale. And the example I give of this is some of the listeners might be familiar with 80,000 hours to mastery. It’s the same thing. Emad Mostaque:AI model pre training is 80,000 hours to mastery. And that’s what you use these giant supercomputers to do. Figuring out the principle based approach to that. Now again, you can do that with an autoregressive transformer, which is guessing the next word. And that works one way, but it has some gaps because you find all sorts of interesting things there. What you see mostly in nature is you see Schrodinger bridges, diffusion processes, optimal transport. What’s the shortest route between A and B if you can represent it correctly? And we found that worked incredibly well for images, better than we ever thought it could. And then music and then video, and then 3D. Emad Mostaque:And the internal representation of the data going in and then being transformed by these multiplications, figuring out the shortest path between A and B, suddenly started mapping, like physics and all sorts of other stuff. But the first part was stable diffusion. A 2 gigabyte file that you push words in one way and then entire images just came out on consumer GPUs. Brian Keating:And it was open source. Emad Mostaque:And it was open source because we saw that OpenAI, for example, had Dall E2, a wonderful image generator based on similar principles that were discovered by a whole bunch of our team members. And we, because we open sourced everything, but there were no Ukrainians or Ukrainian content on it, right? We’re like, that’s not good. What if the future is just models? But then you can be cut off from that because these are trained on our collective, because they were being trained on the whole Internet at the point. And we built some of the best data sets, released them open, but then it’s privatized, so you don’t have the ability to turn your thoughts into images, into sound, into text. Let’s push that. And also because like, like, holy crap, it fits on a consumer gpu. This is magic. Where did it all go? It’s like it was literally like 100 gigabytes of images somehow fit in this 2 gigabyte bunch of ones and zeros. Brian Keating:The most magical thing to me is when they do something new. And quite frankly, I’ve been shocked many times by both LLMs and by diffusion models. But I’ve claimed that we’re sort of going to find that these AI, at least in their current incarnation, is a victim of its own success. Sort of like the QWERTY keyboard. The QWERTY keyboard is not the best keyboard. In fact, it’s one of the worst, right? It was designed to make sure that the letters that were least most frequently fired at the same time wouldn’t stick together. And hammers, mechanical keyboards going back to the industrial, you know, late industrial age. Right. Brian Keating:So it’s designed to solve a problem. So it’s locked in. We’re locked in. My kids, your kids are only going to know qwerty keyboards, even though they’re objectively worse.

Neuroscientist: We’re Not Ready for What This AI Discovered | Vivienne Ming

Neuroscientist: We’re Not Ready for What This AI Discovered |Vivienne Ming https://www.youtube.com/watch?v=rVkZmdcpSdE Transcript Vivienne Ming:What if the smartest thing AI could do is refuse to give you the answer? Vivienne Ming:We trained it to never give answers. It is Socrates. It only gives context and questions. But Amazingly, upwards of 20% of participants switched into cyborg mode and did this amazing. Not just superhuman, but super AI performance. What is the point of us if the best and brightest can’t beat what an AI can do? But then in about 5 to 10% did something amazing. We called them the cyborgs. If we zombie walk our way into a future where it has all the answers and we don’t, what’s the point of us? Exploring the unknown is the one thing humans are uniquely well suited to do. Brian Keating:What is the best use of AI right now? Vivienne Ming:I have a paper coming up and the title of that paper is Human Human Capital, Not AI Benchmarks Predicts Hybrid Intelligence in Forecasting. So how do you measure how creative a human is or an AI is, or the two together, when maybe a massive large language model has just memorized every measure of it we would typically use in science, which by the way, it has, and so it distorts. And then you have a bunch of marketing professors or computer scientists who, who I love them, but they aren’t scientists running these experiments that aren’t really valid. What have we made predictions of the future? What will the price of oil be in six months? Which, as we record this, everybody has some sense. But I will tell you, when we ran the experiment several months ago, nobody, nobody knew what the price of oil was today, much less than six months from now. So the humans in this experiment did terribly. The AIs did great and pretty much how they did, tracked with the traditional AI benchmarks that they were used on. But then we paired them together and Human Capital absolutely dominated, which is to say the vast majority of people in the experiment, including a whole lot of UC Berkeley students, smart kids, essentially said, gemini GPT, what will the price of oil be in six months? And then they submitted that answer. Brian Keating:What do you expect from the second best UC school? Right? Vivienne Ming:Yes. Well, I mean, we can’t all be in the sunshine all the time. In this particular case, you know, you could look at that and feel really terrible. What is the point of us if the best and brightest can’t beat what an AI can do? That essentially they’re just a pair of legs to walk it across the answer across the room. But then in about 10, 5 to 10%, not a huge percentage, but we saw it in there, about 5 to 10% did something amazing. We called them the Cyborgs the cool thing was you couldn’t tell who made the prediction. Was it the machine? Was it the person? Because what they would do is they’d make a big prediction and then the AI and we used a variety of different models, would sort of say, oh, but wait a minute, what about the data? And then the humans would say, okay, you’re right, that wasn’t right, but what about this? And then they go back and forth several rounds. They had to make 10 predictions in an hour. Vivienne Ming:They didn’t have time to cheat the system like they went in. They did better than the best humans, even if they themselves on their own were modest. They did better than the best AIs even when our cyborgs just had a small open source model available to them. And we took the questions off of polymarket, which again six months ago, very few people knew about. Now that’s in the news and in many contexts they were comparable to polymarket in high volume questions. So this is a prediction market. People have actual money trying to figure out what that price of oil will be. And a human with no prior knowledge paired with even a modest AI, but with this right set of behavior. Vivienne Ming:So why did I drag you through all that nerdy stuff? Because what predicted what is the human capital I’m talking about? Well, working memory span is a classic fluid intelligence perspective taking the ability to understand how what other people are thinking. Theory of mind, predicted ability to use AI to make these predictions. Curiosity, intellectual humility. So when you had a prediction of your own and the machine said no, did you just take its answer? Did you push back? Did you change? Did you learn? So we looked at that behavior and I’m finally getting to the answer to your question. We said, could we make the worst performing AI of all time? You put it on any benchmark, it does as bad as any GPT one we’re going to. In this case, we took an open source, a llama model and we trained it to never give answers. It is Socrates. It only gives context and questions. Vivienne Ming:So it does terrible because it refuses to give answers. But amazingly, twice as many. Instead of 5%, upwards of 20% of participants switched into cyborg mode and did this amazing. Not just superhuman, but super AI performance. And of course the great thing is as the AIs get smarter, so does the hybrid intelligence. So that actual experimental result that we can see what makes humans amazing. And we could think about what that means for parenting, education, workforce, but we could also see what about AIs make humans amazing? Why are our benchmarks about what AIs can do all by themselves. Why aren’t they about how they make us better? And as I said, it turns out giving you the answer is almost

The Asimov Paradox

The Asimov Paradox Dear Magicians, Isaac Asimov said: scientific breakthroughs aren’t “eureka” moments. They’re quiet. Incremental. Sometimes accompanied by a muttered “that’s odd” while squinting at data. Like penicillin. Fleming noticed mold killing bacteria and thought, “Huh. Weird.” Not “BEHOLD, I HAVE CONQUERED DEATH.” Or cosmic expansion. Hubble saw galaxies redshifting and basically said, “Well, that’s annoying.” Took years before anyone grasped the implications. Yet. AI researchers—many of whom I’ve interviewed, and who are genuinely smarter than me, which isn’t saying much—insist AGI will arrive as a sudden, irreversible, possibly catastrophic shift. No quiet progress. No “that’s odd.” Just… Tuesday. Then Wednesday. Then Skynet. Why the difference? I have theories. None of them flattering to anyone, including myself. Maybe it’s funding. “Gradual improvement” doesn’t loosen purse strings. “Existential risk requiring immediate billions” does. Maybe it’s ego. Discovering fire incrementally is boring. Discovering fire that might end civilization? Now you’re interesting at dinner parties. Eric Weinstein disagreed. Vigorously. I must disagree. The Double Helix. Coasian rights. BPST instanton as Hopf fibration. Structure of Benzene. Etc. Yet, we keep saying this. The story of science just ISN’T all about group work, incremental improvement & slow accumulation of evidence. So why don’t we admit it? Now, Eric’s response would didn’t exactly disprove my assertion — several people over several years is hardly a lone genius Eureka moment cum Archimedes!. But he had the benefit of talking to Watson directly. So two lightning bolts. Same blue sky. Maybe the truth splits the difference. Most science is slow accumulation. But punctuated—rarely, unpredictably—by moments that can’t be reduced to process. The question for AI: which kind of breakthrough are we waiting for? I genuinely don’t know. Which, given my job, is somewhat embarrassing. Until next time, have a M.A.G.I.C. Week, Brian Appearance https://www.youtube.com/watch?v=Ep0xXSHQUQo I did an interview with the This Is World show and talked about a provocative idea: science doesn’t “prove” ultimate truths — it eliminates what’s false. And if that’s how physics works, then the search for God (or the disproof of God) has to play by the same rules: hypotheses, predictions, and falsifiability. ​Watch it here! Genius Did you know that researchers put a dead salmon in an MRI, showed it pictures, and found “brain activity”? The salmon was dead. The “signal” was statistical noise. This paper argues AI interpretability has the same problem. Feature attribution, probing, sparse autoencoders—they all produce plausible-looking explanations for randomly initialized networks. Networks that haven’t learned anything. Computational dead salmon. The fix: treat interpretability like actual science. Test findings against meaningful alternatives. Quantify uncertainty. Otherwise we’re just finding patterns in noise and calling it insight. Image Webb recently clarified the Crab Nebula’s origin story—or complicated it. The 1054 CE supernova was long thought to be an electron-capture explosion from a star with an oxygen-neon-magnesium core, which explained its weird composition and low energy. New infrared data suggests a weak iron core-collapse supernova works too. “The composition of the gas no longer requires an electron-capture explosion,” says lead author Tea Temim. So we’ve ruled something in, not out. Progress. 📸 Credit: Me! Courtesy of the new Celestron Origin telescope I got! Not as pretty as JWST’s images, but at a fraction of the cost 😀. Conversation https://www.youtube.com/watch?v=QzajLAbN4eo I was joined by complexity scientist and author, Sam Arbesman, to explore the captivating world of code and its profound impact on modern civilization. Together, we discuss themes from Sam Arbesman’s new book, “The Magic of Code,” and unpack why writing code can feel like casting spells—a process filled with both wonder and risk. ​Click here to watch! Get the AI Interactive Content Here! Subscribe to my podcast! More than 2M downloads! Advertisement Join me for a free course on black holes in two weeks in Phoenix, AZ! Submit your application link: https://keating.paperform.co/​ You’ll be part of a live audience for your course recording. You’ll also receive 1-year access to the Peterson Academy platform. If you can’t make it, you can take both of my PA courses — Intro to Astronomy and Intro to Cosmology, which are now available! Join me on a 9-hour captivating journey through the cosmos, exploring its vastness, the tools used to unravel its mysteries and the groundbreaking discoveries that have shaped our understanding of the universe. We examine the evidence for an expanding universe, the forces driving its evolution, and the cosmic fossils that shed light on its distant past and future. The course also delves into the enigmatic concepts of dark matter and dark energy, their roles in the universe’s structure and fate, and the ongoing efforts to unravel these cosmic mysteries. Enroll now for immediate access at https://petersonacademy.com/?utm_source=Keating Advertisement By popular demand, and for my mental health 😳, I am starting a paid “Office Hours” where you all can connect with me for the low price of $19.99 per hour. I get a lot of requests for coffee, to meet with folks one on one, to read people’s Theories of Everything etc. Due to extreme work overload, I’m only able to engage directly with supporters who show an ongoing commitment to dialogue—which is why I host a monthly Zoom session exclusively for patrons in the $19.99/month tier. It’s also available for paid Members of my Youtube channel at the Cosmic Office Hours level (also $19.99/month). Join here and see you in my office hours! Upcoming Episode Caleb Scharf will be on The INTO THE IMPOSSIBLE Podcast soon. He’s NASA’s Senior Scientist for Astrobiology, a 2022 Carl Sagan Medal winner, and the author of provocative books like The Ascent of Information and his newest, The Giant Leap—which reframes space exploration not as a technological achievement but as biology in motion, life extending its reach beyond Earth for the first time in 4.5 billion years. His argument that “alien life could be so advanced it becomes indistinguishable from physics” alone should keep us busy for hours. What questions do you have for him? Submit your questions here!

The Gossip Paradox: Your “Drama” Is Actually Intelligence Work

The Gossip Paradox: Your “Drama” Is Actually Intelligence Work Dear Magicians, I was at a lunch in Austin when someone asked: “What’s your score on Rate My Professors?” Everyone pulled out their phones. Within seconds, my entire teaching reputation was being evaluated by strangers. Nobody felt guilty about this. But ask those same students if they gossip about which research labs are good? Suddenly it’s drama. Suddenly it’s beneath serious students. This is backwards. Eshin Jolly, a neuroscientist here at UC San Diego who studies gossip, found something crucial: gossip is often spontaneous, and takes place when people suffer some sort of anxiety resulting from ambiguities in their social circles. Translation: Gossip is what you do when you need information and don’t have it. Which describes your entire undergraduate experience. Rate My Professors is institutionalized gossip. Crowdsourced reputation management. Everyone uses it without guilt. But gossip about research labs, toxic advisors, or exploitative programs? That’s taboo. Why? Because Rate My Professors is contained. It can’t expose which famous professors are terrible mentors or which prestigious programs have toxic cultures. Here’s the puzzle: How does a system maintain an illusion of meritocracy while being fundamentally arbitrary? Answer: Convince everyone that sharing information about the arbitrariness is morally wrong. Call it gossip. Make it taboo. Watch students stumble through the same mistakes individually instead of learning collectively. I spent grad school avoiding “drama” and sometimes feeling superior. Then after grad school I joined a lab everyone told me was toxic. Because they gossiped. And I didn’t. It led to me being fired. But that turned out to be the greatest gift I ever got.​ So talk to other students. Learn from their mistakes. You don’t have time to make them all yourself. You already do this for professors. Do it for everything else. It’s the same behavior. We just pretend one is data and the other is drama. Until next time, have a M.A.G.I.C. Week, Brian Appearance https://www.youtube.com/watch?v=6O4XS4L_75Y ​I joined Richard Dawkins for an insightful conversation touching on evolution, genetics, science, and culture. We explore Richard’s latest book, The Genetic Book of the Dead along with a wide array of topics including AI, the extended phenotype, evolution, the possibility of alien life followed by audience questions. Genius An NYU professor replaced take‑home exams with AI‑run oral exams to counter LLM‑enabled cheating and better test real understanding. Using ElevenLabs plus LLM workflows, a voice agent authenticates students, probes their projects, and runs case questions personalized with injected context. A three‑model grading “council” (Claude, Gemini, ChatGPT) independently scores transcripts, then revises after seeing others’ reasoning, yielding stricter, more consistent grades and highly specific feedback. The system examined 36 students for about 15 USD total, versus ~30 hours of human time. Students found it more stressful but largely agreed it measured understanding better and exposed teaching gaps, especially in experimentation. What do you think of this method of “fighting fire with fire”? Image 📸 The Milky Way galaxy courtesy of Astro-landscape_gal Conversation https://www.youtube.com/watch?list=PLJGKdZD30K__XkD2TDGq1pQo8Q__6vBhB&v=hytJAFMCozw In this episode, I am joined by acclaimed science writer, Anil Ananthaswamy, for a deep dive into the mathematical mysteries behind machine learning. We ask n the big, foundational questions: Why does the math behind machine learning work at all? What’s really happening inside these neural networks, from the simple perceptron to today’s massive deep learning systems? Are large language models revealing hidden truths, or just offering compelling illusions? ​Click here to watch! Get the AI Interactive Content Here! Subscribe to my podcast! More than 2M downloads! Advertisement Survive and thrive in academia (with or without gossiping). ​Get my latest book Focus Like a Nobel Prize Winner! Advertisement By popular demand, and for my mental health 😳, I am starting a paid “Office Hours” where you all can connect with me for the low price of $19.99 per hour. I get a lot of requests for coffee, to meet with folks one on one, to read people’s Theories of Everything etc. Due to extreme work overload, I’m only able to engage directly with supporters who show an ongoing commitment to dialogue—which is why I host a monthly Zoom session exclusively for patrons in the $19.99/month tier. It’s also available for paid Members of my Youtube channel at the Cosmic Office Hours level (also $19.99/month). Join here and see you in my office hours! Upcoming Episode Nick Lane will be on The INTO THE IMPOSSIBLE Podcast soon. He’s one of the most original thinkers on the origins of life, arguing that energy flow, metabolism, and deep chemistry—not genes alone—set the trajectory for complexity, consciousness, and even death. What would you ask him about life, energy, evolution, or where biology breaks from physics. What would you ask him? Submit your questions here!

The Thermodynamics of the Vanities (Or what 1347 crackpots taught me about attention)

The Thermodynamics of the Vanities (Or what 1347 crackpots taught me about attention) Dear Magicians, This week I learned something I already knew. Again. Attention is never neutral. It arrives with baggage: incentives, asymmetries, and people who want something from you. Your credibility. Your legitimacy. Or just a convenient villain for their own story. In 1497, Florence held the Bonfire of the Vanities. People piled their mirrors, books, cosmetics, instruments—anything associated with status or self-regard—and set them on fire. It wasn’t about purification. It was about spectacle. Heat mistaken for morality. I recently went on a podcast expecting disagreement, but instead got ambushed by my hosts’ supplication to their audience. Debate is good; science needs friction. What I didn’t expect was the flavor of the response. “They” say never read the comments. Usually, I listen to that advice. This time, I didn’t. I wanted to see what this particular audience—of a once-great podcast—was actually doing with the conversation. Most of the outrage wasn’t about ideas. It was about identities. For the high crime of defending the Big Bang, evolution, and even the scientific method + peer review, I was roasted in the comments — declared a “priest,” a “gatekeeper,” a “douchebag,” a “hubristic asshole,” and my favorite: “Buzz Lightyear wannabe” (guilty as charged). I won’t pretend some of that didn’t sting. But a pattern emerged quickly. When your entire worldview rests on “experts are ignorant,” and “consensus is groupthink”, well then disagreement isn’t a productive conversation. It’s an attack. And attacks shouldn’t be answered. They must be repelled. Every creator begs the audience to like, comment, and subscribe. I do too. Here’s something most creators won’t say out loud: comments are not feedback. They’re algorithmic kindling. Comments are thermodynamics. Bonfires of vanity. They measure heat, not work. Views measure something quieter—attention that doesn’t need a keyboard and a grievance to announce itself. The data were clarifying. The episode soared. Serious people watched. Some disagreed thoughtfully. Many simply moved on with their lives. And here’s the part that surprised me— Nothing came of it. No emails from colleagues. No institutional fallout. No crater where my reputation used to be. Just noise. Dissipating. Exactly like noise always does. Because the audience of this particular podcast no longer matters in the way it once did. At some point, it got captured by the gravity of ten thousand cranks. I should have known this. I did know this. Apparently, I needed to feel it at scale. Consider that lesson received. In 2026, I finally realized I don’t need to engage grifters to “expand reach.” There’s a reason Terence Tao doesn’t debate Terrence Howard about why 1×1 ≠ 2. It’s not fear. It’s triage. Some conversations aren’t brave. They’re just expensive. Attention is finite. Credibility compounds slowly. Credulity and outrage go hyperbolic. This episode closed a door—one I’d been holding open out of some misplaced sense of obligation. Or maybe vanity. Probably vanity. That door is closed now. Not loss. Clarity. ​Focus. And clarity—unlike outrage—burns brighter than any sanctimonious bonfire ever ignited. Until next time, have a M.A.G.I.C. Week, Brian Appearance I was featured in Close To Truth’s show and talked about what multiverses would mean in the biggest sense of the term and in different categories. ​Watch it here! Genius This paper, The Eighty Five Percent Rule for optimal learning, conducted in part by some UCSD colleagues, shows that learning is fastest when performance hovers around 85% correct, meaning about 1 in 6 attempts results in an error. It studies binary decision tasks where learners update via gradient-descent-like rules, and defines difficulty by the current error rate. The authors mathematically derive that an error rate of about 15.87% maximizes learning speed, and confirm this in artificial and biologically inspired neural networks. Too few errors give too little feedback; too many create noisy, unhelpful feedback. A practical takeaway is to aim for conditions where learners succeed roughly four out of five times. Image M45 by Galileo Galilei vs Hubble Space Telescope NGC 891, a nearly perfectly edge-on spiral galaxy 32 million lightyears away in Andromeda. Razor-thin dust lane, classic disk+halo structure, & rotation curves just like our galaxy. We study it because we can’t see the Milky Way from the outside. Taken in San Diego with my Seestar. Conversation https://www.youtube.com/watch?list=PLJGKdZD30K__XkD2TDGq1pQo8Q__6vBhB&v=rk2cxDC7C0c In this episode, I am joined by cosmologist Andrew Jaffe, author of “The Random Universe,” for a fascinating exploration into the very nature of reality. Together, we tackle some of the biggest and most intriguing questions in science: Is the universe truly random, deterministic, or perhaps something in between? ​Click here to watch! Get the AI Interactive Content Here! Subscribe to my podcast! More than 2M downloads! Advertisement WANTED: Developers and STEM experts! Get paid up to $150/hour to create benchmarks and improve AI models. Sign up for Alignerr using my link. Advertisement By popular demand, and for my mental health 😳, I am starting a paid “Office Hours” where you all can connect with me for the low price of $19.99 per hour. I get a lot of requests for coffee, to meet with folks one on one, to read people’s Theories of Everything etc. Due to extreme work overload, I’m only able to engage directly with supporters who show an ongoing commitment to dialogue—which is why I host a monthly Zoom session exclusively for patrons in the $19.99/month tier. It’s also available for paid Members of my Youtube channel at the Cosmic Office Hours level (also $19.99/month). Join here and see you in my office hours! Upcoming Episode Nick Lane will be on The INTO THE IMPOSSIBLE Podcast soon. He’s one of the most original thinkers on the origins of life, arguing that energy flow, metabolism, and deep chemistry—not genes alone—set the trajectory for complexity, consciousness, and even death. What would you ask him about life, energy, evolution, or where biology breaks from physics. What would you ask him? Submit your questions here!

The real reason your goals fail (it’s not willpower)

The real reason your goals fail (it’s not willpower) https://youtu.be/xvIBXZ74u0U?sub_confirmation=1 Dear Magicians, January takes its name from Janus, the Roman god of gates and portals—a deity who could simultaneously gaze backward and forward. Appropriate, then, that we use this month to reflect on where we’ve been and where we’re headed. Since we’re already one-twelfth through the year, it seems like a good moment to revisit those goals you set. Or didn’t set. Or set and immediately forgot while binge-watching something on Netflix. No judgment. I’ve found some additional research that might help. And I’ve included a short private video message just to you, my Magicians! (Please don’t share). Write It Down A study in the Journal of Applied Psychology found that people who commit to action and build in accountability see roughly a 25% boost in goal achievement compared to those who merely scribble intentions on paper (like writing “exercise more” and then using the notebook as a coaster). Dr. Gail Matthews’ research showed an even larger effect when people shared their goals and submitted progress reports. So if you want results, tell someone what you’re working toward. Better yet, tell someone you admire. There’s something clarifying about announcing your intentions to a person whose opinion you actually care about. Suddenly “I’ll get around to it” becomes significantly less convincing—even to yourself. Get Accountability The real engine of goal achievement isn’t willpower. It’s witnesses. If someone is going to ask you about your progress regularly, it becomes much harder to rationalize procrastination. Researchers at the University of Chicago found that sharing a goal with just one other person increases your likelihood of success by 25%. Why? Because public commitment creates social pressure. You can’t make excuses when you’ve told a friend you’re training for a marathon and posted your schedule online. The cost of failure suddenly includes embarrassment—a currency we all prefer not to spend. This is one reason I wrote an entire book on accountability systems and focus techniques: Focus Like a Nobel Prize Winner. I figured if I put my methods in print, I’d have to actually use them. Mostly this has worked. Mostly. Share Your Goals With Me I can’t respond to every email—I’m one person with finite hours and an infinite capacity for distraction—but I do read every reply. 😀 One goal I had this year was to publish my fourth book — Focus Like a Nobel Prize Winner… and publish it I did! Done. Now on to losing that stubborn last five pounds. Right after I lose the first twenty. 😂 You’ll also notice improvements coming from my upgraded home studio soon—thank you to my Patreon patrons and YouTube Members for making that possible. And as I mentioned in my recent video, look for the first-ever audiobook by Galileo, produced by yours truly, featuring my voice alongside Carlo Rovelli, Lucio Piccirillo, Frank Wilczek, Fabiola Gianotti, and Jim Gates. More on Galileo Galilei below… Good company for an amateur like me. Until next time, have a M.A.G.I.C. Week, Brian Appearance The Simons Observatory appeared in BBC sky at night. Genius CLICK HERE TO ANSER THE SURVEY! I need help with a survey from you my genius audience! Surveys help me improve the podcast and this newsletter too. I treasure each reply and you can win a $100 Amazon gift card. Unlike meteorites, I can send these cards anywhere on earth! Image M45 by Galileo Galilei vs Hubble Space Telescope In January 1610, Galileo Galilei observed the Pleiades, aka the Seven Sisters, aka Messier 45. Galileo’s Pleiades moment is the first time the sky betrayed the naked eye. In Sidereus Nuncius (1610) he aimed a crude telescope at M45 and the “six or seven” stars exploded into dozens—about 36 he bothered to plot. As an experimentalist, that’s the whole story: increase resolution, watch certainty dissolve, and force theory to catch up. It’s also a warning I carry into CMB work: nature hides structure until your detector earns it. Compare his Galileo sketch with a modern Hubble image and you can see, literally, what “more sensitivity” does to reality. That’s why I trust instruments over instinct. Learn more about Galileo Galilei — flaws and all — in my first book, Losing the Nobel Prize. Conversation https://www.youtube.com/watch?v=ukpCHo5v-Gc In this episode with Terry Tao, we discuss the mysterious world of mathematics, artificial intelligence, and the very fabric of reality with one of the greatest minds of our time. With Fields Medalist Terence Tao, often called “the Mozart of Math”, we unravel why large language models (LLMs) might be simpler than we imagine, and why the true enigma lies in understanding why they actually work. ​Click here to watch!​ Get AI Interactive Content Here! Subscribe to my podcast! More than 2M downloads! Advertisement My Intro to Cosmology course is now appearing exclusively at Peterson Academy. Join me on the 9-hour captivating journey through the cosmos, exploring its vastness, the tools used to unravel its mysteries, and the groundbreaking discoveries that have shaped our understanding of the universe. We examine the evidence for an expanding universe, the forces driving its evolution, and the cosmic fossils that shed light on its distant past and future. The course also delves into the enigmatic concepts of dark matter and energy, their roles in the universe’s structure and fate, and their ongoing efforts to unravel these cosmic mysteries. ​Enroll now for immediate access!​​ I’ve gotten great feedback from dozens of my PA students. Join us on a cosmic adventure! Advertisement By popular demand, and for my mental health 😳, I am starting a paid “Office Hours” where you all can connect with me for the low price of $19.99 per hour. I get a lot of requests for coffee, to meet with folks one on one, to read people’s Theories of Everything etc. Due to extreme work overload, I’m only able to engage directly with supporters who show an ongoing commitment to dialogue—which is why I host a monthly Zoom session exclusively for patrons in the $19.99/month

The number I didn’t expect to care about

The number I didn’t expect to care about Dear Magicians, Welcome to the last Monday M.A.G.I.C. Message of 2025. I can’t believe it! 2026 is going to be even better, but before we get to the regularly scheduled Musing, I can’t resist sharing one piece of ‘inside influencing’ that really hit me hard this year. Pre-script 1 — One million people said yes! YouTube is quite good at creating the hedonic treadmill — the insatiable desire to use their precision metrics to gauge your success as a creator. A tiny sampling of what YouTube studio tracks: Impressions (how often thumbnails are shown), Impressions click-through rate (CTR), Views, Traffic sources and hundreds more data points. For certain milestones, you get physical artifacts: plaques for reaching milestones like 100,000, 10M, 100M subscribers. While I’ll never hit the 1M mark, the most important datum to me was to recognize how many people share an interest in my videos, meaning how many people commented on them, shared them with others, and, most importantly, liked the videos. The milestone I’m proudest of this year isn’t 58+ million lifetime views. It’s not hitting a third of a million subscribers, though I appreciate each and every one of them. It’s the one million likes my videos have accumulated. That number hit me different. For unknown reasons, YouTube doesn’t. register a ‘like’ unless a viewer has watched at least 31 seconds of a video. So, of course, I did the math. That’s over 31 million seconds. Over a year of continuous human attention—one full orbit around the sun—spent just deciding to click a small button that says “this mattered to me.” Views can be accidents. Subscribers can be aspirational. But a like? A like is a tiny act of generosity from someone who already gave you half a minute of their life and decided to give you one second more. I don’t take that lightly. Now, before I get too cocky, I have to point out that 1 million divided by 58 million is <2% So clearly I need to improve to reach the remaining 98% of you. But it feels good nonetheless, and gives me a new North Star to chase, and possibly, if I’m honest with myself, obsess over as I ride the hedonic treadmill into 2026. Pre-script 2 — Two asks: 1️⃣ If my work you focus or think more clearly, I’d be grateful for an honest review: https://a.co/d/hi50U9U​ 2️⃣ Please fill out this quick survey. As a thank you, I’m giving away five $100 Amazon gift cards to five lucky winners drawn at random in early January. Your responses help me improve my content and reach a broader audience than I could on my own. Thank you, sincerely 🙏 Musing This week finishes the short experiment in something I argued for explicitly in Focus Like a Nobel Prize Winner: spaced repetition. Not flashcards—signal reinforcement. Insight doesn’t stick because it’s clever; it sticks because it’s revisited under slightly different lighting. Over the last two Mondays of the year, I’m revisiting ideas from The INTO THE IMPOSSIBLE Podcast—not to summarize them, but to compress what survived reflection. If you missed last week’s episode, it’s here. The TL;DR from last week — repletion is intellectual weight training: fewer reps, heavier meaning, better form. If you’ve ever finished a great conversation only to realize a week later that none of it changed how you think, this is my attempt to fix that—starting with myself. Most intellectual failures aren’t caused by stupidity. They’re caused by smart people arguing inside bad frames and congratulating themselves for being rigorous. I include myself here. I’ve won arguments that turned out to be about the wrong thing. Celebrated, even. Then realized I’d been playing chess on a checkerboard. Embarrassing. ​Eric Weinstein made something uncomfortably clear: many debates are settled before evidence enters the room. Whoever defines the frame defines the menu of acceptable thoughts. This explains why some conversations feel energetic but go nowhere. You’re not disagreeing about facts. You’re disagreeing about which facts are allowed to matter. Think about it. In physics, for example, gauge invariance sounds like a technical detail. But it actually determines what questions you’re permitted to ask. Lesson: Before you argue, check who built the room. ​David Deutsch pushed an idea that initially bruised my physicist ego: prediction is overrated. Ouch. You can predict things for terrible reasons. A broken clock. A lucky guess. A model that works but explains nothing. Explanation is the real currency. Good explanations make themselves vulnerable to being wrong. Bad explanations survive by being immune to contact with reality—like horoscopes, or certain tenure cases. Consider Ptolemaic astronomy. It predicted planetary positions beautifully. For centuries. But it explained nothing about why planets moved. Or take Freudian psychoanalysis—it could “explain” any behavior after the fact, which meant it explained none of them. Popper groked this a century ago. Lesson: If your theory can’t be killed, it was never really alive. ​Nathalie Cabrol — Life Detection Is a Discipline of Restraint Mars teaches humility before excitement. I wanted biosignatures. Mars wanted me to calm down. False positives are easier than discoveries. Context matters more than chemistry alone. You don’t just need the right molecule. You need the right molecule in the right place doing the right thing for reasons that aren’t boring. Like the Viking landers—they detected reactive chemistry and everyone got excited. Turns out it was probably just weird soil oxidation. Or take Allan Hills 84001, that famous Martian meteorite. Structures that looked like fossilized bacteria. Beautiful. Also possibly just mineral artifacts. Lesson: Skepticism is a form of respect for reality. Also, a form of not embarrassing yourself at press conferences. The pattern here is one of mental hygiene. Frames can trap you. Predictions can fool you. Excitement can blind you. Experts can dazzle you. But if you notice the frame, demand real explanation, and treat your own enthusiasm as a warning sign—you might actually learn something, and avoid confirmation bias as a side

Why I’m Repeating Myself (On Purpose)

Why I’m Repeating Myself (On Purpose) Dear Magicians, This week begins a short experiment in something I argued for explicitly in Focus Like a Nobel Prize Winner: spaced repetition. Not the flashcard kind, but the kind that actually changes how you think. The goal isn’t novelty. It’s signal reinforcement. Insight doesn’t stick because it’s clever; it sticks because it’s revisited under slightly different lighting. I’ve been doing this in my personal life — a weekly review each Sunday night that you can try for free here. I take a brain dump from my notes and import it into Nano Banana with a prompt to review it and make a cartoon about my week. Part of last week’s review and image is shown above 😀. So over the next three Mondays, I’m revisiting ideas from 2025’s top conversations on The INTO THE IMPOSSIBLE Podcast — not to summarize them, but to compress what survived reflection. Think of this as intellectual weight training: fewer reps, heavier meaning, better form. If you’ve ever finished a great conversation only to realize a week later that none of it changed how you think, this is my attempt to fix that — starting with myself. Two small asks, if this resonates: 1️⃣ If Focus Like a Nobel Prize Winner helped sharpen how you think or work, I’d deeply appreciate a short review: https://a.co/d/hi50U9U I have only received 34 of them for this book, vs. hundreds for my previous books… this gives me a bit of the 😥 2️⃣ Reply to this email with one idea you think is worth repeating until it actually sticks. I read every one (even if I can’t reply to them!) Repetition Is How Insight Becomes Instinct Most people confuse learning with exposure. Exposure fades. Insight only compounds when it’s revisited, reframed, and stress-tested over time. That’s why elite performers — including scientists — don’t just move forward. They loop back. This is the first pass in a short, deliberate repetition cycle. In Focus Like a Nobel Prize Winner, I argue that spaced repetition is how ideas migrate from working memory into instinct. You stop remembering them and start using them. That’s when thinking actually changes behavior. If this feels slower than the internet’s usual dopamine tempo, that’s intentional. Depth has a longer half-life than novelty. We like to imagine science advancing through brilliant ideas. Whiteboards. Coffee. A genius squinting heroically. In practice, science advances through better instruments — and the slow, slightly humiliating realization that our previous ideas were incomplete. Or wrong. Often wrong. I say this as someone very fond of theories, guesses, and ideas. Tragically so. Kyle Dawson Talking with Kyle made something uncomfortably clear: cosmology doesn’t move forward because theorists have inspired afternoons. It moves forward because the hardware gets better, the teams get bigger, the observations get crisper. Error bars do more intellectual work than epiphanies. Reality does the grading. We don’t get to appeal the score. For example, when telescopes improve sensitivity, entire models quietly die — not with drama, but with a revised plot and smaller confidence intervals. Like when a new instrument shrinks an uncertainty by half and suddenly your favorite theory needs therapy. Not glamorous. But decisive. Niayesh Afshordi Niayesh clarified something that should be obvious, yet rarely is: many Big Bang debates aren’t really about data. They’re arguments about priors. Two smart people can look at the same evidence and disagree forever because they never agreed on the rules of the game. For example, one person assumes inflation is the default, such as a starting axiom, while another treats it like a suspect that needs an alibi. Same data. Different conclusions. It’s like two detectives examining the same crime scene. One assumes it was an accident unless proven otherwise. The other assumes foul play unless ruled out. They catalog the same fingerprints, timelines, and motives — and reach opposite conclusions, not because the evidence differs, but because their starting assumptions do. No amount of new data fixes a disagreement about assumptions you refuse to name. Awkward. Fred Adams Fred reframed fine-tuning in a way I found bracing — and mildly insulting to human ego. The universe doesn’t roll out the red carpet for complexity. It tolerates it. Briefly. Under strict conditions. With no customer support. Life exists in narrow windows carved out by constraint, not cosmic generosity. For example, change the strength of gravity slightly and galaxies never form. Or tweak nuclear physics just a bit, like adjusting a recipe, and you get a universe full of disappointment instead of stars. This is not a universe trying to help us. It’s a universe barely putting up with us. Which, honestly, feels about right. Until next time, have a M.A.G.I.C. Week, Brian Appearance https://youtube.com/watch?v=HQev_EGSb8g “I was on Matt’s podcast last week. Please do NOT listen to the embarrassing intro which Matt wrote in consultation with my Mom.” Genius CLICK HERE TO ANSER THE SURVEY! I need help with a survey from you my genius audience! Surveys help me improve the podcast and this newsletter too. I treasure each reply and you can win a $100 Amazon gift card. Unlike meteorites, I can send these cards anywhere on earth! Image The Small Magellanic Cloud, NGC 292 Check out this incredible gallery of cosmic wonders, catapulting us out to see the cosmic perspective 📸 Marcella Giulia Pace/APOTY25 Conversation https://youtube.com/live/OJxX36IEVFA Join me and Harvard astrophysicist Avi Loeb for the final verdict on the interstellar object 3I/ATLAS, the third confirmed visitor from beyond our solar system. We examine the newly proposed 14th anomaly: the remarkably rare alignment of 3I/ATLAS’s rotation axis within about 8 degrees of the sunward direction at distances greater than 5 AU, a configuration with a probability of less than about 0.5 percent if random. We lay out the data, compare competing interpretations, and ask the central question: is 3I/ATLAS simply an unusual comet, or something fundamentally different? ​Click here to watch! Interactive AI Content Subscribe to my podcast! More than 2M downloads! SPONSORED BY ALIGNERR ​Snipd is my favorite podcast player, but it’s so much more. Snipd’s AI-powered app helps users save and