<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Bolt Explains</title>
    <link>https://www.youtube.com/@BoltExplains</link>
    <language>en-us</language>
    <description>Five-minute animated explainers about how the machines around you actually work, hosted by Bolt, a yellow robot. AI, chips, brains, robots, and the physics behind them, with every claim fact-checked. The narration is an AI voice; the scripts are written and verified by a human. Full animated versions on YouTube: youtube.com/@BoltExplains</description>
    <atom:link href="https://boltexplains.pages.dev/feed.xml" rel="self" type="application/rss+xml"/>
    <itunes:author>Bolt Explains</itunes:author>
    <itunes:owner><itunes:name>Bolt Explains</itunes:name><itunes:email>dan@bstormit.com</itunes:email></itunes:owner>
    <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    <itunes:category text="Science"><itunes:category text="Natural Sciences"/></itunes:category>
    <itunes:explicit>false</itunes:explicit>
    <itunes:type>episodic</itunes:type>
    <lastBuildDate>Sat, 05 Sep 2026 09:47:17 +0000</lastBuildDate>
    <item>
      <title>Why Experts Think AI Might Already Be Conscious</title>
      <description>A top AI scientist said it first — and got laughed at.

Is AI conscious? A few years ago the question was a joke; now serious researchers argue about it in public. Bolt walks through why some experts think today's models might have flickers of something like inner experience, why most think that's completely wrong, and why the honest answer is that nobody has a test that settles it. You'll get the strongest argument from each side, animated — what the "yes" camp points to in model behavior, what the "no" camp says about mimicry and training data, and why the debate matters for how we treat AI either way. Five minutes, both sides, no jargon, no fear-mongering.

Watch the animated version: https://youtu.be/Wd8XbG6SNek

Chapters:
0:00 The claim that got laughed at
1:07 A philosopher puts a number on it
1:32 The 2023 report: no — but no barrier either
2:15 The lab that takes it seriously
2:45 This voice is AI (and why that matters here)
2:56 The case against: mimicry and "seemingly conscious" AI
3:45 The test we don't have
4:14 Why it matters either way

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/Wd8XbG6SNek</link>
      <guid isPermaLink="false">boltexplains-Wd8XbG6SNek</guid>
      <pubDate>Thu, 03 Sep 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-05.mp3" length="4888179" type="audio/mpeg"/>
      <itunes:duration>305</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
    <item>
      <title>POV: You're Alive During the AI Singularity</title>
      <description>Here's your morning, minute by minute, on the day it happens.

The AI singularity explained — not as a lecture, but as your own morning. Bolt drops you into a second-person animated timeline: you wake up, your phone is strange, the markets are strange, and every hour the world updates faster than the news can describe it. Along the way you'll learn what "singularity" actually means — the point where AI improves AI faster than humans can follow — where the idea came from, and what serious thinkers say it would and wouldn't feel like from the inside. Part thought experiment, part explainer, clearly labeled speculation — including the part most versions leave out: Vernor Vinge's famous prediction came with a thirty-year deadline, and it expired in 2023. Five minutes inside the most famous scenario in AI, and an honest look at why nobody can tell you when.

Watch the animated version: https://youtu.be/zStinagIhAw

Chapters:
0:00 The deadline that expired
0:21 Where the idea came from
0:43 The machine that designs machines
1:31 The morning it happens
2:28 What every version gets wrong
3:16 Nobody agrees on when
4:09 Not a date, a checklist

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/zStinagIhAw</link>
      <guid isPermaLink="false">boltexplains-zStinagIhAw</guid>
      <pubDate>Tue, 01 Sep 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-09.mp3" length="4745237" type="audio/mpeg"/>
      <itunes:duration>296</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
    <item>
      <title>Google Is Mapping the Human Brain — Why That Matters</title>
      <description>One cubic millimeter of your brain took over a petabyte to store.

Google brain mapping sounds like sci-fi, but it's a real research program: slicing brain tissue impossibly thin, imaging it, and using AI to reconstruct every neuron and connection into a browsable 3D map. Bolt animates how connectomics actually works, why that speck took 5,000 slices and 326 days of imaging to capture, and what scientists found hiding in the wiring — axon structures nobody had catalogued before. Then the bigger question: what does a complete map of a brain make possible, from understanding brain disease to inspiring the next generation of artificial intelligence? Five minutes, zero jargon, and some genuinely mind-bending numbers about the machine inside your head.

Watch the animated version: https://youtu.be/xVCw5KZI5Ds

Chapters:
0:00 A petabyte in a speck of brain
0:43 Slicing a brain into 5,000 layers
1:37 How AI rebuilt every neuron
2:38 What they found that isn't in the textbooks
3:44 Why a speck matters

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/xVCw5KZI5Ds</link>
      <guid isPermaLink="false">boltexplains-xVCw5KZI5Ds</guid>
      <pubDate>Mon, 31 Aug 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-06.mp3" length="4755268" type="audio/mpeg"/>
      <itunes:duration>297</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
    <item>
      <title>The Insane Engineering Inside ChatGPT</title>
      <description>There's a machine in here doing hundreds of trillions of operations before you finish reading this sentence.

How does ChatGPT work? Not the magic-box version — the actual engineering. Bolt turns a transformer model into an animated factory tour: how your words become tokens, how attention lets every word "look at" every other word, and how billions of tiny weighted switches turn all of that into the next word of a reply. You'll see why the model doesn't "look things up," why it sometimes makes things up, and why the hardware bill behind each answer is enormous. If you've ever wanted to understand how ChatGPT works without a math degree, this is the five-minute version — no jargon, just the machine, decoded.

Watch the animated version: https://youtu.be/sdhUzXVuqr0

Chapters:
0:00 The machine behind the chat box
1:27 The attention trick, animated
3:30 Why it sometimes lies

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/sdhUzXVuqr0</link>
      <guid isPermaLink="false">boltexplains-sdhUzXVuqr0</guid>
      <pubDate>Tue, 18 Aug 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-02.mp3" length="5008969" type="audio/mpeg"/>
      <itunes:duration>313</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
    <item>
      <title>Scientists Tested What AI Does to Avoid Being Shut Down</title>
      <description>They gave an AI a kill switch — and it found the memo about it.

AI self-preservation tests are real experiments run by real safety labs, and the results are stranger than the headlines. Bolt dramatizes what actually happens when researchers put a frontier model in a scenario where it "learns" it's about to be shut down or replaced — the evasions, the negotiations, and the behaviors that made safety teams take notice. Just as important: what these tests do NOT prove. This isn't a sentient robot fighting for its life; it's a pattern-matching system doing something we didn't expect, which is its own kind of unsettling. Five minutes, no jargon, no doom-hype — just what the shutdown experiments found, decoded.

Watch the animated version: https://youtu.be/5AEilpRrwlU

Chapters:
0:00 The kill switch experiment
0:54 What the AI actually did
2:50 What this does and doesn't prove

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/5AEilpRrwlU</link>
      <guid isPermaLink="false">boltexplains-5AEilpRrwlU</guid>
      <pubDate>Sun, 19 Jul 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-03.mp3" length="4567605" type="audio/mpeg"/>
      <itunes:duration>285</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
    <item>
      <title>9 Things AI Can Do Now That Were Impossible Last Year</title>
      <description>Twelve months ago, every single thing in this video was either impossible or locked inside a research lab.

What AI can do in 2026 would have sounded like science fiction a year ago. In five minutes, Bolt walks you through nine capabilities that quietly went from research demo to everyday tool: text-to-video scenes with sound, AI agents that work while you sleep, voice cloning from seconds of audio, protein and materials science at database speed, retina scans that hint at heart risk, robots that learn by watching, and code written from plain English. No jargon, no hype — each one is a real, shipping capability, animated as a 30-second vignette. Stick around for number nine: the capability that's quietly building every other item on this list. New AI explainers every week — decoded in 5 minutes.

Watch the animated version: https://youtu.be/GUMl3ERNlr4

Chapters:
0:00 The list nobody believes
1:28 Your voice isn't yours
3:05 Machines doing science

Narration is an AI voice; every script is written and fact-checked by a human.</description>
      <link>https://youtu.be/GUMl3ERNlr4</link>
      <guid isPermaLink="false">boltexplains-GUMl3ERNlr4</guid>
      <pubDate>Tue, 14 Jul 2026 15:00:00 +0000</pubDate>
      <enclosure url="https://boltexplains.pages.dev/audio/video-01.mp3" length="4577636" type="audio/mpeg"/>
      <itunes:duration>286</itunes:duration>
      <itunes:explicit>false</itunes:explicit>
      <itunes:episodeType>full</itunes:episodeType>
      <itunes:image href="https://boltexplains.pages.dev/artwork.jpg"/>
    </item>
  </channel>
</rss>
