WEBVTT 00:00:00.000 --> 00:00:02.120 Hey there. Welcome back to another Deep Talk. 00:00:02.259 --> 00:00:06.219 Today, we are absolutely tearing into a stack 00:00:06.219 --> 00:00:09.640 of recent insights all about AI, but specifically, 00:00:09.900 --> 00:00:12.039 you know, what it's actually doing to the world 00:00:12.039 --> 00:00:14.199 of work in technology right now, like hitting 00:00:14.199 --> 00:00:15.839 the ground. It feels like everyone's talking 00:00:15.839 --> 00:00:18.219 about AI, right? But you often hear, OK, but 00:00:18.219 --> 00:00:21.519 what's the real impact? Is this just hype? Exactly. 00:00:21.519 --> 00:00:24.640 So we've got this. Really meaty report from PWC 00:00:24.640 --> 00:00:26.800 plus a whole bunch of other fascinating snippets, 00:00:26.820 --> 00:00:30.600 news items, tool releases, you name it. Our mission 00:00:30.600 --> 00:00:33.020 is to cut through the noise for you, you know? 00:00:33.060 --> 00:00:35.320 Yeah. Like, are jobs just vanishing into thin 00:00:35.320 --> 00:00:37.520 air like some folks worry? Or are there these 00:00:37.520 --> 00:00:39.920 weird, unexpected opportunities popping up that 00:00:39.920 --> 00:00:42.320 maybe we're not seeing? That's really what we 00:00:42.320 --> 00:00:44.119 want to understand for you. What does the data, 00:00:44.259 --> 00:00:45.979 what does the actual news out there, what does 00:00:45.979 --> 00:00:49.299 it really show beyond the sometimes wild headlines? 00:00:49.619 --> 00:00:52.539 Totally. I got to tell you, buried in this data, 00:00:52.640 --> 00:00:54.719 there's one finding that honestly surprised me. 00:00:55.039 --> 00:00:59.000 It's about how getting AI skills. It can lead 00:00:59.000 --> 00:01:01.539 to a pay bump, but not just like a little one. 00:01:01.600 --> 00:01:03.859 We're talking a really big one, like surprisingly 00:01:03.859 --> 00:01:06.019 big. OK, I'm intrigued. We'll unpack that in 00:01:06.019 --> 00:01:09.040 just a sec. So let's kick off with this big anchor 00:01:09.040 --> 00:01:11.959 piece, this major report from PwC. They did this 00:01:11.959 --> 00:01:15.200 just massive thing. They looked at nearly a billion 00:01:15.200 --> 00:01:19.519 job ads globally. Wow. I mean, that's like. A 00:01:19.519 --> 00:01:21.659 wild amount of data to chew through, right? Yeah. 00:01:21.920 --> 00:01:24.180 Analyzing close to a billion job ads from six 00:01:24.180 --> 00:01:25.900 continents. That gives you a pretty powerful 00:01:25.900 --> 00:01:28.280 lens into what employers are actually looking 00:01:28.280 --> 00:01:31.040 for, what skills are in demand, and how things 00:01:31.040 --> 00:01:33.200 are changing on the ground. It's not survey data. 00:01:33.379 --> 00:01:35.879 It's actual job market signals. Right. It's not 00:01:35.879 --> 00:01:38.659 just theory floating around. And their headline 00:01:38.659 --> 00:01:41.579 finding, the one that really jumps out, it's 00:01:41.579 --> 00:01:43.780 this huge wage premium for people with AI skills. 00:01:44.159 --> 00:01:48.180 That number. 56%. What's fascinating here is 00:01:48.180 --> 00:01:50.959 that 56 % wage premium. That's significant. The 00:01:50.959 --> 00:01:53.099 report even calls it the AI bubble bonus, which 00:01:53.099 --> 00:01:55.799 is a pretty strong term, suggesting it's a significant 00:01:55.799 --> 00:01:58.500 market signal for these specific skills right 00:01:58.500 --> 00:02:00.760 now. The AI bubble bonus. I like that. It kind 00:02:00.760 --> 00:02:02.079 of feels accurate for the moment, doesn't it? 00:02:02.200 --> 00:02:03.480 It does. But why do you think it's that high? 00:02:03.900 --> 00:02:07.359 Is it purely demand? Or is it that finding people 00:02:07.359 --> 00:02:09.460 who can actually effectively use these tools 00:02:09.460 --> 00:02:11.840 is still pre -rare? I think it's a combination, 00:02:12.060 --> 00:02:13.919 honestly. It's demand for the capability, but 00:02:13.919 --> 00:02:16.520 it also signals that true fluency with these 00:02:16.520 --> 00:02:18.379 tools, you know, applying them effectively in 00:02:18.379 --> 00:02:21.740 a job context, that isn't ubiquitous yet. It's 00:02:21.740 --> 00:02:23.960 the application that's rare and therefore valuable. 00:02:24.240 --> 00:02:26.500 Okay, that makes sense. It's the doing, not just 00:02:26.500 --> 00:02:29.400 knowing the name. And they also found AI is making 00:02:29.400 --> 00:02:34.340 jobs evolve way faster. like 66 % faster in roles 00:02:34.340 --> 00:02:37.639 that are really exposed to AI. Yeah, 66 % faster 00:02:37.639 --> 00:02:40.379 change in skills required. That's rapid. This 00:02:40.379 --> 00:02:42.180 raises an important question about what jobs 00:02:42.180 --> 00:02:44.419 are in the modern economy. The Thors give the 00:02:44.419 --> 00:02:46.840 example comparing financial analysts and physical 00:02:46.840 --> 00:02:50.360 therapy. Financial analyst roles, which can heavily 00:02:50.360 --> 00:02:52.960 leverage AI for things like data analysis, forecasting, 00:02:53.340 --> 00:02:56.060 even report writing, their required skills are 00:02:56.060 --> 00:02:58.680 shifting much, much more rapidly than, say, a 00:02:58.680 --> 00:03:01.360 physical therapist role. That relies deeply on 00:03:01.360 --> 00:03:04.280 hands -on interaction, empathy, physical assessment, 00:03:04.479 --> 00:03:07.460 you know, things AI isn't doing yet. So it's 00:03:07.460 --> 00:03:10.599 not just if your job might change, but how quickly 00:03:10.599 --> 00:03:12.800 the actual requirements for it are shifting into 00:03:12.800 --> 00:03:14.879 your feet. That's kind of intense to think about. 00:03:15.000 --> 00:03:17.080 Definitely. And another thing from the reports 00:03:17.080 --> 00:03:20.259 that I thought was key, AI isn't just like a 00:03:20.259 --> 00:03:22.620 Silicon Valley or tech company thing anymore. 00:03:22.900 --> 00:03:25.180 Oh, yeah. If we connect this to the bigger picture 00:03:25.180 --> 00:03:27.219 they're painting, AI is definitely spreading. 00:03:27.259 --> 00:03:29.419 It's moving into industries you might not immediately 00:03:29.419 --> 00:03:32.080 think of. They mentioned agriculture, construction. 00:03:32.680 --> 00:03:35.990 It's getting embedded everywhere. Beyond just 00:03:35.990 --> 00:03:38.930 software development or the typical tech hubs. 00:03:39.189 --> 00:03:41.969 Like what, AI helping analyze soil data or something? 00:03:42.090 --> 00:03:45.370 Exactly. Or optimizing complex construction schedules. 00:03:46.009 --> 00:03:48.689 Practical stuff. Which, yeah, if it's doing those 00:03:48.689 --> 00:03:50.090 practical things, it makes sense it's spreading. 00:03:50.270 --> 00:03:52.590 But OK, here's the big one that everyone worries 00:03:52.590 --> 00:03:56.509 about, right? The headline fear. Does more AI 00:03:56.509 --> 00:03:59.599 automatically mean fewer jobs overall? Well, 00:03:59.680 --> 00:04:01.919 according to this report, often it's actually 00:04:01.919 --> 00:04:05.099 the opposite. The core finding is that AI isn't 00:04:05.099 --> 00:04:07.900 primarily leading to mass automation that simply 00:04:07.900 --> 00:04:11.139 eliminates roles. Instead, it's enabling people 00:04:11.139 --> 00:04:14.639 to do more valuable work. It augments human capabilities 00:04:14.639 --> 00:04:17.939 rather than simply replacing them entirely. That's 00:04:17.939 --> 00:04:20.540 the key finding here. Hmm. Okay, that's a pretty 00:04:20.540 --> 00:04:23.360 crucial insight, right? It shifts the focus from 00:04:23.360 --> 00:04:26.879 replacement to, like... enhancement. So instead 00:04:26.879 --> 00:04:29.120 of replacing a financial analyst, AI helps them 00:04:29.120 --> 00:04:32.100 analyze data way faster, freeing them up to do 00:04:32.100 --> 00:04:33.899 more strategic thinking or client work, that 00:04:33.899 --> 00:04:36.220 kind of thing. Exactly. It takes away some of 00:04:36.220 --> 00:04:38.819 the repetitive or data heavy tasks, allowing 00:04:38.819 --> 00:04:41.019 the human expertise to be applied to higher level 00:04:41.019 --> 00:04:43.899 problems or more creative challenges. And they 00:04:43.899 --> 00:04:46.689 link this to company performance, too. Companies 00:04:46.689 --> 00:04:48.970 using AI, well, they aren't just cutting costs. 00:04:49.069 --> 00:04:50.990 They're actually growing revenue and productivity. 00:04:51.269 --> 00:04:53.129 That's where these really striking numbers come 00:04:53.129 --> 00:04:55.850 in. Industries effectively integrating AI, the 00:04:55.850 --> 00:04:57.350 source in its software publishing as a prime 00:04:57.350 --> 00:05:00.470 example, are seeing significant gains. The data 00:05:00.470 --> 00:05:02.990 points to a 3x growth in revenue per employee. 00:05:03.209 --> 00:05:05.990 Few times revenue per employee. Yeah. And a remarkable 00:05:05.990 --> 00:05:09.069 4x jump in productivity since ChatGPT's public 00:05:09.069 --> 00:05:14.170 launch in late 2022. Wow. A 4x productivity jump 00:05:14.170 --> 00:05:17.629 in under two years. That's massive. What does 00:05:17.629 --> 00:05:19.769 that even look like on the ground in a company? 00:05:19.910 --> 00:05:22.189 Are people just getting way more done with the 00:05:22.189 --> 00:05:24.589 same amount of effort? Pretty much. Think about 00:05:24.589 --> 00:05:27.370 using AI to draft code, write documentation, 00:05:27.829 --> 00:05:30.889 analyze user feedback, automate testing. All 00:05:30.889 --> 00:05:33.449 tasks that previously took significant human 00:05:33.449 --> 00:05:36.850 time can now be done much faster or even automatically. 00:05:37.689 --> 00:05:40.290 This frees up the skilled engineers and developers 00:05:40.290 --> 00:05:43.129 to focus on building new features, solving complex 00:05:43.129 --> 00:05:46.459 bugs, or innovating. Circling back to that skills 00:05:46.459 --> 00:05:49.160 premium, that 56%, what's interesting is they 00:05:49.160 --> 00:05:51.279 say even in roles you might think would be easily 00:05:51.279 --> 00:05:53.220 automatable, like customer support or coding, 00:05:53.439 --> 00:05:55.899 those roles are still growing. It's the skills 00:05:55.899 --> 00:05:57.860 within them that are changing. Right. What's 00:05:57.860 --> 00:06:00.019 fascinating here is the emphasis isn't solely 00:06:00.019 --> 00:06:02.339 on the job title itself, but on the skills and 00:06:02.339 --> 00:06:05.220 proficiency with AI tools within that role. The 00:06:05.220 --> 00:06:07.519 report suggests that for these AI -heavy positions, 00:06:07.800 --> 00:06:09.779 what you can do with AI tools is becoming more 00:06:09.779 --> 00:06:12.680 important than formal qualifications alone. It's 00:06:12.680 --> 00:06:15.149 about practical AI fluency, that ability. to 00:06:15.149 --> 00:06:17.870 apply the tool effectively to get that productivity 00:06:17.870 --> 00:06:20.889 jump. Yeah, it's kind of like your ability to 00:06:20.889 --> 00:06:23.430 wield the tool matters more than just, you know, 00:06:23.449 --> 00:06:25.689 ticking a box on a traditional diploma. And they 00:06:25.689 --> 00:06:28.250 have that interesting stat about more women being 00:06:28.250 --> 00:06:31.129 in AI -affected roles globally, too. That's framed 00:06:31.129 --> 00:06:33.939 as a significant opportunity. Given the potential 00:06:33.939 --> 00:06:36.500 for higher wages and evolving roles with these 00:06:36.500 --> 00:06:39.579 AI fluency skills, this demographic distribution 00:06:39.579 --> 00:06:42.360 presents a chance to ensure diverse participation 00:06:42.360 --> 00:06:45.139 and benefit from the AI -driven future of work 00:06:45.139 --> 00:06:48.000 through targeted reskilling. It's a really interesting 00:06:48.000 --> 00:06:50.240 point. So if we pull this all together from the 00:06:50.240 --> 00:06:53.819 PwC report. The big takeaway isn't this simple 00:06:53.819 --> 00:06:56.579 story of mass unemployment driven by robots taking 00:06:56.579 --> 00:06:59.839 over everything. It's way more nuanced. Precisely. 00:06:59.839 --> 00:07:03.139 The why it matters section from the source summarizes 00:07:03.139 --> 00:07:06.459 it well. The primary impact isn't mass layoffs, 00:07:06.500 --> 00:07:09.560 but rather a pivot towards reskilling. significant 00:07:09.560 --> 00:07:11.800 opportunities for pay growth for those who adapt, 00:07:12.060 --> 00:07:14.420 and strategic upgrading of roles for individuals 00:07:14.420 --> 00:07:17.199 who become proficient in using AI tools effectively. 00:07:17.379 --> 00:07:19.399 Okay, so mostly good news for adapting folks. 00:07:19.639 --> 00:07:22.040 Mostly. However, it's important to note the caveat 00:07:22.040 --> 00:07:24.860 they add. A potential economic downturn could 00:07:24.860 --> 00:07:28.209 alter this picture. How so? Well, companies might 00:07:28.209 --> 00:07:31.689 leverage this increased productivity to maintain 00:07:31.689 --> 00:07:34.850 or grow output with smaller teams if things got 00:07:34.850 --> 00:07:38.750 tough economically. So it's not entirely without 00:07:38.750 --> 00:07:40.850 potential risk, depending on the broader economy. 00:07:41.069 --> 00:07:42.850 Right. There's that nuance. It's mostly good 00:07:42.850 --> 00:07:45.029 news for skill development right now, but economic 00:07:45.029 --> 00:07:47.370 conditions always play a part in how technology 00:07:47.370 --> 00:07:51.870 impacts jobs. Okay. So that was the big picture 00:07:51.870 --> 00:07:54.009 report looking at global trends and data. Let's 00:07:54.009 --> 00:07:55.550 switch gears and look at some of the specific 00:07:55.550 --> 00:07:57.930 examples and tools that are popping up today 00:07:57.930 --> 00:08:00.389 from some of these other sources. What does AI 00:08:00.389 --> 00:08:03.250 in action actually look like right now? This 00:08:03.250 --> 00:08:04.910 is where you see the rubber meeting the road. 00:08:05.029 --> 00:08:07.569 You know, we're seeing new applications launching 00:08:07.569 --> 00:08:10.149 or being discussed across various areas. Like 00:08:10.149 --> 00:08:13.089 in the messaging space, launching XChat, trying 00:08:13.089 --> 00:08:16.509 to muscle in on WhatsApp and Telegram territory. 00:08:16.769 --> 00:08:19.129 And they're using Bitcoin style encryption, which 00:08:19.129 --> 00:08:21.870 sounds... Intense. Yes, that's an interesting 00:08:21.870 --> 00:08:24.750 angle. Bitcoin style in this context likely means 00:08:24.750 --> 00:08:27.350 leveraging principles from blockchain or cryptography 00:08:27.350 --> 00:08:30.610 to ensure a very high level of privacy and security 00:08:30.610 --> 00:08:33.309 for messages. It highlights a growing focus on 00:08:33.309 --> 00:08:36.230 security features in communication tools. OK, 00:08:36.330 --> 00:08:38.700 meaning like really focused on privacy. Got it. 00:08:38.759 --> 00:08:41.139 Yeah. And we're seeing AI writing and content 00:08:41.139 --> 00:08:43.940 tools continue to evolve. Anthropic, you know, 00:08:43.940 --> 00:08:46.259 they're really showing off Claude's writing skills 00:08:46.259 --> 00:08:49.039 with that Claude Explains blog. Yeah, that's 00:08:49.039 --> 00:08:50.600 like a public demo, right? It's kind of like 00:08:50.600 --> 00:08:53.139 a quiet flex, right? Yeah. Look what our model 00:08:53.139 --> 00:08:55.580 can do in a public -facing way. It demonstrates 00:08:55.580 --> 00:08:57.700 the model's capacity for generating coherent, 00:08:58.019 --> 00:09:00.820 well -structured, and informative text across 00:09:00.820 --> 00:09:03.639 various topics. And we're seeing other tools 00:09:03.639 --> 00:09:06.220 emerge applying this, like PenDown, specifically 00:09:06.220 --> 00:09:09.500 for writing proposals quickly, or Tilda for creating 00:09:09.500 --> 00:09:11.919 entire websites easily, often without needing 00:09:11.919 --> 00:09:15.009 to code. So applying AI to specific practical 00:09:15.009 --> 00:09:17.490 business tasks to speed things up. There's also 00:09:17.490 --> 00:09:20.250 tools like Firecrawl for just grabbing data from 00:09:20.250 --> 00:09:22.970 entire websites easily for analysis. Web scraping 00:09:22.970 --> 00:09:25.169 made easy. Or these automation blueprints using 00:09:25.169 --> 00:09:28.350 platforms like N8n. These tools address workflow 00:09:28.350 --> 00:09:30.970 efficiency head on, whether it's streamlining 00:09:30.970 --> 00:09:33.009 the process of gathering information from the 00:09:33.009 --> 00:09:35.950 web or automating complex sequences of tasks 00:09:35.950 --> 00:09:38.710 that connect different applications. Very practical. 00:09:38.929 --> 00:09:42.059 And then you get the more creative. or futuristic 00:09:42.059 --> 00:09:44.679 stuff like Mirage Studio, creating these really 00:09:44.679 --> 00:09:48.419 lifelike AI actors. Ooh, yeah. That's kind of 00:09:48.419 --> 00:09:50.759 mind -bending to think about applying that. It 00:09:50.759 --> 00:09:53.220 shows the expansion of AI into creative production, 00:09:53.639 --> 00:09:56.059 potentially changing how content is made in film, 00:09:56.220 --> 00:09:59.179 gaming, or virtual reality. Beyond the tools, 00:09:59.320 --> 00:10:02.879 there's also just the industry buzz and sometimes 00:10:02.879 --> 00:10:05.649 drama that reflects what's happening. The reports 00:10:05.649 --> 00:10:07.970 about deep seek possibly training on other models 00:10:07.970 --> 00:10:10.929 data. Gemini Chat GPT outputs. Oh, right. That 00:10:10.929 --> 00:10:13.649 controversy. This definitely highlights the complex 00:10:13.649 --> 00:10:16.269 and sometimes contentious nature of model training 00:10:16.269 --> 00:10:18.490 data sources in the hyper -competitive landscape. 00:10:18.889 --> 00:10:21.789 There are ongoing questions globally about intellectual 00:10:21.789 --> 00:10:24.610 property, fair use, and data licensing in this 00:10:24.610 --> 00:10:27.750 rapidly developing field. It's maybe a bit messy 00:10:27.750 --> 00:10:30.669 behind the scenes. Yeah, it shows how the foundation 00:10:30.669 --> 00:10:33.210 of these models can be a whole area of contention. 00:10:33.350 --> 00:10:35.299 And then you have someone like... Yoshua Bengio, 00:10:35.460 --> 00:10:38.399 one of the real godfathers of modern AI, talking 00:10:38.399 --> 00:10:42.320 openly about models potentially lying and launching 00:10:42.320 --> 00:10:45.820 this initiative called Law Zero for Honest AI. 00:10:46.139 --> 00:10:48.659 This raises an important fundamental question 00:10:48.659 --> 00:10:51.799 about model reliability and the potential for 00:10:51.799 --> 00:10:54.519 models to generate plausible sounding but factually 00:10:54.519 --> 00:10:56.990 incorrect or misleading information. what we 00:10:56.990 --> 00:10:59.970 call hallucinations, but perhaps in more complex 00:10:59.970 --> 00:11:02.990 ways now. Initiatives like Law Zero reflect a 00:11:02.990 --> 00:11:05.110 growing concern within the AI research community 00:11:05.110 --> 00:11:08.649 about ensuring alignment, truthfulness, and trustworthiness 00:11:08.649 --> 00:11:11.250 in advanced AI systems, especially as they become 00:11:11.250 --> 00:11:13.490 more capable. It makes you think about the underlying 00:11:13.490 --> 00:11:15.929 principles, doesn't it? Can we even trust what 00:11:15.929 --> 00:11:18.230 these models tell us without verification? It's 00:11:18.230 --> 00:11:20.740 a huge question. And speaking of drama... There's 00:11:20.740 --> 00:11:22.519 already a movie being made about the whole open 00:11:22.519 --> 00:11:25.720 AI Sam Altman situation. Wow. That just speaks 00:11:25.720 --> 00:11:28.159 to how incredibly quickly things are moving in 00:11:28.159 --> 00:11:31.679 this space and how much attention and drama that 00:11:31.679 --> 00:11:34.539 particular episode generated externally. Yeah, 00:11:34.559 --> 00:11:36.899 seriously. It really captured public imagination, 00:11:37.159 --> 00:11:40.279 for better or worse. Totally. It felt like a 00:11:40.279 --> 00:11:43.419 tech thriller unfolding in real time. And money 00:11:43.419 --> 00:11:46.460 is still flowing. You know, we saw that 10 billion 00:11:46.460 --> 00:11:49.919 yuan fund launched in Guangdong just for AI and 00:11:49.919 --> 00:11:52.080 robotics. Big investment there. Or big investment 00:11:52.080 --> 00:11:54.580 still going to companies like Grammarly or Neuralink. 00:11:54.960 --> 00:11:57.399 Funding continues to be a key indicator of where 00:11:57.399 --> 00:12:00.200 investors see future growth and potential, supporting 00:12:00.200 --> 00:12:03.000 both foundational AI research like in robotics 00:12:03.000 --> 00:12:06.000 and specific applications like enhancing communication 00:12:06.000 --> 00:12:09.759 or developing brain -computer interfaces. The 00:12:09.759 --> 00:12:11.659 bets are still being placed, and they're large. 00:12:12.019 --> 00:12:13.960 And just some quick surprising hits from the 00:12:13.960 --> 00:12:15.899 sources that show the sheer breadth of where 00:12:15.899 --> 00:12:18.879 AI is touching things. AlphaFold3 predicting 00:12:18.879 --> 00:12:22.019 protein structures so accurately huge for science, 00:12:22.120 --> 00:12:24.740 obviously. That's a massive breakthrough in biological 00:12:24.740 --> 00:12:27.480 research, accelerating dread discovery, and our 00:12:27.480 --> 00:12:29.960 fundamental understanding of life sciences by 00:12:29.960 --> 00:12:32.679 predicting shapes crucial for function. Just 00:12:32.679 --> 00:12:34.559 incredible potential there. But then on the flip 00:12:34.559 --> 00:12:38.100 side, Google had to pause its Ask Photos feature 00:12:38.100 --> 00:12:41.179 after some issues. Okay, so not everything works 00:12:41.179 --> 00:12:43.840 perfectly right away, you know. Yeah. AI development 00:12:43.840 --> 00:12:45.940 is iterative, and sometimes things need pulling 00:12:45.940 --> 00:12:48.659 back. That definitely underscores the iterative 00:12:48.659 --> 00:12:51.980 nature of AI development and deployment. Features 00:12:51.980 --> 00:12:54.700 may look promising in testing, but require significant 00:12:54.700 --> 00:12:57.600 adjustments based on real -world usage and feedback. 00:12:57.960 --> 00:13:01.039 And sometimes they're just not ready. Yeah, it's 00:13:01.039 --> 00:13:02.759 not always smooth sailing, even for the biggest 00:13:02.759 --> 00:13:06.720 players. And this last one, this AI foot scanner. 00:13:06.960 --> 00:13:09.539 Oh, yeah, I saw that. That can alert patients 00:13:09.539 --> 00:13:12.399 about heart failure signs 13 days in advance. 00:13:12.580 --> 00:13:16.220 Wow. That's incredibly specific and, like, potentially 00:13:16.220 --> 00:13:18.740 life -saving. That's a powerful example of AI's 00:13:18.740 --> 00:13:21.080 potential in preventative health care, leveraging 00:13:21.080 --> 00:13:24.139 seemingly unrelated or subtle data points like 00:13:24.139 --> 00:13:26.779 changes detectable in a foot scan to identify 00:13:26.779 --> 00:13:29.019 early indicators of serious conditions like heart 00:13:29.019 --> 00:13:31.360 failure. It's an unexpected application with 00:13:31.360 --> 00:13:33.799 profound potential impact. Really makes you think. 00:13:33.940 --> 00:13:35.559 OK, that one really sticks with you, doesn't 00:13:35.559 --> 00:13:37.399 it? Yeah. Let's shift gears again. There's this 00:13:37.399 --> 00:13:39.840 section in the sources specifically about voice 00:13:39.840 --> 00:13:41.860 AI. It feels like voice AI was getting a lot 00:13:41.860 --> 00:13:43.600 of hype. Remember, like everything was going 00:13:43.600 --> 00:13:46.320 to be a voice bot. Oh, yeah. The big conversational 00:13:46.320 --> 00:13:49.360 AI push. But this source does a bit of a reality 00:13:49.360 --> 00:13:51.600 check on that. Yes. The source points out that 00:13:51.600 --> 00:13:55.600 2024 saw a lot of initial AI first launches that 00:13:55.600 --> 00:13:58.440 were voice centric, often featuring conversational 00:13:58.440 --> 00:14:02.490 agents. And many of them didn't quite land. way 00:14:02.490 --> 00:14:04.809 they expected? Like Klarn and Duolingo rolling 00:14:04.809 --> 00:14:06.710 back features they'd announced? It sounds like 00:14:06.710 --> 00:14:08.909 those polished demos didn't always translate 00:14:08.909 --> 00:14:11.230 into practical, useful, everyday experiences 00:14:11.230 --> 00:14:14.049 for people? Exactly. The demos often showcased 00:14:14.049 --> 00:14:16.710 the potential, but the actual user experience 00:14:16.710 --> 00:14:18.929 with those agent -led features when deployed 00:14:18.929 --> 00:14:22.029 at scale was often mixed. The source notes that 00:14:22.029 --> 00:14:24.769 style outpaced substance. In some cases, the 00:14:24.769 --> 00:14:27.090 capability was there, but it wasn't necessarily 00:14:27.090 --> 00:14:30.009 solving a real user problem in a smooth way or 00:14:30.009 --> 00:14:32.490 was still too error -prone. Over substance sometimes. 00:14:32.649 --> 00:14:35.029 Yeah. But the source also talks about what is 00:14:35.029 --> 00:14:37.570 actually working today in voice AI. And it sounds 00:14:37.570 --> 00:14:39.409 like it's not the flashy stuff. It's the quietly 00:14:39.409 --> 00:14:42.549 embedded tools. Right. Things like transcription, 00:14:42.870 --> 00:14:45.549 real time speech recognition, operating behind 00:14:45.549 --> 00:14:48.950 the scenes, analyzing conversations. These are 00:14:48.950 --> 00:14:51.149 the applications that are demonstrating measurable 00:14:51.149 --> 00:14:53.549 return on investment for businesses right now. 00:14:54.350 --> 00:14:56.289 They aren't necessarily the main interface you 00:14:56.289 --> 00:14:59.429 talk to, but they are enabling significant efficiency 00:14:59.429 --> 00:15:01.830 gains in the background. So they're working best 00:15:01.830 --> 00:15:04.549 when they're just helping out, like you said, 00:15:04.590 --> 00:15:06.450 not being the star of the show. Pretty much. 00:15:06.669 --> 00:15:09.149 Which feels a bit like that PwC reports point 00:15:09.149 --> 00:15:12.610 about augmentation. Exactly. And where is voice 00:15:12.610 --> 00:15:16.409 AI? really finding demand, these quiet tools. 00:15:16.710 --> 00:15:19.190 The source highlights environments that are inherently 00:15:19.190 --> 00:15:22.190 challenging for traditional tech or human capacity. 00:15:22.590 --> 00:15:25.110 Noisy settings like clinics where it's hard for 00:15:25.110 --> 00:15:27.850 a doctor to type notes accurately. High pressure 00:15:27.850 --> 00:15:30.169 environments like customer call centers where 00:15:30.169 --> 00:15:32.909 agents need quick information access or situations 00:15:32.909 --> 00:15:35.610 involving multiple languages or complex code 00:15:35.610 --> 00:15:37.669 switching. Code switching, like mixing languages. 00:15:37.990 --> 00:15:41.129 Yeah, where people blend languages mid -sentence. 00:15:41.230 --> 00:15:44.080 Voice AI solutions. that can handle these complexities 00:15:44.080 --> 00:15:46.240 are finding real traction because they solve 00:15:46.240 --> 00:15:48.779 specific difficult problems. That makes total 00:15:48.779 --> 00:15:51.379 sense. Places where a human might struggle to 00:15:51.379 --> 00:15:54.419 keep up or accurately record information. And 00:15:54.419 --> 00:15:56.759 the source lists some key trends for voice AI's 00:15:56.759 --> 00:15:59.059 success going forward. What stands out to you 00:15:59.059 --> 00:16:01.710 there? What's fascinating here is this reinforcement 00:16:01.710 --> 00:16:04.690 of the shift we just mentioned. It's less about 00:16:04.690 --> 00:16:07.529 creating fully autonomous conversational agents 00:16:07.529 --> 00:16:09.450 for the front end that try to replace a person, 00:16:09.570 --> 00:16:12.169 and much more about powering back -end operations 00:16:12.169 --> 00:16:15.070 and workflows using the voice data. Okay, so 00:16:15.070 --> 00:16:18.090 processing the conversation, summarizing it instantly, 00:16:18.370 --> 00:16:20.610 pulling out key data points or action items, 00:16:20.809 --> 00:16:23.370 stuff like that. Precisely. And this connects 00:16:23.370 --> 00:16:26.250 directly back to the PwC Reports augmentation 00:16:26.250 --> 00:16:29.649 theme. It's about being assistive rather than 00:16:29.649 --> 00:16:32.950 fully autonomous. The voice AI handles the administrative 00:16:32.950 --> 00:16:35.470 heavy lifting like generating a call summary, 00:16:35.710 --> 00:16:39.049 identifying compliance risks, or drafting a follow 00:16:39.049 --> 00:16:41.980 -up email. freeing up the human, whether it's 00:16:41.980 --> 00:16:44.539 a doctor or a call agent, to focus on the strategic 00:16:44.539 --> 00:16:47.759 aspects, the empathy, the complex problem solving 00:16:47.759 --> 00:16:50.460 that only a human can do. That fits perfectly. 00:16:50.700 --> 00:16:53.600 AI helping humans do the higher value work by 00:16:53.600 --> 00:16:56.039 taking care of the grunt work. And you need measurable 00:16:56.039 --> 00:16:59.100 results, right? Like clear ROI, especially for 00:16:59.100 --> 00:17:01.039 these backend tools. Absolutely critical. The 00:17:01.039 --> 00:17:03.559 source mentions examples like 50, 60 % time saved 00:17:03.559 --> 00:17:06.839 on transcription. 50 to 60%. Wow. Yeah. That's 00:17:06.839 --> 00:17:10.630 a clear quantifiable benefit for business. Demonstrable 00:17:10.630 --> 00:17:12.809 efficiency gains are key drivers for adoption 00:17:12.809 --> 00:17:14.930 in this space. You have to show it's actually 00:17:14.930 --> 00:17:16.930 saving time or money. Yeah, that's a massive 00:17:16.930 --> 00:17:19.029 time saver that adds up fast. And what about 00:17:19.029 --> 00:17:20.349 the language part? You mentioned code switching 00:17:20.349 --> 00:17:22.950 earlier. Multilingual capability is becoming 00:17:22.950 --> 00:17:25.750 essential, not optional in many environments. 00:17:25.849 --> 00:17:28.289 Systems need to seamlessly handle code switching. 00:17:28.470 --> 00:17:31.309 Where speakers naturally blend languages within 00:17:31.309 --> 00:17:34.049 a single conversation, as well as just operating 00:17:34.049 --> 00:17:36.369 accurately across multiple distinct languages. 00:17:37.000 --> 00:17:40.000 For global or diverse teams and customer bases, 00:17:40.359 --> 00:17:43.200 this is rapidly becoming a must -have. Okay, 00:17:43.240 --> 00:17:45.660 so it needs to be really flexible and robust 00:17:45.660 --> 00:17:49.480 with language and accuracy. That seems super 00:17:49.480 --> 00:17:51.240 important, especially if the stakes are high, 00:17:51.299 --> 00:17:53.380 like in a doctor's office or a financial setting. 00:17:53.559 --> 00:17:56.579 Oh, absolutely. This raises an important question 00:17:56.579 --> 00:18:00.000 about the threshold for accuracy. In highly regulated 00:18:00.000 --> 00:18:03.000 fields like compliance, health care and finance, 00:18:03.240 --> 00:18:06.480 even small hallucinations or errors by the AI 00:18:06.480 --> 00:18:09.279 in transcribing or summarizing could have severe 00:18:09.279 --> 00:18:12.539 consequences, including legal or financial repercussions, 00:18:12.619 --> 00:18:14.960 potentially even lawsuits. Yikes. The need for 00:18:14.960 --> 00:18:17.619 extremely high near perfect accuracy is paramount 00:18:17.619 --> 00:18:20.079 in these areas as the stakes are incredibly high. 00:18:20.400 --> 00:18:22.720 Yeah, you absolutely can't have AI messing up 00:18:22.720 --> 00:18:24.579 a critical medical record or a financial compliance 00:18:24.579 --> 00:18:27.180 note because of a transcription error. And the 00:18:27.180 --> 00:18:29.299 source finishes with this analogy about voice 00:18:29.299 --> 00:18:31.299 AI entering its cloud moment. What's that mean? 00:18:31.400 --> 00:18:34.619 This connects it to a past technological shift 00:18:34.619 --> 00:18:37.119 that profoundly changed computing infrastructure. 00:18:37.960 --> 00:18:41.000 Remember maybe 10, 15 years ago when the cloud 00:18:41.000 --> 00:18:43.779 seemed maybe a bit risky or just for niche applications? 00:18:44.279 --> 00:18:46.859 Yeah, people were kind of skeptical. Right. Now 00:18:46.859 --> 00:18:49.400 it underpins vast amounts of computing infrastructure. 00:18:49.480 --> 00:18:52.740 It runs everything, often quietly, reliably, 00:18:53.099 --> 00:18:55.720 and invisibly in the background. Uh -huh. It 00:18:55.720 --> 00:18:58.380 just became the default. Exactly. The source 00:18:58.380 --> 00:19:01.319 suggests voice AI is undergoing a similar transformation. 00:19:02.039 --> 00:19:05.200 It might not always be a flashy AI chatbot you 00:19:05.200 --> 00:19:07.019 talk to directly on the front end that grabs 00:19:07.019 --> 00:19:10.160 headlines. But its capabilities, accurate transcription, 00:19:10.579 --> 00:19:13.000 real -time recognition, intelligent summarization, 00:19:13.119 --> 00:19:15.799 voice analysis, will become deeply embedded and 00:19:15.799 --> 00:19:18.000 essential across various platforms and workflows. 00:19:18.359 --> 00:19:20.779 The shift, as the source says, won't be loud, 00:19:20.960 --> 00:19:23.599 but it will be everywhere, becoming the new infrastructure 00:19:23.599 --> 00:19:26.119 for how we process spoken information. That's 00:19:26.119 --> 00:19:27.539 a powerful way to put it. It's not going to be 00:19:27.539 --> 00:19:29.359 this big, splashy arrival. It's just going to 00:19:29.359 --> 00:19:32.019 quietly seep into everything until you can't 00:19:32.019 --> 00:19:34.259 imagine working without it. Kind of like Wi -Fi 00:19:34.259 --> 00:19:36.960 or the cloud. Exactly, that kind of quiet revolution. 00:19:37.630 --> 00:19:39.769 Okay, so we did a lot in this deep dive, right? 00:19:39.829 --> 00:19:42.009 We covered the big global job trends and the 00:19:42.009 --> 00:19:44.829 potential for serious pay bumps based on skills 00:19:44.829 --> 00:19:48.710 from that huge PwC report. That 56 % number still 00:19:48.710 --> 00:19:51.210 stands out. Totally. Looked at a rapid -fire 00:19:51.210 --> 00:19:54.309 round of specific new tools and some of the industry 00:19:54.309 --> 00:19:57.269 buzz and drama, and then dug into the quiet reality 00:19:57.269 --> 00:19:59.849 of voice AI and how it's actually finding its 00:19:59.849 --> 00:20:02.660 useful place. It definitely shows that the impact 00:20:02.660 --> 00:20:06.039 of AI is far from simple or monolithic. It's 00:20:06.039 --> 00:20:08.359 not just a linear story of technology replacing 00:20:08.359 --> 00:20:11.400 people wholesale. It's much more about jobs changing, 00:20:11.640 --> 00:20:14.500 new practical skills becoming incredibly valuable 00:20:14.500 --> 00:20:17.440 and commanding a premium, and companies finding 00:20:17.440 --> 00:20:20.779 ways to leverage AI not just to cut costs, but 00:20:20.779 --> 00:20:23.240 to become significantly more productive and find 00:20:23.240 --> 00:20:25.660 new avenues for growth and capability. And that 00:20:25.660 --> 00:20:28.420 PwC report really, really hammers home the importance 00:20:28.420 --> 00:20:31.130 of adapting, of learning. these new skills. You 00:20:31.130 --> 00:20:32.730 know, that's where the opportunity is clearly 00:20:32.730 --> 00:20:35.230 presenting itself right now. Reskilling and upgrading 00:20:35.230 --> 00:20:38.029 capabilities, focusing on that practical fluency 00:20:38.029 --> 00:20:40.990 with AI tools seems to be the key strategy identified 00:20:40.990 --> 00:20:43.109 by the data for individuals and organizations 00:20:43.109 --> 00:20:45.829 navigating this transformation effectively. So 00:20:45.829 --> 00:20:47.490 here's a final thought for you to mull over after 00:20:47.490 --> 00:20:50.390 this deep dive. Given that AI seems to be quietly 00:20:50.390 --> 00:20:53.470 embedding itself everywhere, from massive global 00:20:53.470 --> 00:20:56.549 industries and the way companies operate to specific 00:20:56.549 --> 00:20:59.400 tools you might use every day. And maybe even 00:20:59.400 --> 00:21:01.559 those surprising future healthcare applications 00:21:01.559 --> 00:21:04.599 like that foot scanner. And given that mastering 00:21:04.599 --> 00:21:06.900 these tools is showing a real tangible benefit 00:21:06.900 --> 00:21:09.920 like that potential 56 % pay bump we talked about, 00:21:10.019 --> 00:21:12.380 what part of your world, maybe something that 00:21:12.380 --> 00:21:15.359 seems totally unconnected to AI right now, might 00:21:15.359 --> 00:21:18.019 be the next to experience this quiet transformation? 00:21:18.859 --> 00:21:21.720 And what simple AI skill, maybe just exploring 00:21:21.720 --> 00:21:23.839 a new tool or learning how AI works in a specific 00:21:23.839 --> 00:21:26.619 context, could you explore that might give you 00:21:26.619 --> 00:21:29.059 that edge, that little boost as things continue 00:21:29.059 --> 00:21:31.539 to shift? Something to think about as you continue 00:21:31.539 --> 00:21:33.900 to navigate this evolving landscape, for sure. 00:21:34.279 --> 00:21:36.319 It's about being ready for that quiet shift. 00:21:36.579 --> 00:21:38.380 Definitely. Thanks for going on this deep dive 00:21:38.380 --> 00:21:38.740 with us.