1 00:00:00,000 --> 00:00:02,280 Welcome to the Daily AI News Podcast. 2 00:00:02,280 --> 00:00:05,640 We're diving deep into AI agents today 3 00:00:05,640 --> 00:00:08,880 and how AI regulation is changing. 4 00:00:08,880 --> 00:00:11,480 Yeah, it's a really interesting time for AI right now. 5 00:00:11,480 --> 00:00:12,320 It is. 6 00:00:12,320 --> 00:00:13,160 Things are moving so fast. 7 00:00:13,160 --> 00:00:14,960 Yeah, we've got some pretty cool sources to look at. 8 00:00:14,960 --> 00:00:15,800 Yeah. 9 00:00:15,800 --> 00:00:19,200 New AI agents like Magnetic One from Microsoft 10 00:00:19,200 --> 00:00:21,680 and Jarvis AI from Google. 11 00:00:21,680 --> 00:00:23,160 And then of course the implications 12 00:00:23,160 --> 00:00:26,840 of the recent US elections on AI regulations. 13 00:00:26,840 --> 00:00:28,640 Definitely a lot to unpack there. 14 00:00:28,640 --> 00:00:29,560 Yeah, for sure. 15 00:00:29,560 --> 00:00:32,160 You know, it's exciting to watch AI evolve 16 00:00:32,160 --> 00:00:35,720 from just a simple tool to more of a real collaborator, 17 00:00:35,720 --> 00:00:39,480 something that can handle those complex real world tasks. 18 00:00:39,480 --> 00:00:40,320 Yeah. 19 00:00:40,320 --> 00:00:42,280 But the more advanced these AI agents get, 20 00:00:42,280 --> 00:00:44,440 the more we need to think about how we manage them. 21 00:00:44,440 --> 00:00:45,720 For sure how we regulate them. 22 00:00:45,720 --> 00:00:46,560 Exactly. 23 00:00:46,560 --> 00:00:48,840 Keep everything safe, responsible. 24 00:00:48,840 --> 00:00:51,040 So let's start with Magnetic One. 25 00:00:51,040 --> 00:00:53,600 This new multi-agent system from Microsoft. 26 00:00:53,600 --> 00:00:54,800 Sounds good. 27 00:00:54,800 --> 00:00:56,920 It's designed to automate those complex tasks. 28 00:00:56,920 --> 00:01:00,120 Ordering food, ranging deliveries all through a system 29 00:01:00,120 --> 00:01:02,720 of these specialized AI agents. 30 00:01:02,720 --> 00:01:06,160 It's really fascinating how they've structured Magnetic One. 31 00:01:06,160 --> 00:01:08,240 They have this orchestrator agent 32 00:01:08,240 --> 00:01:11,040 that's basically like the project manager, you know? 33 00:01:11,040 --> 00:01:11,880 Okay. 34 00:01:11,880 --> 00:01:16,720 It delegates tasks to four specialized sub agents. 35 00:01:16,720 --> 00:01:17,560 Okay. 36 00:01:17,560 --> 00:01:20,880 Web surfer, file surfer, coder, and computer terminal. 37 00:01:20,880 --> 00:01:22,840 So each one has a specific role. 38 00:01:22,840 --> 00:01:23,680 Exactly. 39 00:01:23,680 --> 00:01:25,880 And they all work together to get the job done. 40 00:01:25,880 --> 00:01:27,840 Kind of like a team of specialists, 41 00:01:27,840 --> 00:01:29,560 each with their own skill set. 42 00:01:29,560 --> 00:01:31,640 So like web surfer would browse the internet. 43 00:01:31,640 --> 00:01:32,480 Right. 44 00:01:32,480 --> 00:01:35,880 File surfer digs through files, coder writes code. 45 00:01:35,880 --> 00:01:36,720 Yeah. 46 00:01:36,720 --> 00:01:39,760 And computer terminal interacts with like a console shell. 47 00:01:39,760 --> 00:01:40,600 Exactly. 48 00:01:40,600 --> 00:01:42,320 Like if you needed to book a flight, for example. 49 00:01:42,320 --> 00:01:43,160 Okay. 50 00:01:43,160 --> 00:01:44,560 The orchestrator would tell web surfer 51 00:01:44,560 --> 00:01:47,880 to go find the best deals based on your criteria. 52 00:01:47,880 --> 00:01:50,840 File surfer might check your travel preferences 53 00:01:50,840 --> 00:01:52,000 from the past. 54 00:01:52,000 --> 00:01:53,800 Then coder fills out all the forms 55 00:01:53,800 --> 00:01:55,600 and computer terminal makes sure the payment 56 00:01:55,600 --> 00:01:56,760 goes through securely. 57 00:01:56,760 --> 00:01:57,600 Wow. 58 00:01:57,600 --> 00:01:58,920 All these agents working together 59 00:01:58,920 --> 00:02:00,600 is what lets magnetic one tackle 60 00:02:00,600 --> 00:02:02,440 these more complex tasks. 61 00:02:02,440 --> 00:02:04,440 Yeah, they used to need a person to do them. 62 00:02:04,440 --> 00:02:05,280 Exactly. 63 00:02:05,280 --> 00:02:06,120 That's pretty incredible. 64 00:02:06,120 --> 00:02:06,960 It is. 65 00:02:06,960 --> 00:02:09,080 So how does magnetic one stack up 66 00:02:09,080 --> 00:02:11,840 against other AI agents that are out there? 67 00:02:11,840 --> 00:02:12,680 Okay. 68 00:02:12,680 --> 00:02:13,520 Like the ones from Anthropic. 69 00:02:13,520 --> 00:02:15,640 Well, one big difference is how much access it has 70 00:02:15,640 --> 00:02:16,520 to your computer. 71 00:02:16,520 --> 00:02:17,360 Okay. 72 00:02:17,360 --> 00:02:19,120 Magnetic one can actually interact 73 00:02:19,120 --> 00:02:21,800 with your computer's command line interface. 74 00:02:21,800 --> 00:02:22,640 Oh, God. 75 00:02:22,640 --> 00:02:25,320 Which opens up a ton of possibilities for automation. 76 00:02:25,320 --> 00:02:26,160 Yeah, a bit. 77 00:02:26,160 --> 00:02:28,760 But it also brings up some security issues. 78 00:02:28,760 --> 00:02:29,600 Of course, yeah. 79 00:02:29,600 --> 00:02:31,200 Which Microsoft is trying to address 80 00:02:31,200 --> 00:02:35,200 by recommending things like containerization 81 00:02:35,200 --> 00:02:37,800 and limiting its access to the internet. 82 00:02:37,800 --> 00:02:38,640 Okay. 83 00:02:38,640 --> 00:02:39,880 They're really stressing the importance 84 00:02:39,880 --> 00:02:41,040 of human oversight. 85 00:02:41,040 --> 00:02:41,880 Right. 86 00:02:41,880 --> 00:02:42,840 Which is super important 87 00:02:42,840 --> 00:02:44,880 as these systems become more powerful. 88 00:02:44,880 --> 00:02:48,000 Yeah, it's not just about making a super intelligent AI. 89 00:02:48,000 --> 00:02:48,840 Right. 90 00:02:48,840 --> 00:02:50,200 But about doing it responsibly. 91 00:02:50,200 --> 00:02:51,040 Absolutely. 92 00:02:51,040 --> 00:02:52,520 And another thing to think about 93 00:02:52,520 --> 00:02:54,560 is the language model it uses. 94 00:02:54,560 --> 00:02:55,300 Okay. 95 00:02:55,300 --> 00:02:57,920 The magnetic one currently runs on GPT-4, 96 00:02:57,920 --> 00:03:01,160 which is a large language model, or LLM. 97 00:03:01,160 --> 00:03:02,680 Okay, I've heard that term LLM. 98 00:03:02,680 --> 00:03:03,520 Yeah. 99 00:03:03,520 --> 00:03:05,760 But what makes it so important for these AI agents? 100 00:03:05,760 --> 00:03:08,560 Think of it like the brain of the AI agent. 101 00:03:08,560 --> 00:03:09,400 Okay. 102 00:03:09,400 --> 00:03:10,520 It's what allows it to understand 103 00:03:10,520 --> 00:03:13,320 and generate text that sounds like a person wrote it. 104 00:03:14,440 --> 00:03:18,000 So without a good LLM, like GPT-4, 105 00:03:18,000 --> 00:03:20,400 it wouldn't be able to understand our commands. 106 00:03:20,400 --> 00:03:21,240 Right. 107 00:03:21,240 --> 00:03:22,400 Or process information. 108 00:03:22,400 --> 00:03:24,520 Or even respond in a way that makes sense. 109 00:03:24,520 --> 00:03:25,280 Exactly. 110 00:03:25,280 --> 00:03:28,280 Basically, the better the LLM, the smarter the AI. 111 00:03:28,280 --> 00:03:29,120 You got it. 112 00:03:29,120 --> 00:03:32,280 And magnetic one isn't limited to just GPT-4, right? 113 00:03:32,280 --> 00:03:33,120 That's right. 114 00:03:33,120 --> 00:03:36,480 Microsoft has said that other LLMs could be used to. 115 00:03:36,480 --> 00:03:39,360 Yeah, as long as they have strong reasoning capabilities. 116 00:03:39,360 --> 00:03:42,560 Okay, so staying on the topic of AI agents, 117 00:03:42,560 --> 00:03:44,200 let's talk about Google 118 00:03:44,200 --> 00:03:47,120 and their highly anticipated Jarvis AI. 119 00:03:47,120 --> 00:03:48,080 Okay. 120 00:03:48,080 --> 00:03:51,560 It's Google's answer to the demand for AI 121 00:03:51,560 --> 00:03:52,840 help with web browsing. 122 00:03:52,840 --> 00:03:53,680 Right. 123 00:03:53,680 --> 00:03:56,520 They're trying to work inside the Chrome browser 124 00:03:56,520 --> 00:03:59,400 to automate things like online shopping, booking flights, 125 00:03:59,400 --> 00:04:01,040 or even research. 126 00:04:01,040 --> 00:04:03,160 So it's like your own personal assistant for the internet. 127 00:04:03,160 --> 00:04:04,000 Exactly. 128 00:04:04,000 --> 00:04:06,520 I remember hearing rumors about Jarvis a while ago. 129 00:04:06,520 --> 00:04:08,720 Seems like things are finally moving forward. 130 00:04:08,720 --> 00:04:11,160 I think there was even a little mishap with the release. 131 00:04:11,160 --> 00:04:14,800 Oh yeah, there was that accidental leak of a prototype 132 00:04:14,800 --> 00:04:16,640 on the Chrome extension store. 133 00:04:16,640 --> 00:04:17,560 They took it down quickly, 134 00:04:17,560 --> 00:04:20,080 but it gave us a little peek into what they're working on. 135 00:04:20,080 --> 00:04:22,000 I'm really curious to see how it compares 136 00:04:22,000 --> 00:04:25,440 to other AI agents. 137 00:04:25,440 --> 00:04:26,280 Yeah. 138 00:04:26,280 --> 00:04:29,560 Like Anthropics, and that rumored one from OpenAI. 139 00:04:29,560 --> 00:04:31,520 Yeah, it seems like everyone is jumping 140 00:04:31,520 --> 00:04:33,840 on the AI agent train these days. 141 00:04:33,840 --> 00:04:36,480 Yeah, it's a testament to how powerful this tech can be. 142 00:04:36,480 --> 00:04:37,560 For sure. 143 00:04:37,560 --> 00:04:39,960 We're moving towards a future where AI agents 144 00:04:39,960 --> 00:04:44,120 are just seamlessly integrated into our digital lives, 145 00:04:44,120 --> 00:04:46,000 taking care of all sorts of tasks for us. 146 00:04:46,000 --> 00:04:48,080 Yeah, freeing up our time for other things. 147 00:04:48,080 --> 00:04:48,960 That sounds nice. 148 00:04:48,960 --> 00:04:49,920 It does. 149 00:04:49,920 --> 00:04:52,160 But the more we rely on these AI agents, 150 00:04:52,160 --> 00:04:54,880 the more we have to think about regulations 151 00:04:54,880 --> 00:04:55,920 and ethical stuff. 152 00:04:55,920 --> 00:04:57,040 Yeah, absolutely. 153 00:04:57,040 --> 00:04:58,080 And speaking of regulation, 154 00:04:58,080 --> 00:04:59,480 that's been a big topic lately. 155 00:04:59,480 --> 00:05:00,320 Oh yeah. 156 00:05:00,320 --> 00:05:03,360 Especially when it comes to using AI and hiring. 157 00:05:03,360 --> 00:05:05,440 Definitely, AI is popping up everywhere 158 00:05:05,440 --> 00:05:08,440 and hiring these days from screening resumes 159 00:05:08,440 --> 00:05:11,840 and picking candidates to doing interviews 160 00:05:11,840 --> 00:05:14,240 and even keeping an eye on how employees are doing. 161 00:05:14,240 --> 00:05:15,920 It seems like a lot of companies are using it 162 00:05:15,920 --> 00:05:17,480 to make hiring more efficient. 163 00:05:17,480 --> 00:05:18,320 Yeah. 164 00:05:18,320 --> 00:05:19,840 But I wonder what people think about it. 165 00:05:19,840 --> 00:05:21,400 There's definitely some hesitation. 166 00:05:21,400 --> 00:05:22,240 I bet. 167 00:05:22,240 --> 00:05:24,680 A recent Pew Research Center survey 168 00:05:24,680 --> 00:05:28,040 showed that most people in the US are against using AI 169 00:05:28,040 --> 00:05:30,320 to make final hiring decisions. 170 00:05:30,320 --> 00:05:31,160 Makes sense. 171 00:05:31,160 --> 00:05:34,040 They're worried about things like fairness, transparency, 172 00:05:34,040 --> 00:05:36,320 and bias in these AI systems. 173 00:05:36,320 --> 00:05:38,200 Yeah, those are all valid concerns. 174 00:05:38,200 --> 00:05:40,440 We've seen examples where AI systems 175 00:05:40,440 --> 00:05:43,360 just end up perpetuating existing biases. 176 00:05:43,360 --> 00:05:46,160 Right, and that leads to unfair outcomes. 177 00:05:46,160 --> 00:05:48,520 Exactly like the case of Laura Cummings. 178 00:05:48,520 --> 00:05:49,560 I don't think I'm familiar. 179 00:05:49,560 --> 00:05:51,160 She's a qualified candidate who thinks 180 00:05:51,160 --> 00:05:54,560 she was unfairly screened out by an AI resume tool. 181 00:05:54,560 --> 00:05:55,800 Oh wow. 182 00:05:55,800 --> 00:05:59,920 This really highlights the black box problem with AI. 183 00:05:59,920 --> 00:06:01,400 Yeah, it's hard to understand 184 00:06:01,400 --> 00:06:03,240 how these systems make their decisions. 185 00:06:03,240 --> 00:06:05,720 Right, which makes it tricky to hold them accountable. 186 00:06:05,720 --> 00:06:06,960 And to make sure they're being fair. 187 00:06:06,960 --> 00:06:07,800 Exactly. 188 00:06:07,800 --> 00:06:10,360 So it seems like companies want to use AI 189 00:06:10,360 --> 00:06:11,600 to make hiring easier. 190 00:06:11,600 --> 00:06:12,440 Yeah. 191 00:06:12,440 --> 00:06:13,920 But they also need to make sure it's done ethically. 192 00:06:13,920 --> 00:06:14,840 For sure. 193 00:06:14,840 --> 00:06:18,160 So what can companies do to address these concerns? 194 00:06:18,160 --> 00:06:19,360 That's the big question. 195 00:06:19,360 --> 00:06:21,160 Well, first they need to educate themselves 196 00:06:21,160 --> 00:06:24,240 about potential biases in these AI systems 197 00:06:24,240 --> 00:06:27,640 and understand the ethics of using them in hiring. 198 00:06:27,640 --> 00:06:28,520 Makes sense. 199 00:06:28,520 --> 00:06:31,720 Then they should do thorough risk assessments 200 00:06:31,720 --> 00:06:36,080 to identify and fix any potential biases in their AI tools. 201 00:06:36,080 --> 00:06:37,960 So kind of like a checkup for their AI. 202 00:06:37,960 --> 00:06:38,800 Exactly. 203 00:06:38,800 --> 00:06:42,840 And finally, they need to test these systems internally 204 00:06:42,840 --> 00:06:46,480 really rigorously to make sure they're working properly 205 00:06:46,480 --> 00:06:48,440 and not discriminating against anyone. 206 00:06:48,440 --> 00:06:51,920 So transparency and accountability are super important 207 00:06:51,920 --> 00:06:55,480 when it comes to this new world of AI-driven hiring. 208 00:06:55,480 --> 00:06:56,320 Absolutely. 209 00:06:56,320 --> 00:06:58,360 And government regulation could play a big role too. 210 00:06:58,360 --> 00:06:59,280 Okay, how so? 211 00:06:59,280 --> 00:07:02,000 We'll take the Minnesota Consumer Data Privacy Act, 212 00:07:02,000 --> 00:07:03,160 for example. 213 00:07:03,160 --> 00:07:06,160 It goes into effect in July 2025. 214 00:07:06,160 --> 00:07:07,000 Okay. 215 00:07:07,000 --> 00:07:07,960 And it aims to regulate 216 00:07:07,960 --> 00:07:10,160 automated decision-making processes. 217 00:07:10,160 --> 00:07:11,560 Including the ones used in hiring. 218 00:07:11,560 --> 00:07:13,680 Right, so this could be a good step towards 219 00:07:13,680 --> 00:07:16,560 fairness and transparency in AI hiring. 220 00:07:16,560 --> 00:07:18,880 So it sounds like both companies and policymakers 221 00:07:18,880 --> 00:07:20,320 have a responsibility here. 222 00:07:20,320 --> 00:07:21,160 Definitely. 223 00:07:21,160 --> 00:07:23,680 To make sure AI is used ethically and fairly. 224 00:07:23,680 --> 00:07:25,120 But what about the job seekers? 225 00:07:25,120 --> 00:07:25,960 Yeah. 226 00:07:25,960 --> 00:07:28,320 Is there anything they can do to adapt to this new landscape? 227 00:07:28,320 --> 00:07:29,720 Absolutely. 228 00:07:29,720 --> 00:07:32,920 They need to know that AI is being used more and more 229 00:07:32,920 --> 00:07:33,760 in hiring. 230 00:07:33,760 --> 00:07:34,600 Okay. 231 00:07:34,600 --> 00:07:37,640 And take steps to make their applications AI-friendly. 232 00:07:37,640 --> 00:07:38,800 Like what kind of steps? 233 00:07:38,800 --> 00:07:41,200 Well, things like optimizing their resumes 234 00:07:41,200 --> 00:07:44,120 and cover letters with relevant keywords. 235 00:07:44,120 --> 00:07:44,960 Okay. 236 00:07:44,960 --> 00:07:47,480 AI-powered tools to improve their writing 237 00:07:47,480 --> 00:07:48,800 and how they present themselves. 238 00:07:48,800 --> 00:07:52,080 So job hunting in the age of AI requires some new skills. 239 00:07:52,080 --> 00:07:52,920 It really does. 240 00:07:52,920 --> 00:07:53,920 It's a whole new ball game. 241 00:07:53,920 --> 00:07:54,760 Okay. 242 00:07:54,760 --> 00:07:55,600 Shifting gears a bit. 243 00:07:55,600 --> 00:07:56,440 Okay. 244 00:07:56,440 --> 00:07:58,640 Let's talk about another challenge in the AI world. 245 00:07:58,640 --> 00:08:00,000 AI hallucinations. 246 00:08:00,000 --> 00:08:00,840 Right. 247 00:08:00,840 --> 00:08:02,920 Those instances where AI systems generate 248 00:08:02,920 --> 00:08:05,000 false or misleading information. 249 00:08:05,000 --> 00:08:05,840 Yeah. 250 00:08:05,840 --> 00:08:07,720 Which can be a real problem when we're relying on them 251 00:08:07,720 --> 00:08:08,960 for accurate insights. 252 00:08:08,960 --> 00:08:10,600 It's a big concern for sure. 253 00:08:10,600 --> 00:08:13,360 AI hallucinations are a major roadblock 254 00:08:13,360 --> 00:08:15,720 to building trust in these systems. 255 00:08:15,720 --> 00:08:16,560 Right. 256 00:08:16,560 --> 00:08:17,960 Like when your AI assistant suddenly tells you 257 00:08:17,960 --> 00:08:19,920 to eat rocks or glue pizza together. 258 00:08:19,920 --> 00:08:21,040 Yeah. Exactly. 259 00:08:21,040 --> 00:08:23,000 That's funny, but also a little alarming. 260 00:08:23,000 --> 00:08:25,520 It shows the limits of current AI. 261 00:08:25,520 --> 00:08:27,480 And the need to keep improving it. 262 00:08:27,480 --> 00:08:28,320 Exactly. 263 00:08:28,320 --> 00:08:29,920 I remember reading about those examples. 264 00:08:29,920 --> 00:08:30,760 Yeah. 265 00:08:30,760 --> 00:08:33,280 It makes you question how reliable AI really is. 266 00:08:33,280 --> 00:08:34,120 Right. 267 00:08:34,120 --> 00:08:36,040 Especially when we're using it for important stuff. 268 00:08:36,040 --> 00:08:39,120 So what are companies doing to tackle this problem? 269 00:08:39,120 --> 00:08:41,680 Companies like Microsoft are working on solutions. 270 00:08:41,680 --> 00:08:42,520 Okay. Good. 271 00:08:42,520 --> 00:08:44,400 They've filed a patent for something called 272 00:08:44,400 --> 00:08:47,120 a response augmenting system or RAS. 273 00:08:47,120 --> 00:08:47,960 RAS. 274 00:08:47,960 --> 00:08:50,880 Yeah. Think of it like a built in lie detector for AI. 275 00:08:50,880 --> 00:08:52,000 A lie detector for AI. 276 00:08:52,000 --> 00:08:52,840 That's interesting. 277 00:08:52,840 --> 00:08:53,680 Right. 278 00:08:53,680 --> 00:08:54,520 How does that work? 279 00:08:54,520 --> 00:08:56,720 So RAS is designed to double check the information 280 00:08:56,720 --> 00:08:58,120 that an AI generates. 281 00:08:58,120 --> 00:08:58,960 Okay. 282 00:08:58,960 --> 00:09:00,680 It compares it to external sources 283 00:09:00,680 --> 00:09:02,040 to make sure it's accurate. 284 00:09:02,040 --> 00:09:03,840 Kind of like a fact checker built right into the system. 285 00:09:03,840 --> 00:09:04,680 Exactly. 286 00:09:04,680 --> 00:09:06,640 So if the AI starts hallucinating, 287 00:09:06,640 --> 00:09:09,680 RAS can flag it and provide more reliable information. 288 00:09:09,680 --> 00:09:10,520 Right. 289 00:09:10,520 --> 00:09:13,720 That will help reduce the risk of AI spreading misinformation. 290 00:09:13,720 --> 00:09:15,240 Okay. That sounds promising. 291 00:09:15,240 --> 00:09:18,600 Are there any other tools or strategies in development? 292 00:09:18,600 --> 00:09:19,440 Definitely. 293 00:09:19,440 --> 00:09:23,000 Microsoft also has Azure AI content safety. 294 00:09:23,000 --> 00:09:23,840 Okay. 295 00:09:23,840 --> 00:09:27,000 Which is focused on detecting and filtering out harmful 296 00:09:27,000 --> 00:09:29,880 or inappropriate content that AI generates. 297 00:09:29,880 --> 00:09:30,840 So it's like a safety net. 298 00:09:30,840 --> 00:09:31,680 Yeah. 299 00:09:31,680 --> 00:09:32,960 To catch those problematic outputs 300 00:09:32,960 --> 00:09:34,240 before they get out there. 301 00:09:34,240 --> 00:09:36,320 So it's a multi-pronged approach. 302 00:09:36,320 --> 00:09:37,160 Yeah. 303 00:09:37,160 --> 00:09:39,800 Improving the AI to reduce hallucinations. 304 00:09:39,800 --> 00:09:40,480 Right. 305 00:09:40,480 --> 00:09:41,920 And putting safety measures in place 306 00:09:41,920 --> 00:09:43,560 to catch any that slip through. 307 00:09:43,560 --> 00:09:44,440 Exactly. 308 00:09:44,440 --> 00:09:48,320 It's all about making AI as accurate, reliable, 309 00:09:48,320 --> 00:09:49,960 and trustworthy as possible. 310 00:09:49,960 --> 00:09:51,440 Speaking of trust and reliability. 311 00:09:51,440 --> 00:09:52,280 Yeah. 312 00:09:52,280 --> 00:09:54,960 The recent US election brought up some interesting questions 313 00:09:54,960 --> 00:09:58,320 about the role of AI in providing election information. 314 00:09:58,320 --> 00:09:59,280 It did. 315 00:09:59,280 --> 00:10:01,680 A lot of AI labs decided to stay away 316 00:10:01,680 --> 00:10:03,160 from election predictions. 317 00:10:03,160 --> 00:10:04,000 Really? 318 00:10:04,000 --> 00:10:05,280 Because they were worried about hallucinations 319 00:10:05,280 --> 00:10:06,720 and spreading misinformation. 320 00:10:06,720 --> 00:10:07,560 I can see why. 321 00:10:07,560 --> 00:10:10,080 But one company, Perplexity, 322 00:10:10,080 --> 00:10:11,480 decided to go for it. 323 00:10:11,480 --> 00:10:12,360 Perplexity? 324 00:10:12,360 --> 00:10:13,200 Yeah. 325 00:10:13,200 --> 00:10:15,000 They're known for their AI powered search engine 326 00:10:15,000 --> 00:10:16,000 and chatbot. 327 00:10:16,000 --> 00:10:16,840 Okay. 328 00:10:16,840 --> 00:10:18,240 During the election, they used their chatbot 329 00:10:18,240 --> 00:10:20,480 to give real time election insights. 330 00:10:20,480 --> 00:10:21,560 Wow, that's bold. 331 00:10:21,560 --> 00:10:22,400 It was. 332 00:10:22,400 --> 00:10:24,800 Especially considering how much attention is on elections. 333 00:10:24,800 --> 00:10:25,640 Oh wow. 334 00:10:25,640 --> 00:10:26,480 So how did they do? 335 00:10:26,480 --> 00:10:27,320 Yeah. 336 00:10:27,320 --> 00:10:28,320 Were they able to provide accurate information? 337 00:10:28,320 --> 00:10:30,200 They actually did a pretty good job. 338 00:10:30,200 --> 00:10:34,480 Their strategy was to rely on solid data sources. 339 00:10:34,480 --> 00:10:35,320 Okay. 340 00:10:35,320 --> 00:10:37,960 Like democracy works and the associated press. 341 00:10:37,960 --> 00:10:40,280 So they weren't trying to predict the outcome themselves. 342 00:10:40,280 --> 00:10:41,120 Right. 343 00:10:41,120 --> 00:10:42,760 They were more about gathering and presenting info 344 00:10:42,760 --> 00:10:44,040 from trusted sources. 345 00:10:44,040 --> 00:10:45,560 That seems like a smarter approach. 346 00:10:45,560 --> 00:10:46,400 It was. 347 00:10:46,400 --> 00:10:48,080 And while there were a few small hiccups 348 00:10:48,080 --> 00:10:50,240 and some minor hallucinations, 349 00:10:50,240 --> 00:10:53,800 overall, their chatbot provided pretty accurate 350 00:10:53,800 --> 00:10:56,240 and reliable election information. 351 00:10:56,240 --> 00:10:57,080 That's impressive. 352 00:10:57,080 --> 00:11:00,480 Definitely more so than some other AI chatbots out there. 353 00:11:00,480 --> 00:11:04,040 It's interesting to see how they were able to use AI 354 00:11:04,040 --> 00:11:06,720 for real time election insights 355 00:11:06,720 --> 00:11:09,520 without spreading misinformation. 356 00:11:09,520 --> 00:11:10,360 It is. 357 00:11:10,360 --> 00:11:12,800 It makes you wonder about the future of AI in elections. 358 00:11:12,800 --> 00:11:13,640 Definitely. 359 00:11:13,640 --> 00:11:15,440 Well, we see more companies doing this in the future. 360 00:11:15,440 --> 00:11:16,280 Right. 361 00:11:16,280 --> 00:11:17,880 And how will it change how we consume 362 00:11:17,880 --> 00:11:20,000 and interpret election info? 363 00:11:20,000 --> 00:11:22,680 It's something to think about as AI continues to evolve. 364 00:11:22,680 --> 00:11:23,520 For sure. 365 00:11:23,520 --> 00:11:27,040 So we've talked about AI agents and AI in hiring. 366 00:11:27,040 --> 00:11:27,880 Right. 367 00:11:27,880 --> 00:11:29,960 And even AI hallucinations. 368 00:11:29,960 --> 00:11:31,800 Now let's switch gears to the food industry. 369 00:11:31,800 --> 00:11:32,640 Okay. 370 00:11:32,640 --> 00:11:34,600 I was reading about a company called Profile Print. 371 00:11:34,600 --> 00:11:37,560 That's using AI to revolutionize 372 00:11:37,560 --> 00:11:40,160 how we assess the quality of food ingredients. 373 00:11:40,160 --> 00:11:43,680 Profile Print is a cool example of how AI is being used 374 00:11:43,680 --> 00:11:46,440 to solve real world problems in the food industry. 375 00:11:46,440 --> 00:11:47,320 Yeah, it sounds it. 376 00:11:47,320 --> 00:11:50,240 They've developed a system that uses wave detectors 377 00:11:50,240 --> 00:11:54,160 and AI analysis to create digital fingerprints 378 00:11:54,160 --> 00:11:55,560 of food ingredients. 379 00:11:55,560 --> 00:11:56,640 Digital fingerprints? 380 00:11:56,640 --> 00:11:58,560 That sounds like something out of a sci-fi movie. 381 00:11:58,560 --> 00:11:59,520 It does, right? 382 00:11:59,520 --> 00:12:00,440 How does that work? 383 00:12:00,440 --> 00:12:02,040 It's actually pretty clever. 384 00:12:02,040 --> 00:12:05,000 They shine a special light wave onto the food ingredient. 385 00:12:05,000 --> 00:12:05,840 Okay. 386 00:12:05,840 --> 00:12:07,480 And then they analyze how the light interacts 387 00:12:07,480 --> 00:12:09,600 with the molecules in the ingredient. 388 00:12:09,600 --> 00:12:12,320 This creates a unique spectral signature. 389 00:12:12,320 --> 00:12:13,160 Okay. 390 00:12:13,160 --> 00:12:14,320 Basically a digital fingerprint. 391 00:12:14,320 --> 00:12:15,160 Right. 392 00:12:15,160 --> 00:12:16,800 That captures all this detailed information 393 00:12:16,800 --> 00:12:20,520 about the ingredients, chemical makeup and quality. 394 00:12:20,520 --> 00:12:23,200 So they're using light to identify 395 00:12:23,200 --> 00:12:26,120 the unique characteristics of each ingredient. 396 00:12:26,120 --> 00:12:26,960 That's it. 397 00:12:26,960 --> 00:12:27,800 I see. 398 00:12:27,800 --> 00:12:29,320 So what are the benefits of this tech? 399 00:12:29,320 --> 00:12:31,160 Well, for one, it makes quality control 400 00:12:31,160 --> 00:12:32,280 way more streamlined. 401 00:12:32,280 --> 00:12:33,120 Okay. 402 00:12:33,120 --> 00:12:34,840 Instead of relying on human inspectors 403 00:12:34,840 --> 00:12:36,120 to check ingredients. 404 00:12:36,120 --> 00:12:38,480 Which can take a long time and be prone to errors. 405 00:12:38,480 --> 00:12:39,360 Exactly. 406 00:12:39,360 --> 00:12:41,840 Profile Prince tech gives you a fast 407 00:12:41,840 --> 00:12:44,000 and objective assessment of quality. 408 00:12:44,000 --> 00:12:46,320 That could save companies a lot of time and money. 409 00:12:46,320 --> 00:12:47,160 For sure. 410 00:12:47,160 --> 00:12:49,480 And it ensures that their ingredients are consistent. 411 00:12:49,480 --> 00:12:50,320 That's great. 412 00:12:50,320 --> 00:12:51,840 Are there any other advantages? 413 00:12:51,840 --> 00:12:52,680 Definitely. 414 00:12:52,680 --> 00:12:54,400 It also helps reduce food waste. 415 00:12:54,400 --> 00:12:55,240 Okay. 416 00:12:55,240 --> 00:12:59,000 By identifying bad or low quality ingredients early on, 417 00:12:59,000 --> 00:13:02,080 companies can avoid using them in their products. 418 00:13:02,080 --> 00:13:04,040 And that minimizes waste 419 00:13:04,040 --> 00:13:05,600 and makes the industry more sustainable. 420 00:13:05,600 --> 00:13:06,520 Exactly. 421 00:13:06,520 --> 00:13:08,080 So it's not just about efficiency 422 00:13:08,080 --> 00:13:09,800 but about sustainability too. 423 00:13:09,800 --> 00:13:10,800 For sure. 424 00:13:10,800 --> 00:13:14,200 Maybe most importantly, it helps ensure food safety. 425 00:13:14,200 --> 00:13:15,040 Okay. 426 00:13:15,040 --> 00:13:17,080 By detecting contaminants or things 427 00:13:17,080 --> 00:13:18,840 that shouldn't be in the ingredients. 428 00:13:18,840 --> 00:13:19,680 Uh huh. 429 00:13:19,680 --> 00:13:22,600 Profile Prince tech can stop those substances 430 00:13:22,600 --> 00:13:24,040 from getting into the food supply. 431 00:13:24,040 --> 00:13:24,880 Right. 432 00:13:24,880 --> 00:13:26,600 Protecting consumers from potential harm. 433 00:13:26,600 --> 00:13:27,440 Exactly. 434 00:13:27,440 --> 00:13:28,280 So it's a win-win-win. 435 00:13:28,280 --> 00:13:28,960 It is. 436 00:13:28,960 --> 00:13:29,920 Proved efficiency. 437 00:13:29,920 --> 00:13:30,760 Uh huh. 438 00:13:30,760 --> 00:13:31,600 Reduced waste. 439 00:13:31,600 --> 00:13:32,440 Yep. 440 00:13:32,440 --> 00:13:33,280 And enhanced food safety. 441 00:13:33,280 --> 00:13:34,120 That's right. 442 00:13:34,120 --> 00:13:36,080 Are there any specific examples 443 00:13:36,080 --> 00:13:37,760 of how this is being used? 444 00:13:37,760 --> 00:13:40,160 They've been very successful in the coffee industry. 445 00:13:40,160 --> 00:13:41,000 Coffee? 446 00:13:41,000 --> 00:13:41,840 Yeah. 447 00:13:41,840 --> 00:13:44,560 Their technology can identify defective coffee deans. 448 00:13:44,560 --> 00:13:47,320 Which are hard to spot with the human eye. 449 00:13:47,320 --> 00:13:48,240 Exactly. 450 00:13:48,240 --> 00:13:50,160 So they're helping coffee producers 451 00:13:50,160 --> 00:13:51,960 make sure their beans are high quality. 452 00:13:51,960 --> 00:13:52,800 Right. 453 00:13:52,800 --> 00:13:55,080 Which means a better cup of coffee for us. 454 00:13:55,080 --> 00:13:57,520 So next time I'm enjoying a good cup of coffee, 455 00:13:57,520 --> 00:13:58,920 I might have AI to think. 456 00:13:58,920 --> 00:13:59,880 You just might. 457 00:13:59,880 --> 00:14:03,040 And Profile Prince isn't the only company using AI 458 00:14:03,040 --> 00:14:05,080 to transform the food industry. 459 00:14:05,080 --> 00:14:05,920 There are others. 460 00:14:05,920 --> 00:14:08,040 There's a whole wave of innovation happening. 461 00:14:08,040 --> 00:14:08,880 Like what? 462 00:14:08,880 --> 00:14:11,280 Well, there's a company called Aeromics. 463 00:14:11,280 --> 00:14:12,120 Aeromics. 464 00:14:12,120 --> 00:14:12,940 Yeah. 465 00:14:12,940 --> 00:14:14,640 They're using AI to digitally capture 466 00:14:14,640 --> 00:14:16,760 and reproduce flavors and scents. 467 00:14:16,760 --> 00:14:17,600 Wow. 468 00:14:17,600 --> 00:14:19,040 That sounds like something straight out of Willy Wonka. 469 00:14:19,040 --> 00:14:20,000 It does, doesn't it? 470 00:14:20,000 --> 00:14:21,200 And then there's Brightseed. 471 00:14:21,200 --> 00:14:22,040 Brightseed. 472 00:14:22,040 --> 00:14:24,520 They're using AI to figure out the hidden health 473 00:14:24,520 --> 00:14:26,000 benefits of plants. 474 00:14:26,000 --> 00:14:26,840 Oh wow. 475 00:14:26,840 --> 00:14:29,760 They analyze tons of plant compounds 476 00:14:29,760 --> 00:14:31,640 to find the ones that might have medicinal 477 00:14:31,640 --> 00:14:33,040 or nutritional properties. 478 00:14:33,040 --> 00:14:35,440 So that could lead to like new natural remedies 479 00:14:35,440 --> 00:14:36,440 and healthier food. 480 00:14:36,440 --> 00:14:37,360 Exactly. 481 00:14:37,360 --> 00:14:40,160 AI is really impacting every part of the food industry. 482 00:14:40,160 --> 00:14:41,280 It really is. 483 00:14:41,280 --> 00:14:45,040 From how food is produced and quality control 484 00:14:45,040 --> 00:14:47,000 to developing new products. 485 00:14:47,000 --> 00:14:49,360 And even discovering new health benefits. 486 00:14:49,360 --> 00:14:50,920 It's pretty amazing. 487 00:14:50,920 --> 00:14:53,920 So we've talked about AI agents that can help us online. 488 00:14:53,920 --> 00:14:54,520 Yeah. 489 00:14:54,520 --> 00:14:57,760 And AI that's changing how companies hire people. 490 00:14:57,760 --> 00:14:59,360 And AI hallucinations. 491 00:14:59,360 --> 00:14:59,840 Right. 492 00:14:59,840 --> 00:15:02,440 And even AI-powered taste testers. 493 00:15:02,440 --> 00:15:05,000 And digital flavor creators. 494 00:15:05,000 --> 00:15:07,840 It's clear that AI is rapidly changing the world. 495 00:15:07,840 --> 00:15:08,600 Yeah. 496 00:15:08,600 --> 00:15:09,520 It's an exciting time. 497 00:15:09,520 --> 00:15:10,000 It is. 498 00:15:10,000 --> 00:15:11,560 But like we've been talking about, 499 00:15:11,560 --> 00:15:14,720 it's super important to approach these developments 500 00:15:14,720 --> 00:15:16,160 thoughtfully and responsibly. 501 00:15:16,160 --> 00:15:17,080 I completely agree. 502 00:15:17,080 --> 00:15:19,720 We need to make sure AI is used ethically 503 00:15:19,720 --> 00:15:21,280 and for the good of everyone. 504 00:15:21,280 --> 00:15:21,780 Right. 505 00:15:21,780 --> 00:15:25,640 AI gets more powerful and becomes a bigger part of our lives. 506 00:15:25,640 --> 00:15:28,680 It's up to all of us to guide its development 507 00:15:28,680 --> 00:15:31,840 and make sure it's used in a way that benefits everyone. 508 00:15:31,840 --> 00:15:32,440 Well said. 509 00:15:32,440 --> 00:15:34,040 We need to ask the hard questions. 510 00:15:34,040 --> 00:15:34,440 Yeah. 511 00:15:34,440 --> 00:15:36,760 Think about the potential consequences 512 00:15:36,760 --> 00:15:40,480 and work together to create a future where AI serves humanity. 513 00:15:40,480 --> 00:15:41,480 Not the other way around. 514 00:15:41,480 --> 00:15:42,280 Exactly. 515 00:15:42,280 --> 00:15:44,280 This deep dive has been really insightful. 516 00:15:44,280 --> 00:15:45,320 It has. 517 00:15:45,320 --> 00:15:49,120 We've seen just how much potential AI has. 518 00:15:49,120 --> 00:15:51,840 But also how important it is to be careful. 519 00:15:51,840 --> 00:15:52,480 Absolutely. 520 00:15:52,480 --> 00:15:54,840 We need ethical guidelines and open discussions. 521 00:15:54,840 --> 00:15:55,360 For sure. 522 00:15:55,360 --> 00:15:57,960 To navigate this new technological frontier. 523 00:15:57,960 --> 00:15:59,480 It's a journey we're all on together. 524 00:15:59,480 --> 00:16:00,000 Yeah. 525 00:16:00,000 --> 00:16:03,360 And the choices we make now will shape the world of tomorrow. 526 00:16:03,360 --> 00:16:04,200 For sure. 527 00:16:04,200 --> 00:16:06,920 Thanks for listening to the Daily AI News podcast. 528 00:16:06,920 --> 00:16:19,760 And stay tuned for more.