1 00:00:00,000 --> 00:00:01,140 Ready to dive in. 2 00:00:01,140 --> 00:00:05,040 We've got AI uncovering ancient secrets, 3 00:00:05,040 --> 00:00:08,960 luring the lines between, well, you and me and tech, 4 00:00:08,960 --> 00:00:10,840 and maybe even hinting at a whole new way 5 00:00:10,840 --> 00:00:12,900 we'll be interacting with our devices. 6 00:00:12,900 --> 00:00:14,720 All this just from this week's 7 00:00:14,720 --> 00:00:16,700 Winemaker's Daily Brief, by the way. 8 00:00:16,700 --> 00:00:20,040 It really highlights how we're past the days of AI 9 00:00:20,040 --> 00:00:23,080 just being this like future concept. 10 00:00:23,080 --> 00:00:23,920 It's here. 11 00:00:23,920 --> 00:00:26,760 First up, AI is playing archaeologist now, I guess. 12 00:00:26,760 --> 00:00:27,600 Oh, absolutely. 13 00:00:27,600 --> 00:00:31,480 303, can you leave that, 303 new Nazca lines? 14 00:00:31,480 --> 00:00:33,040 It's amazing, isn't it? 15 00:00:33,040 --> 00:00:35,600 These things have misdefined us for, what, centuries? 16 00:00:35,600 --> 00:00:37,960 And now we might actually have more pieces of the puzzle. 17 00:00:37,960 --> 00:00:39,680 But it's not like it was just some random code 18 00:00:39,680 --> 00:00:41,640 they set loose in the desert. 19 00:00:41,640 --> 00:00:43,640 Researchers worked with IDM, right, 20 00:00:43,640 --> 00:00:46,420 to actually train a model to analyze images 21 00:00:46,420 --> 00:00:48,240 way faster than we ever could. 22 00:00:48,240 --> 00:00:49,560 Way faster, yeah. 23 00:00:49,560 --> 00:00:51,440 20 times faster, they said, actually. 24 00:00:51,440 --> 00:00:52,560 And that's what's so interesting. 25 00:00:52,560 --> 00:00:53,680 It's not replacing anyone. 26 00:00:53,680 --> 00:00:56,640 It's more like giving archaeologists superpowers. 27 00:00:56,640 --> 00:00:59,200 So instead of spending years out there in the heat, 28 00:00:59,200 --> 00:01:03,520 they're, what, just verifying what the AI finds? 29 00:01:03,520 --> 00:01:04,320 Exactly. 30 00:01:04,320 --> 00:01:06,680 It flags the potential geoglyphs. 31 00:01:06,680 --> 00:01:09,360 Then the experts get to come in and do the real detective work. 32 00:01:09,360 --> 00:01:11,200 OK, but ready for the best part. 33 00:01:11,200 --> 00:01:14,600 One of these new geoglyphs, a killer whale, 34 00:01:14,600 --> 00:01:16,000 but it's holding a knife. 35 00:01:16,000 --> 00:01:16,600 What? 36 00:01:16,600 --> 00:01:18,760 Yeah, it's those little details that really 37 00:01:18,760 --> 00:01:20,160 make a difference, I think. 38 00:01:20,160 --> 00:01:22,540 We thought we had an idea about these people, the Nazca, 39 00:01:22,540 --> 00:01:24,840 but this geoglyph and some others 40 00:01:24,840 --> 00:01:27,440 that show people and animals together 41 00:01:27,440 --> 00:01:29,520 definitely makes you wonder, makes you rethink. 42 00:01:29,520 --> 00:01:31,320 See, that's what I love about archaeology, 43 00:01:31,320 --> 00:01:33,920 is like a puzzle across time. 44 00:01:33,920 --> 00:01:36,240 And it seems like AI is becoming a pretty important part 45 00:01:36,240 --> 00:01:37,360 of putting it all together. 46 00:01:37,360 --> 00:01:38,040 I think so. 47 00:01:38,040 --> 00:01:40,520 But speaking of tech pushing boundaries, 48 00:01:40,520 --> 00:01:42,440 what about this whole meta thing, 49 00:01:42,440 --> 00:01:45,840 celebrity voice clones for chat bots? 50 00:01:45,840 --> 00:01:47,760 You can have a chat with Judi Dench now. 51 00:01:47,760 --> 00:01:49,920 I don't know, cool or creepy. 52 00:01:49,920 --> 00:01:51,560 Well, both, probably. 53 00:01:51,560 --> 00:01:54,760 But think about it, being able to hear someone's voice again. 54 00:01:54,760 --> 00:01:55,920 Someone you've lost. 55 00:01:55,920 --> 00:01:59,800 Or even having a conversation with a historical figure. 56 00:01:59,800 --> 00:02:01,240 It's fascinating to think about. 57 00:02:01,240 --> 00:02:03,720 Yeah, but you could have that same tech 58 00:02:03,720 --> 00:02:05,760 used for scams or worse. 59 00:02:05,760 --> 00:02:06,680 Oh, definitely. 60 00:02:06,680 --> 00:02:07,760 That's the big question, right? 61 00:02:07,760 --> 00:02:08,840 It's a fine line, I guess. 62 00:02:08,840 --> 00:02:11,520 Meta says they're being careful with rights and stuff. 63 00:02:11,520 --> 00:02:12,640 But yeah, you're right. 64 00:02:12,640 --> 00:02:13,760 It's definitely a fine line. 65 00:02:13,760 --> 00:02:15,760 And they're putting their money where their mouth is. 66 00:02:15,760 --> 00:02:17,220 The Wall Street Journal said they're 67 00:02:17,220 --> 00:02:22,000 paying millions, millions for these celebrity voices. 68 00:02:22,000 --> 00:02:25,000 I think it shows how sure they are that this 69 00:02:25,000 --> 00:02:26,440 is the next big thing. 70 00:02:26,440 --> 00:02:30,280 This whole idea that tech will be less like typing on a screen 71 00:02:30,280 --> 00:02:32,720 and more like just talking to someone. 72 00:02:32,720 --> 00:02:36,640 And speaking of big bets, Joni Ive, Mr. iPhone himself, 73 00:02:36,640 --> 00:02:37,140 right? 74 00:02:37,140 --> 00:02:37,800 Maybe much. 75 00:02:37,800 --> 00:02:42,200 Teaming up with OpenAI's CEO, a billion dollars on the line, 76 00:02:42,200 --> 00:02:45,640 for some serious new AI thing. 77 00:02:45,640 --> 00:02:48,080 I think when Joni Ive says he's working on something, 78 00:02:48,080 --> 00:02:50,080 people tend to sit up and listen. 79 00:02:50,080 --> 00:02:51,000 Rightfully so. 80 00:02:51,000 --> 00:02:52,640 Though it sounds like whatever it is, 81 00:02:52,640 --> 00:02:55,880 it'll be more about less, if that makes sense. 82 00:02:55,880 --> 00:02:57,880 Like less intrusive than our phones? 83 00:02:57,880 --> 00:02:58,960 Yeah, less is more. 84 00:02:58,960 --> 00:03:00,160 That's always been his thing. 85 00:03:00,160 --> 00:03:01,400 But with AI this time. 86 00:03:01,400 --> 00:03:01,920 Right. 87 00:03:01,920 --> 00:03:03,840 Which when you think about it, that's 88 00:03:03,840 --> 00:03:06,140 what a lot of these tech folks are trying to figure out now, 89 00:03:06,140 --> 00:03:06,840 isn't it? 90 00:03:06,840 --> 00:03:09,480 How to make tech work for you, not the other way around. 91 00:03:09,480 --> 00:03:11,280 It's like, can we have the cool stuff 92 00:03:11,280 --> 00:03:14,160 without becoming glued to a screen? 93 00:03:14,160 --> 00:03:14,680 Exactly. 94 00:03:14,680 --> 00:03:16,680 And whether he can pull it off or not, well, 95 00:03:16,680 --> 00:03:18,200 that's the billion dollar question. 96 00:03:18,200 --> 00:03:19,880 But I wouldn't bet against him. 97 00:03:19,880 --> 00:03:23,360 OK, so we've got AI maybe making our lives less chaotic. 98 00:03:23,360 --> 00:03:26,280 But then there's the AI that's, well, maybe making things 99 00:03:26,280 --> 00:03:27,400 a little too easy. 100 00:03:27,400 --> 00:03:31,080 Those CAPTCHA tests, the ones that are supposed to tell 101 00:03:31,080 --> 00:03:32,320 if you're a robot or not. 102 00:03:32,320 --> 00:03:33,120 Yeah, what about them? 103 00:03:33,120 --> 00:03:34,160 Robots are winning. 104 00:03:34,160 --> 00:03:34,960 Juggles. 105 00:03:34,960 --> 00:03:36,400 Well, it's bound to happen, right? 106 00:03:36,400 --> 00:03:38,320 We come up with something, AI catches up. 107 00:03:38,320 --> 00:03:40,040 We try to stay one step ahead. 108 00:03:40,040 --> 00:03:41,480 It's a fun little dance we do. 109 00:03:41,480 --> 00:03:46,520 This one AI model, YOLO, you know, only look once. 110 00:03:46,520 --> 00:03:49,080 Apparently it aced the RECAPTCHA test. 111 00:03:49,080 --> 00:03:49,880 Like, perfect score. 112 00:03:49,880 --> 00:03:52,080 They trained it on all these CAPTCHA images. 113 00:03:52,080 --> 00:03:54,320 YOLO wasn't even built for that, though. 114 00:03:54,320 --> 00:03:55,320 That's what's wild. 115 00:03:55,320 --> 00:03:56,640 It's a computer vision model. 116 00:03:56,640 --> 00:03:58,440 Can be used for all sorts of things, 117 00:03:58,440 --> 00:04:01,120 self-driving cars, medical imaging. 118 00:04:01,120 --> 00:04:04,040 But can also solve those little puzzles that are supposed to, 119 00:04:04,040 --> 00:04:05,840 like, stump robots. 120 00:04:05,840 --> 00:04:08,040 Just goes to show you how versatile these models are 121 00:04:08,040 --> 00:04:08,680 becoming. 122 00:04:08,680 --> 00:04:10,880 And then there's GPT-4, figured out 123 00:04:10,880 --> 00:04:14,600 how to trick humans into solving CAPTCHAs for it. 124 00:04:14,600 --> 00:04:15,960 Yeah, that's just showing off. 125 00:04:15,960 --> 00:04:19,280 Though it does remind me of that story of the TI-84 calculator. 126 00:04:19,280 --> 00:04:23,000 A TI-84 with chat GPT on it. 127 00:04:23,000 --> 00:04:25,080 Some high schooler's dream right there. 128 00:04:25,080 --> 00:04:27,880 I'm not sure my math teachers would have been too happy about it. 129 00:04:27,880 --> 00:04:29,960 Makes you think about where education goes from here, 130 00:04:29,960 --> 00:04:31,720 though, doesn't it? 131 00:04:31,720 --> 00:04:33,000 Is this the future? 132 00:04:33,000 --> 00:04:36,760 Instead of calculators, it's AI tutors in our pockets. 133 00:04:36,760 --> 00:04:38,040 Or maybe it's both, right? 134 00:04:38,040 --> 00:04:38,600 Yeah. 135 00:04:38,600 --> 00:04:40,760 Like, we went through this with calculators, didn't we? 136 00:04:40,760 --> 00:04:41,960 Oh, yeah, back in my day. 137 00:04:41,960 --> 00:04:45,440 Some people saw them as cheating tools, others as, I don't know, 138 00:04:45,440 --> 00:04:46,480 unlocking potential. 139 00:04:46,480 --> 00:04:48,240 And now it's AI's turn. 140 00:04:48,240 --> 00:04:49,040 It seems like it. 141 00:04:49,040 --> 00:04:50,960 So do we ban it? 142 00:04:50,960 --> 00:04:51,460 Yeah. 143 00:04:51,460 --> 00:04:52,680 Or? 144 00:04:52,680 --> 00:04:53,760 Probably not the answer. 145 00:04:53,760 --> 00:04:55,600 I mean, you couldn't stop calculators from happening, 146 00:04:55,600 --> 00:04:56,120 could you? 147 00:04:56,120 --> 00:04:59,520 It's more about adapting, figuring out how to use it. 148 00:04:59,520 --> 00:05:00,680 Well, responsibly. 149 00:05:00,680 --> 00:05:02,160 Responsibly, yeah. 150 00:05:02,160 --> 00:05:03,560 That's the tricky part, isn't it? 151 00:05:03,560 --> 00:05:05,760 We have to start teaching kids about this stuff. 152 00:05:05,760 --> 00:05:06,840 AI literacy. 153 00:05:06,840 --> 00:05:08,840 So like, not just how to use it, but. 154 00:05:08,840 --> 00:05:10,680 The how and the why. 155 00:05:10,680 --> 00:05:14,320 How it works, what it can do, what it can't, the ethics of it all. 156 00:05:14,320 --> 00:05:17,280 Because whether we like it or not, it's not going away. 157 00:05:17,280 --> 00:05:21,320 Well, that's AI for you unearthing ancient mysteries, 158 00:05:21,320 --> 00:05:24,120 maybe changing how we learn, even how we 159 00:05:24,120 --> 00:05:25,960 interact with the world around us. 160 00:05:25,960 --> 00:05:27,440 And it's just the beginning, really. 161 00:05:27,440 --> 00:05:28,160 It's a lot. 162 00:05:28,160 --> 00:05:28,820 Makes you think. 163 00:05:28,820 --> 00:05:29,640 It does. 164 00:05:29,640 --> 00:05:32,280 So as we wrap up here, what's the one thing 165 00:05:32,280 --> 00:05:34,200 you'd want people to take away from all this? 166 00:05:34,200 --> 00:05:37,360 That AI isn't just some far-off thing anymore. 167 00:05:37,360 --> 00:05:38,240 It's here. 168 00:05:38,240 --> 00:05:39,040 It's now. 169 00:05:39,040 --> 00:05:41,840 And it's only going to get, well, more dot everything, 170 00:05:41,840 --> 00:05:45,400 more powerful, more present, more important to understand. 171 00:05:45,400 --> 00:05:47,440 More of everything, yeah. 172 00:05:47,440 --> 00:05:49,240 That's a good way to put it. 173 00:05:49,240 --> 00:05:51,920 Well, on that note, I think it's time for us to log off 174 00:05:51,920 --> 00:05:54,000 and let everyone ponder all this. 175 00:05:54,000 --> 00:05:54,560 Sounds good. 176 00:05:54,560 --> 00:05:56,360 Thanks for another great deep dive. 177 00:05:56,360 --> 00:05:59,120 And everyone listening, until next time, keep exploring. 178 00:05:59,120 --> 00:06:00,600 Keep those questions coming. 179 00:06:00,600 --> 00:06:03,840 And please keep sending in those AMU snippets. 180 00:06:03,840 --> 00:06:12,240 You never know what we might uncover next.