1 00:00:00,000 --> 00:00:01,360 Welcome to the deep dive. 2 00:00:01,360 --> 00:00:02,320 Are you ready for this? 3 00:00:02,320 --> 00:00:07,920 Today we're diving into how AI is being used to decode ancient languages and texts. 4 00:00:07,920 --> 00:00:11,600 We've got all these amazing news articles about projects and discoveries 5 00:00:11,600 --> 00:00:14,160 that are honestly like rewriting history. 6 00:00:14,160 --> 00:00:15,920 Are you ready to have your mind blown? 7 00:00:15,920 --> 00:00:18,400 I think mind blown is an understatement. 8 00:00:18,400 --> 00:00:21,040 It's just incredible how AI isn't just translating. 9 00:00:21,040 --> 00:00:24,640 It's actually like revealing these hidden connections and patterns 10 00:00:24,640 --> 00:00:26,960 that have been buried for centuries. 11 00:00:26,960 --> 00:00:29,760 Like having a new set of eyes to look at the past. 12 00:00:29,760 --> 00:00:34,160 Yeah. Like can you imagine like being able to read a 5,000 year old shopping list? 13 00:00:34,160 --> 00:00:38,640 Or what about cracking the enigma code in 10 minutes faster than Alan Turing? 14 00:00:38,640 --> 00:00:41,680 We're talking about languages that have stumped experts for generations. 15 00:00:42,320 --> 00:00:45,120 But now thanks to AI they're starting to give up their secrets. 16 00:00:45,120 --> 00:00:49,760 It really is like changing the scale of what's possible when it comes to understanding the past. 17 00:00:49,760 --> 00:00:53,920 A great example of this is the work being done on Hidite tablets over in Turkey. 18 00:00:53,920 --> 00:00:57,200 So they've digitized 2,000 of these ancient tablets. 19 00:00:57,200 --> 00:01:00,880 And the AI they're using is already at 75% accuracy in translation. 20 00:01:00,880 --> 00:01:03,600 Wow. Okay. So 75% might not sound like a lot. 21 00:01:04,560 --> 00:01:05,760 But think about it this way. 22 00:01:05,760 --> 00:01:08,080 Deciphering these tablets used to take years. 23 00:01:08,720 --> 00:01:12,480 And we're talking about just for a single scholar to translate one tablet. 24 00:01:12,480 --> 00:01:16,000 Now we're talking about thousands of tablets being translated so fast. 25 00:01:16,000 --> 00:01:19,440 I mean, what mysteries are you most curious about solving with this technology? 26 00:01:19,440 --> 00:01:21,920 Oh, the possibilities are really endless. 27 00:01:21,920 --> 00:01:23,840 And it's not even just about the speed. 28 00:01:23,840 --> 00:01:25,680 Take the Dead Sea Scrolls, for example. 29 00:01:25,680 --> 00:01:28,800 When AI analyzed the Great Isaiah Scroll, 30 00:01:28,800 --> 00:01:31,280 it actually found out that it was written by two different scribes. 31 00:01:31,280 --> 00:01:33,120 Wait a minute. How is that even possible? 32 00:01:33,120 --> 00:01:39,280 Handwriting. AI can pick up on these tiny, almost invisible differences that humans often miss. 33 00:01:39,280 --> 00:01:41,840 It's like using a microscope on historical documents. 34 00:01:41,840 --> 00:01:45,120 Wow. So what if we could apply this to like any historical text? 35 00:01:45,120 --> 00:01:48,960 Could we end up like rewriting what we know about major historical events 36 00:01:48,960 --> 00:01:51,680 or even like challenge the authorship of famous works? 37 00:01:51,680 --> 00:01:53,920 It's like we're holding a magnifying glass up to history. 38 00:01:53,920 --> 00:01:55,120 Exactly. 39 00:01:55,120 --> 00:01:57,040 And speaking of rewriting history, 40 00:01:57,040 --> 00:02:00,080 there's this really interesting story of Linear B. 41 00:02:00,080 --> 00:02:03,600 This script baffled experts for decades. 42 00:02:03,600 --> 00:02:07,600 It was like they were trying to read hieroglyphics without the Rosetta Stone. 43 00:02:07,600 --> 00:02:10,240 Oh yeah. Linear B. That sounds familiar. 44 00:02:10,240 --> 00:02:12,240 But can you remind me what was the breakthrough there? 45 00:02:12,240 --> 00:02:14,480 Well, AI finally helped crack the code. 46 00:02:14,480 --> 00:02:17,680 Turns out Linear B is an early form of Greek. 47 00:02:17,680 --> 00:02:22,000 This discovery pushed back the timeline of written Greek by centuries, 48 00:02:22,000 --> 00:02:25,440 which completely changed what we thought we knew about ancient Greek history. 49 00:02:25,440 --> 00:02:29,280 Wow. So we basically had to rewrite textbooks because of AI. That's huge. 50 00:02:29,280 --> 00:02:30,480 Yeah, it really is. 51 00:02:30,480 --> 00:02:36,000 And now researchers are using similar AI techniques to tackle even more challenging languages 52 00:02:36,000 --> 00:02:40,960 like Etruscan and Iberian, which have no close modern relatives at all. 53 00:02:40,960 --> 00:02:44,800 It makes you wonder how many cultures and stories we thought were just lost. 54 00:02:44,800 --> 00:02:46,800 It might actually be recovered with this technology. 55 00:02:46,800 --> 00:02:48,400 It's a very real possibility. 56 00:02:48,400 --> 00:02:52,800 We might be super close to understanding civilizations that we thought were gone forever. 57 00:02:52,800 --> 00:02:53,840 It's an exciting time. 58 00:02:53,840 --> 00:02:58,320 It's like all of a sudden we have this key that unlocks doors we didn't even know existed. 59 00:02:58,320 --> 00:03:00,320 And that key is AI. 60 00:03:00,320 --> 00:03:04,800 But what's even cooler is that this isn't limited to just ancient languages. 61 00:03:04,800 --> 00:03:09,120 Researchers are using AI to decode other complex communication systems as well. 62 00:03:09,120 --> 00:03:11,040 Oh wow. Like what? Give me an example. 63 00:03:11,040 --> 00:03:12,480 Like the waggle dance of bees. 64 00:03:12,480 --> 00:03:16,800 No way. Are you telling me that AI can understand what bees are saying to each other? 65 00:03:16,800 --> 00:03:21,200 It's still early, but researchers are using AI to analyze these dances 66 00:03:21,200 --> 00:03:26,240 and try to understand how bees communicate the location of food or potential threats, 67 00:03:26,240 --> 00:03:28,720 even like the overall health of the hive. 68 00:03:28,720 --> 00:03:31,120 So like theoretically we could talk to bees. 69 00:03:31,120 --> 00:03:33,920 That's insane. What could we even learn from them? 70 00:03:33,920 --> 00:03:35,440 Imagine the possibilities. 71 00:03:35,440 --> 00:03:40,880 Understanding bee communication could revolutionize agriculture and even conservation efforts. 72 00:03:40,880 --> 00:03:45,120 We could learn how to better protect them, optimize pollination strategies, 73 00:03:45,120 --> 00:03:47,440 and even gain insights into the health of ecosystems. 74 00:03:47,440 --> 00:03:51,520 That's mind-blowing. It's like we're on the edge of this whole new era of communication. 75 00:03:51,520 --> 00:03:55,040 Not just with ancient civilizations, but with nature itself. 76 00:03:55,040 --> 00:03:57,040 It's a pretty exciting time to be alive, right? 77 00:03:57,040 --> 00:03:59,840 Absolutely. But before we get ahead of ourselves, 78 00:03:59,840 --> 00:04:04,720 let's talk about one of the biggest mysteries of all time, the Voynich manuscript. 79 00:04:04,720 --> 00:04:07,520 Ah yes. The Voynich manuscript. 80 00:04:07,520 --> 00:04:14,320 That 600-year-old book filled with weird illustrations and undecipherable scripts. 81 00:04:14,320 --> 00:04:17,680 It's been called the most mysterious manuscript in the world for a reason. 82 00:04:17,680 --> 00:04:19,920 It sounds like something out of a conspiracy theory. 83 00:04:19,920 --> 00:04:22,240 What are some of the theories about what it is? 84 00:04:22,240 --> 00:04:28,240 Oh, there are tons. From medieval medical texts to coded messages from secret societies. 85 00:04:28,240 --> 00:04:31,600 It's a puzzle that has stumped the smartest people for centuries. 86 00:04:31,600 --> 00:04:35,760 And now AI is bringing some really interesting new possibilities to the table. 87 00:04:35,760 --> 00:04:37,280 Okay, can't leave me hanging. 88 00:04:37,280 --> 00:04:39,680 What is AI telling us about this mysterious book? 89 00:04:39,680 --> 00:04:43,520 Well, researchers have used AI to analyze the manuscript structure. 90 00:04:43,520 --> 00:04:47,120 And it's suggesting a possible Hebrew code, even more exciting. 91 00:04:47,120 --> 00:04:49,680 AI has been able to offer partial translations. 92 00:04:49,680 --> 00:04:52,960 Hold on. So AI is translating the Voynich manuscript. What's it saying? 93 00:04:52,960 --> 00:04:54,640 One of the early translations was, 94 00:04:54,640 --> 00:04:58,400 she made recommendations to the priest, man of the house, and me and people. 95 00:04:58,400 --> 00:04:59,600 Oh, that's cryptic. 96 00:04:59,600 --> 00:05:04,960 But still, it's amazing to think that AI might be the key to unlocking this book. 97 00:05:04,960 --> 00:05:06,480 It's a big step forward. 98 00:05:06,480 --> 00:05:09,760 But as with anything related to the Voynich manuscript, 99 00:05:09,760 --> 00:05:14,960 there's still a ton of debate and skepticism. A lot of experts are still cautious, 100 00:05:14,960 --> 00:05:18,400 pointing out that previous attempts have suggested other languages. 101 00:05:18,400 --> 00:05:20,720 So is this a real breakthrough? 102 00:05:20,720 --> 00:05:24,880 Or just another theory in a long line of failed attempts? 103 00:05:24,880 --> 00:05:26,240 I guess only time will tell. 104 00:05:26,240 --> 00:05:29,600 I think that's the best part about mysteries like this. They keep us guessing. 105 00:05:29,600 --> 00:05:32,400 And remind us that there's still so much that we don't know. 106 00:05:32,400 --> 00:05:34,080 I don't know about you, but I am hooked. 107 00:05:34,080 --> 00:05:37,280 I can't wait to see what other secrets AI helps us unlock. 108 00:05:37,280 --> 00:05:41,760 We've barely scratched the surface. There's a whole world of knowledge waiting to be discovered. 109 00:05:41,760 --> 00:05:45,280 And we'll be right back to dive even deeper into these amazing breakthroughs. 110 00:05:47,280 --> 00:05:51,680 Okay, so we're back and ready to keep going with all the ways AI is helping us understand the past. 111 00:05:51,680 --> 00:05:56,160 Last time we were talking about AI being able to translate those ancient languages 112 00:05:56,160 --> 00:06:01,520 and even deciphering those complex communication systems like the waggle dance of bees. 113 00:06:01,520 --> 00:06:03,200 Yeah, that was pretty wild, huh? 114 00:06:03,200 --> 00:06:06,160 It would be crazy if we could actually talk to others about AI. 115 00:06:06,160 --> 00:06:08,160 If we could actually talk to other species someday. 116 00:06:08,160 --> 00:06:08,720 Imagine that. 117 00:06:08,720 --> 00:06:12,400 But okay, for now, let's stick with these breakthroughs that are happening with ancient languages. 118 00:06:12,400 --> 00:06:17,520 I was reading about this AI model that can translate dead languages in seconds. 119 00:06:17,520 --> 00:06:18,640 What can you tell me about that? 120 00:06:18,640 --> 00:06:20,160 Yeah, that's a really interesting one. 121 00:06:20,720 --> 00:06:26,880 That model is focused on Akkadian, one of the oldest and most complex languages in the world. 122 00:06:26,880 --> 00:06:29,520 It was spoken in ancient Mesopotamia. 123 00:06:29,520 --> 00:06:35,520 And they used cuneiform, which is that writing system that uses those wedge shaped symbols, 124 00:06:35,520 --> 00:06:37,120 you know, etched into clay tablets. 125 00:06:37,120 --> 00:06:38,720 Oh yeah, I've seen pictures of those tablets. 126 00:06:38,720 --> 00:06:40,080 They always look so mysterious. 127 00:06:40,880 --> 00:06:43,440 Like they're holding the secrets of some lost civilization. 128 00:06:43,440 --> 00:06:44,320 Well, they do. 129 00:06:44,320 --> 00:06:48,000 And for a long time translating Akkadian was a really time consuming process. 130 00:06:48,000 --> 00:06:50,400 First, you had to transliterate it into Latin characters, 131 00:06:50,960 --> 00:06:52,720 and then you could translate it into English. 132 00:06:52,720 --> 00:06:55,520 But this AI model can do it all in just seconds. 133 00:06:55,520 --> 00:06:57,920 Wow, so it's like a Google translate for ancient languages? 134 00:06:57,920 --> 00:06:59,040 Pretty much, yeah. 135 00:06:59,040 --> 00:07:04,160 And it's really accurate too, like 97% accuracy rate for Latin transliterations. 136 00:07:04,160 --> 00:07:05,200 That's incredible. 137 00:07:05,200 --> 00:07:09,680 It does have some trouble with like literary or more creative texts, 138 00:07:09,680 --> 00:07:10,400 which makes sense. 139 00:07:10,400 --> 00:07:11,840 Those are always harder to translate. 140 00:07:11,840 --> 00:07:13,040 Yeah, I can imagine that. 141 00:07:13,040 --> 00:07:16,800 So what does this mean for like our understanding of Akkadian texts? 142 00:07:16,800 --> 00:07:21,680 I think it has the potential to really accelerate research in this area. 143 00:07:21,680 --> 00:07:26,240 Imagine scholars being able to scan through thousands of tablets super fast, 144 00:07:26,960 --> 00:07:29,520 and pick out the ones that are most important to their work. 145 00:07:29,520 --> 00:07:32,000 It could open up whole new areas of research. 146 00:07:32,000 --> 00:07:32,560 That's so cool. 147 00:07:32,560 --> 00:07:37,280 It seems like AI is like unlocking a whole new era of discovery for ancient history. 148 00:07:38,000 --> 00:07:41,840 Speaking of unlocking secrets, what about this whole thing with plants and AI? 149 00:07:42,800 --> 00:07:44,480 That's a weird one I'd love to hear more about. 150 00:07:44,480 --> 00:07:46,240 Oh yeah, the language of plants. 151 00:07:46,800 --> 00:07:48,480 It sounds like science fiction, I know. 152 00:07:49,040 --> 00:07:54,080 But there are some companies out there developing AI models that can interpret the signals that 153 00:07:54,080 --> 00:07:56,880 plants give off in response to their environment. 154 00:07:56,880 --> 00:07:58,480 Signals, what kind of signals are we talking about here? 155 00:07:58,480 --> 00:08:02,000 Well, it can be anything from electrical signals in their routes 156 00:08:02,000 --> 00:08:04,480 to ultrasonic sounds they make when they're stressed. 157 00:08:04,480 --> 00:08:07,280 It's like this whole world of communication we didn't even know about. 158 00:08:07,280 --> 00:08:12,720 So plants are talking to each other, and AI is listening in on their conversations. 159 00:08:12,720 --> 00:08:13,920 That is wild. 160 00:08:13,920 --> 00:08:15,840 What could we even learn from that? 161 00:08:15,840 --> 00:08:17,040 Well, think about it. 162 00:08:17,040 --> 00:08:20,480 What if farmers could use AI to understand what their crops need? 163 00:08:21,120 --> 00:08:24,720 Then they could optimize their water usage and fertilizer, 164 00:08:25,440 --> 00:08:29,760 leading to better farming practices, more efficient, more sustainable. 165 00:08:29,760 --> 00:08:34,880 Some early studies have even shown a 30% reduction in energy use in greenhouses, 166 00:08:34,880 --> 00:08:36,800 while still growing the same amount of food. 167 00:08:36,800 --> 00:08:38,320 Wow, that's really impressive. 168 00:08:38,320 --> 00:08:40,960 So AI isn't just helping us understand the past, 169 00:08:40,960 --> 00:08:43,600 it's also helping us create a more sustainable future. 170 00:08:43,600 --> 00:08:48,400 Exactly, it's a perfect example of how we can use AI to tackle some of the biggest problems. 171 00:08:48,400 --> 00:08:48,960 Definitely. 172 00:08:48,960 --> 00:08:53,760 Before we get too far ahead though, let's go back to one more cool AI discovery from the past. 173 00:08:53,760 --> 00:08:56,320 Have you ever heard of the Archimedes polympsest? 174 00:08:56,320 --> 00:08:57,760 Oh yeah, for sure. 175 00:08:57,760 --> 00:08:59,440 I feel like I've heard that name before. 176 00:08:59,440 --> 00:09:00,640 But I can't quite remember what it is. 177 00:09:00,640 --> 00:09:05,520 So it's this ancient manuscript that has some previously unknown texts by Archimedes, 178 00:09:05,520 --> 00:09:07,520 that famous Greek mathematician. 179 00:09:07,520 --> 00:09:09,760 But here's the thing, it's a polympsest, 180 00:09:09,760 --> 00:09:12,000 which means the original writing was scraped off, 181 00:09:12,640 --> 00:09:14,960 and they reused the page for something new. 182 00:09:14,960 --> 00:09:17,040 Oh wow, so it's like an ancient version of recycling. 183 00:09:17,040 --> 00:09:19,600 But how can you even read the original if it's been erased? 184 00:09:19,600 --> 00:09:20,960 Well, that's where AI comes in. 185 00:09:20,960 --> 00:09:23,680 They use this thing called multispectral imaging, 186 00:09:23,680 --> 00:09:27,760 and it can basically like peel back the layers of text to see what's underneath. 187 00:09:27,760 --> 00:09:29,360 It's like digital archaeology. 188 00:09:29,360 --> 00:09:30,400 That's amazing. 189 00:09:30,400 --> 00:09:33,840 So what kind of stuff did they find on the Archimedes polympsest? 190 00:09:33,840 --> 00:09:38,880 They found some of his lost works on topics like infinity, geometry, 191 00:09:39,440 --> 00:09:43,520 even a method that's super similar to what we call calculus today. 192 00:09:43,520 --> 00:09:47,840 That's mind-blowing that we're still learning new things from Archimedes. 193 00:09:47,840 --> 00:09:48,960 After all these centuries. 194 00:09:48,960 --> 00:09:53,280 It really shows how smart he was, and how powerful technology can be. 195 00:09:53,280 --> 00:09:55,600 To rediscover things we thought were gone forever. 196 00:09:55,600 --> 00:09:58,080 Totally. Speaking of rediscovering lost knowledge, 197 00:09:58,080 --> 00:10:01,840 there was this story I read about an AI model called fragmentary. 198 00:10:01,840 --> 00:10:05,120 It was used to analyze all these clay tablets from ancient Babylon. 199 00:10:05,120 --> 00:10:06,240 Oh yeah, fragmentary. 200 00:10:06,240 --> 00:10:08,960 That's a huge one for understanding Babylonian literature. 201 00:10:08,960 --> 00:10:12,160 Those tablets were so hard for humans to piece together, 202 00:10:12,160 --> 00:10:14,720 because a lot of them were broken or had missing pieces. 203 00:10:14,720 --> 00:10:18,000 So it's like a jigsaw puzzle, but with ancient writing instead of pictures. 204 00:10:18,000 --> 00:10:18,960 Exactly. 205 00:10:18,960 --> 00:10:24,080 And fragmentary was able to reconstruct inscriptions that had stumped experts for years. 206 00:10:24,080 --> 00:10:25,600 Wow, so what did they find? 207 00:10:25,600 --> 00:10:28,720 One of the coolest discoveries was this beautiful hymn. 208 00:10:28,720 --> 00:10:31,920 It describes the arrival of spring in Babylon. 209 00:10:31,920 --> 00:10:33,600 Really vivid imagery. 210 00:10:33,600 --> 00:10:38,000 Like rivers nourishing the plains, and grains growing in the fields. 211 00:10:38,000 --> 00:10:41,040 It gives us a glimpse into how they experience the world. 212 00:10:41,040 --> 00:10:41,600 That's amazing. 213 00:10:41,600 --> 00:10:44,480 It's like looking through a window into their culture and beliefs. 214 00:10:44,480 --> 00:10:44,960 I know. 215 00:10:44,960 --> 00:10:51,760 And what makes it even more interesting is that it's a whole new genre of hymns in Babylonian literature. 216 00:10:51,760 --> 00:10:53,520 Hymns dedicated to cities. 217 00:10:53,520 --> 00:10:57,200 This was a foreign that had been lost for thousands of years. 218 00:10:57,200 --> 00:11:00,240 And now, thanks to AI, we're rediscovering it. 219 00:11:00,240 --> 00:11:00,880 That's so cool. 220 00:11:00,880 --> 00:11:03,680 It's incredible that we're still learning so much about these ancient cultures. 221 00:11:03,680 --> 00:11:04,080 I know. 222 00:11:04,080 --> 00:11:04,960 It's awesome. 223 00:11:04,960 --> 00:11:08,080 And you know, AI isn't just helping us understand ancient languages. 224 00:11:08,080 --> 00:11:12,560 It's also being used to document and revitalize languages that are still spoken today, 225 00:11:13,200 --> 00:11:14,960 but are in danger of disappearing. 226 00:11:14,960 --> 00:11:16,000 Oh wow, yeah. 227 00:11:16,000 --> 00:11:17,120 That's so important. 228 00:11:17,120 --> 00:11:19,680 Language is such a huge part of any culture. 229 00:11:19,680 --> 00:11:23,200 Losing a language is like losing a whole way of seeing the world. 230 00:11:23,200 --> 00:11:23,920 Exactly. 231 00:11:23,920 --> 00:11:27,040 And AI is playing a crucial role in this effort. 232 00:11:27,040 --> 00:11:32,960 For example, there are researchers using AI to create digital archive of indigenous languages 233 00:11:32,960 --> 00:11:34,240 all over the world. 234 00:11:34,240 --> 00:11:37,280 These archives include recordings of native speakers, 235 00:11:37,280 --> 00:11:40,240 written texts, even grammatical analyses. 236 00:11:40,240 --> 00:11:43,360 It's like creating a digital time capsule for these languages. 237 00:11:43,360 --> 00:11:44,800 That's a beautiful way to put it. 238 00:11:44,800 --> 00:11:47,440 So future generations can still connect with their heritage. 239 00:11:47,440 --> 00:11:48,240 Exactly. 240 00:11:48,240 --> 00:11:53,520 It's such a powerful example of how we can use technology to preserve cultural diversity. 241 00:11:53,520 --> 00:11:54,240 I totally agree. 242 00:11:54,240 --> 00:11:57,280 It's amazing to see all the good that AI can do. 243 00:11:57,280 --> 00:11:58,400 It really is. 244 00:11:58,400 --> 00:12:00,000 But we also have to be careful. 245 00:12:00,000 --> 00:12:05,760 Like with any powerful technology, we need to make sure that AI is used ethically and responsibly, 246 00:12:05,760 --> 00:12:08,720 especially when we're dealing with something as sensitive as language, 247 00:12:09,440 --> 00:12:12,480 which is so connected to people's identities and cultures. 248 00:12:12,480 --> 00:12:13,440 That's such a good point. 249 00:12:13,440 --> 00:12:16,800 We have to make sure AI is used to empower people not to exploit them. 250 00:12:16,800 --> 00:12:17,200 Exactly. 251 00:12:17,200 --> 00:12:21,200 It's a conversation we need to keep having as AI continues to get more advanced. 252 00:12:21,200 --> 00:12:22,000 Yeah, definitely. 253 00:12:22,000 --> 00:12:27,680 But for now, let's stay positive and celebrate all the incredible ways AI is helping us connect with language. 254 00:12:27,680 --> 00:12:28,400 Couldn't agree more. 255 00:12:28,400 --> 00:12:30,160 There's so much to explore. 256 00:12:30,160 --> 00:12:32,800 Speaking of exploring, there's one more story I wanted to talk about 257 00:12:33,680 --> 00:12:38,400 about this project that's using AI to help decode the Iberian language. 258 00:12:38,400 --> 00:12:40,080 Oh, yes, Iberian. 259 00:12:40,080 --> 00:12:42,560 That one has been stumping linguists forever. 260 00:12:42,560 --> 00:12:43,120 Tell me more. 261 00:12:43,760 --> 00:12:45,040 Why is it so hard to decipher? 262 00:12:45,040 --> 00:12:49,440 Well, unlike a lot of ancient languages, Iberian doesn't have a direct descendant. 263 00:12:49,440 --> 00:12:51,920 It's what we call a linguistic isolate. 264 00:12:51,920 --> 00:12:55,040 It's not related to any other language families that we know of. 265 00:12:55,040 --> 00:12:57,440 So there's no Rosetta Stone for Iberian. 266 00:12:57,440 --> 00:12:58,480 It's really tough. 267 00:12:58,480 --> 00:13:01,280 So how are researchers even starting to figure it out? 268 00:13:01,280 --> 00:13:02,640 That's where the AI comes in. 269 00:13:02,640 --> 00:13:06,720 Researchers at MIT created this machine learning algorithm, 270 00:13:06,720 --> 00:13:10,400 specifically to crack the code of extinct languages like Iberian. 271 00:13:10,400 --> 00:13:14,400 Instead of relying on a related modern language, it looks for sound changes. 272 00:13:14,400 --> 00:13:16,960 Sound changes across all kinds of languages. 273 00:13:16,960 --> 00:13:21,200 Trying to connect Iberian words with similar sounding terms in languages we do understand. 274 00:13:21,200 --> 00:13:22,960 Wow, so it's like a linguistic detective. 275 00:13:22,960 --> 00:13:23,520 I like that. 276 00:13:23,520 --> 00:13:25,520 Looking for clues across time and space. 277 00:13:25,520 --> 00:13:26,480 Exactly. 278 00:13:26,480 --> 00:13:27,360 And it's working. 279 00:13:27,360 --> 00:13:28,800 They're getting some interesting results. 280 00:13:28,800 --> 00:13:33,120 It's even challenging some old ideas about how Iberian might be related to BASC. 281 00:13:33,120 --> 00:13:37,600 So AI is rewriting what we thought we knew about this ancient language. 282 00:13:37,600 --> 00:13:38,640 It really is. 283 00:13:38,640 --> 00:13:43,440 And it's giving us hope that someday we might be able to fully understand the people who spoke it. 284 00:13:43,440 --> 00:13:45,040 This is all so crazy. 285 00:13:45,760 --> 00:13:50,720 I can't believe how much AI is changing the way we see the world. 286 00:13:50,720 --> 00:13:51,760 It's remarkable. 287 00:13:51,760 --> 00:13:55,120 But wait, before we move on, there's one more AI project I wanted to mention. 288 00:13:55,120 --> 00:13:56,800 It's about the Vindalanda tablets. 289 00:13:57,360 --> 00:14:02,080 They were discovered in an ancient Roman fort, your Hadrian's Wall in England. 290 00:14:02,080 --> 00:14:02,800 Oh, right. 291 00:14:02,800 --> 00:14:03,440 I know those. 292 00:14:03,440 --> 00:14:06,320 They're full of information about daily life in Roman Britain. 293 00:14:06,320 --> 00:14:06,880 Yeah, they are. 294 00:14:07,440 --> 00:14:10,880 But like a lot of these old texts, they've been damaged over time. 295 00:14:10,880 --> 00:14:12,720 Thankfully, AI is coming to the rescue. 296 00:14:12,720 --> 00:14:13,600 You got it. 297 00:14:13,600 --> 00:14:19,280 Researchers are using these machine learning algorithms to analyze pictures of the tablets 298 00:14:19,280 --> 00:14:21,440 and figure out patterns in the faded writing. 299 00:14:21,440 --> 00:14:23,360 So it's like a digital restoration project. 300 00:14:23,360 --> 00:14:24,240 Yeah, it is. 301 00:14:24,240 --> 00:14:28,480 And it's helping to make these documents much easier to read for scholars. 302 00:14:28,480 --> 00:14:29,120 That's awesome. 303 00:14:29,120 --> 00:14:32,800 It's like AI is building a bridge between the past and the present. 304 00:14:32,800 --> 00:14:33,440 I love that. 305 00:14:33,440 --> 00:14:37,120 It makes you wonder what other amazing things are out there just waiting to be found. 306 00:14:37,120 --> 00:14:37,760 I know, right? 307 00:14:38,480 --> 00:14:39,920 I guess we'll have to stay tuned. 308 00:14:39,920 --> 00:14:41,840 But okay, for now, we're going to take a quick break. 309 00:14:41,840 --> 00:14:46,560 And when we come back, we'll wrap up our deep dive into the world of AI and ancient languages. 310 00:14:49,920 --> 00:14:55,520 All right, we are back for the final part of our deep dive into how AI is changing the way we 311 00:14:55,520 --> 00:14:57,440 understand ancient languages. 312 00:14:57,440 --> 00:15:01,040 It's amazing to think about how much this technology is changing, 313 00:15:01,040 --> 00:15:04,720 how we think about what we can know, and what's lost forever. 314 00:15:04,720 --> 00:15:05,840 You're telling me. 315 00:15:05,840 --> 00:15:08,880 We've talked about AI translating all these ancient texts, 316 00:15:08,880 --> 00:15:14,640 reconstructing damaged documents, and even understanding how bees communicate. 317 00:15:14,640 --> 00:15:17,840 Yeah, and now we're going to tackle one of the biggest mysteries out there, 318 00:15:17,840 --> 00:15:19,120 the Voynich manuscript. 319 00:15:19,120 --> 00:15:20,160 Oh, yeah, that's a good one. 320 00:15:20,160 --> 00:15:24,320 The 600-year-old book with those bizarre illustrations and that crazy script that 321 00:15:24,320 --> 00:15:27,600 nobody can read, it's puzzled experts for centuries. 322 00:15:27,600 --> 00:15:29,440 It's like something right out of a movie. 323 00:15:29,440 --> 00:15:31,200 People have come up with some wild theories about it, 324 00:15:31,200 --> 00:15:34,800 everything from alien communication to medieval medical guides. 325 00:15:34,800 --> 00:15:36,480 Where do you even start with something like that? 326 00:15:36,480 --> 00:15:40,560 I know it's a puzzle that's captured people's imaginations for generations, 327 00:15:40,560 --> 00:15:44,320 but the exciting thing is that AI is starting to offer some new clues. 328 00:15:44,320 --> 00:15:45,520 Okay, I'm listening. 329 00:15:45,520 --> 00:15:47,920 What's AI telling us about the Voynich manuscript? 330 00:15:48,480 --> 00:15:52,720 Well, researchers have been using AI to analyze the language in the manuscript, 331 00:15:53,280 --> 00:15:56,400 and it's suggesting that it might be a Hebrew code. 332 00:15:56,400 --> 00:16:00,240 And get this, AI has even been able to make some partial translations. 333 00:16:00,240 --> 00:16:04,320 Wait a second, AI is actually translating the Voynich manuscript. 334 00:16:04,320 --> 00:16:05,120 What's it saying? 335 00:16:05,120 --> 00:16:07,440 One of the first translations they got was, 336 00:16:07,440 --> 00:16:12,240 she made recommendations to the priest, man of the house, and me and people. 337 00:16:12,240 --> 00:16:14,160 Huh, well that's definitely cryptic, 338 00:16:14,800 --> 00:16:18,960 but it's still incredible that AI might be able to finally unlock the secrets of this book. 339 00:16:18,960 --> 00:16:19,680 I know, right? 340 00:16:19,680 --> 00:16:21,200 It's a huge step forward. 341 00:16:21,200 --> 00:16:23,600 But like we've said with the Voynich manuscript, 342 00:16:23,600 --> 00:16:27,040 there are always a lot of different opinions and a lot of skepticism. 343 00:16:27,040 --> 00:16:29,600 A lot of experts are still being careful, 344 00:16:29,600 --> 00:16:34,000 because there have been so many attempts to decode the Voynich manuscript before, 345 00:16:34,000 --> 00:16:34,880 and they haven't worked. 346 00:16:34,880 --> 00:16:36,720 Right, so is this a real breakthrough? 347 00:16:37,280 --> 00:16:40,880 Or is this just another theory that's going to turn out to be wrong? 348 00:16:40,880 --> 00:16:42,080 I guess the only time we'll tell. 349 00:16:42,080 --> 00:16:44,560 I think that's what makes these kinds of mysteries so fascinating. 350 00:16:44,560 --> 00:16:45,760 They keep us guessing, 351 00:16:45,760 --> 00:16:48,720 and remind us that there's still so much we don't understand. 352 00:16:48,720 --> 00:16:50,880 Okay, so let's move on to another unsolved mystery. 353 00:16:51,840 --> 00:16:53,040 What about Linear A? 354 00:16:53,840 --> 00:16:56,160 That was the writing system of the Minoans, right? 355 00:16:56,160 --> 00:16:56,640 Right. 356 00:16:56,640 --> 00:16:59,840 That ancient civilization that lived on the island of Crete. 357 00:16:59,840 --> 00:17:00,400 Exactly. 358 00:17:00,400 --> 00:17:01,520 I remember learning about them. 359 00:17:01,520 --> 00:17:04,160 They had that whole palace culture with the bull leaping. 360 00:17:04,160 --> 00:17:04,800 Yeah. 361 00:17:04,800 --> 00:17:07,520 And the myths about the Minotaur and the Labyrinth. 362 00:17:07,520 --> 00:17:10,800 Oh yeah, they were a really interesting civilization, 363 00:17:10,800 --> 00:17:12,160 very advanced for their time. 364 00:17:12,720 --> 00:17:16,880 But unlike Linear B, which we now know is an early form of Greek, 365 00:17:16,880 --> 00:17:19,440 Linear A is still a complete mystery. 366 00:17:19,440 --> 00:17:21,840 So we can't read anything written in Linear A. 367 00:17:21,840 --> 00:17:22,480 Not yet. 368 00:17:22,480 --> 00:17:23,520 We know it's not Greek, 369 00:17:23,520 --> 00:17:27,680 and it's probably connected to some languages that were spoken around the agency, 370 00:17:27,680 --> 00:17:28,960 back in the Bronze Age. 371 00:17:28,960 --> 00:17:30,720 But nobody's been able to crack the code. 372 00:17:30,720 --> 00:17:32,400 But AI might be able to help, right? 373 00:17:32,400 --> 00:17:34,880 It's been so successful with other ancient languages. 374 00:17:34,880 --> 00:17:39,040 Yeah, researchers are hopeful they're using machine learning to analyze the script, 375 00:17:39,040 --> 00:17:42,320 looking for patterns that might tell us what the symbols mean. 376 00:17:42,320 --> 00:17:43,520 It's like a giant puzzle. 377 00:17:43,520 --> 00:17:44,080 Yeah. 378 00:17:44,080 --> 00:17:46,240 But with ancient symbols, you said pictures. 379 00:17:46,240 --> 00:17:47,040 Exactly. 380 00:17:47,040 --> 00:17:48,720 And as AI gets better, 381 00:17:48,720 --> 00:17:50,400 I think we're going to see some big breakthroughs 382 00:17:50,400 --> 00:17:52,560 in our understanding of the Minoans. 383 00:17:52,560 --> 00:17:54,560 So we might be able to read their writing someday. 384 00:17:54,560 --> 00:17:55,680 It'd be amazing. 385 00:17:55,680 --> 00:17:57,440 So we've talked about Linear A. 386 00:17:57,440 --> 00:17:59,200 We've talked about the Voynich manuscript. 387 00:17:59,200 --> 00:18:03,840 What other ancient civilizations are we learning more about, thanks to AI? 388 00:18:03,840 --> 00:18:07,920 Well, one really fascinating civilization that researchers are focusing on 389 00:18:07,920 --> 00:18:09,280 is the Etruscans. 390 00:18:09,280 --> 00:18:12,880 They lived in central Italy, before the Roman Empire. 391 00:18:12,880 --> 00:18:13,520 Oh, yeah. 392 00:18:13,520 --> 00:18:14,800 I remember reading about them. 393 00:18:14,800 --> 00:18:16,560 They have those elaborate tombs. 394 00:18:16,560 --> 00:18:17,120 Oh. 395 00:18:17,120 --> 00:18:18,800 And they love to have feasts. 396 00:18:18,800 --> 00:18:19,280 Yeah. 397 00:18:19,280 --> 00:18:21,440 And their language has always been a mystery. 398 00:18:21,440 --> 00:18:22,080 You got it. 399 00:18:22,080 --> 00:18:25,280 And even though we have a lot of Etruscan artifacts, 400 00:18:25,280 --> 00:18:28,080 their language has been a real challenge to understand. 401 00:18:28,080 --> 00:18:30,880 It doesn't seem to be related to any other language we know of. 402 00:18:30,880 --> 00:18:32,160 Oh, so it's like Iberian. 403 00:18:32,160 --> 00:18:32,720 Exactly. 404 00:18:32,720 --> 00:18:33,840 Another linguistic isolator. 405 00:18:33,840 --> 00:18:34,320 Wow. 406 00:18:34,320 --> 00:18:38,000 But researchers are using some really cool technology to try and decode it. 407 00:18:38,000 --> 00:18:40,640 They're using neural machine translation models, 408 00:18:40,640 --> 00:18:42,800 similar to the ones they used for Linear B, 409 00:18:42,800 --> 00:18:46,240 to analyze Etruscan texts and look for patterns. 410 00:18:46,240 --> 00:18:48,240 It's like having a digital Rosetta Stone. 411 00:18:48,240 --> 00:18:49,040 Pretty much. 412 00:18:49,040 --> 00:18:54,800 One really exciting project is focused on this 2,500-year-old steel artifact. 413 00:18:54,800 --> 00:18:56,240 It was found near Florence. 414 00:18:56,240 --> 00:18:58,640 And it's covered in Etruscan inscriptions. 415 00:18:58,640 --> 00:19:02,480 Researchers are hoping AI can help them figure out what it says. 416 00:19:02,480 --> 00:19:03,760 That's so cool. 417 00:19:03,760 --> 00:19:07,200 Imagine if we could finally understand the Etruscans. 418 00:19:07,200 --> 00:19:10,160 That would completely change how we think about ancient Italy. 419 00:19:10,160 --> 00:19:11,360 It really would. 420 00:19:11,360 --> 00:19:15,600 It shows how AI can help us connect with the past in ways we never imagined. 421 00:19:15,600 --> 00:19:17,680 And speaking of connecting with the past, 422 00:19:17,680 --> 00:19:22,640 you mentioned earlier that AI is being used to decode the waggle dance of bees. 423 00:19:22,640 --> 00:19:23,440 Oh, yeah. 424 00:19:23,440 --> 00:19:24,000 That's right. 425 00:19:24,000 --> 00:19:25,280 Can you tell me more about that? 426 00:19:25,280 --> 00:19:29,280 So researchers built this special beehive with see-through walls 427 00:19:29,280 --> 00:19:31,680 so they can film the bees with cameras. 428 00:19:31,680 --> 00:19:34,240 And then they feed all that data into an AI system. 429 00:19:34,240 --> 00:19:36,080 That can understand the waggle dances. 430 00:19:36,080 --> 00:19:38,080 Wow. So it's like a bee dance translator. 431 00:19:38,080 --> 00:19:38,880 Exactly. 432 00:19:38,880 --> 00:19:42,080 The AI looks at how the bees move, like the angles and how long they waggle. 433 00:19:42,080 --> 00:19:44,080 And it can figure out what they're trying to say. 434 00:19:44,080 --> 00:19:47,280 It can even pinpoint where food sources are based on the dance. 435 00:19:47,280 --> 00:19:48,080 That's wild. 436 00:19:48,080 --> 00:19:52,080 What could we do if we could actually understand bee communication? 437 00:19:52,080 --> 00:19:54,480 I think it could have a huge impact on things like 438 00:19:54,480 --> 00:19:56,480 agriculture and conservation. 439 00:19:56,480 --> 00:20:01,680 Imagine being able to understand how bees pollinate, how they navigate, 440 00:20:01,680 --> 00:20:04,880 and how we can protect them from things like pesticides. 441 00:20:04,880 --> 00:20:05,680 That would be incredible. 442 00:20:05,680 --> 00:20:09,680 It just shows that we can learn so much from nature if we pay attention. 443 00:20:09,680 --> 00:20:10,480 Well said. 444 00:20:10,480 --> 00:20:15,680 This whole field of study is really about being open to new possibilities. 445 00:20:15,680 --> 00:20:16,480 Definitely. 446 00:20:16,480 --> 00:20:19,680 It's been an incredible journey exploring AI and ancient languages. 447 00:20:19,680 --> 00:20:22,480 We've seen how it's changing, how we think about the past, 448 00:20:22,480 --> 00:20:24,480 and giving us hope for the future. 449 00:20:24,480 --> 00:20:25,280 I agree. 450 00:20:25,280 --> 00:20:29,280 It all comes back to human curiosity and our desire to learn. 451 00:20:29,280 --> 00:20:30,080 It really does. 452 00:20:30,080 --> 00:20:33,280 And to everyone listening, thank you for joining us on this deep dive. 453 00:20:33,280 --> 00:20:37,680 We hope you learned some cool stuff and that you're feeling inspired to see the world in new ways. 454 00:20:37,680 --> 00:20:53,680 That's all for today.