July 26, 2025 • 12 mins
Is AI turbocharging our productivity or quietly dulling our thinking? This episode examines the science behind cognitive offloading, fresh business case studies, and smart strategies to ensure AI helps us get sharper, not just quicker at work.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:05):
Hey everyone.
I'm Andy and welcome to the AI breakdown.
Today we're diving into a fascinating topic.
Is AI making us smarter, or are we slowly outsourcing our thinking to machines? We've got some eye-opening stories and research to unpack, so stick around and let's get into it. 5 00:00:30,783.000983889 --> 00:00:34,983.000983889 Let's kick off with a story that made headlines last year, and not in the good kind of way. 6 00:00:35,853.000983889 --> 00:00:40,443.000983889 It's that infamous Air Canada chatbot disaster where a grieving customer. 7 00:00:40,788.000983889 --> 00:00:51,468.000983889 Promised bereavement Faires by the airlines Bot acted on the advice only to be denied a refund and told essentially the bot got it wrong, sorry, but the courts didn't buy that. 8 00:00:51,948.000983889 --> 00:00:59,748.000983889 Air Canada was held responsible, got clobbered in the media and learned the hard way that blaming the robot is not a get out of jail free card. 9 00:01:00,408.000983889 --> 00:01:03,78.000983889 Now I've got no problem with tech making mistakes. 10 00:01:03,318.000983889 --> 00:01:07,668.000983889 Some level of quality issues is expected, especially in this new AI era. 11 00:01:08,388.000983889 --> 00:01:11,658.000983889 What I do take issue with is when leadership Dodges accountability. 12 00:01:12,168.000983889 --> 00:01:24,528.000983889 In situations like this, they should step up, do the right thing, and then treat it as a learning moment, which begs the question when things go wrong, who's actually responsible, the algorithm or the humans behind it. 13 00:01:25,158.000983889 --> 00:01:31,578.000983889 Now this isn't just a one-off in workplaces everywhere, AI is quietly revolutionizing how we work. 14 00:01:31,908.000983889 --> 00:01:35,718.000983889 It drafts our emails, decides what matters in monstrous reports. 15 00:01:36,3.000983889 --> 00:01:42,483.00098389 And even recommends who gets hired or fired, the pace is breathtaking and the efficiency gains are impressive. 16 00:01:42,933.00098389 --> 00:01:54,243.00098389 But are we on the path to new heights or are we bit by bit outsourcing? So much of our thinking that our own critical faculties are quietly eroding is AI making us dumber. 17 00:01:54,903.00098389 --> 00:01:57,153.00098389 The science is beginning to suggest it could be. 18 00:01:57,573.00098389 --> 00:02:03,693.00098389 The Landmark 2025 UK Study by Gerlich and team found a strong negative correlation. 19 00:02:04,38.00098389 --> 00:02:09,858.00098389 Between frequent AI tool use and critical thinking ability, especially among younger professionals. 20 00:02:10,308.00098389 --> 00:02:23,148.00098389 In other words, the more often you rely on ai, the worse you do on tasks that demand independent judgment, complex reasoning, or actually questioning information instead of accepting it at face value. 21 00:02:23,568.00098389 --> 00:02:27,858.00098389 The study's key finding revolves around something called cognitive offloading. 22 00:02:28,278.00098389 --> 00:02:33,588.00098389 That's when we hand off mental effort to a machine instead of flexing the mental muscles ourselves. 23 00:02:34,228.00098389 --> 00:02:35,518.00098389 It's not entirely new. 24 00:02:35,728.00098389 --> 00:02:36,298.00098389 Hands up. 25 00:02:36,298.00098389 --> 00:02:41,600.65404511 Who remembers when calculators were classed as cheating? But today's AI isn't just solving sums. 26 00:02:42,20.65404511 --> 00:02:49,490.65404511 It's suggesting ideas, making recommendations, even weighing in on decisions that used to be the sole domain of humans. 27 00:02:49,940.65404511 --> 00:02:54,175.65404511 And the temptation to just go with the flow grows every time the tech gets better. 28 00:02:55,10.65404511 --> 00:02:57,590.65404511 It's a concern I hear frequently from our customers. 29 00:02:58,160.65404511 --> 00:03:05,630.65404511 What does this actually look like in the workplace? The researchers interviewed workers who openly admitted letting the AI do the thinking. 30 00:03:06,170.65404511 --> 00:03:10,700.65404511 Some acknowledged they check the output less and fear their own skills are getting rusty. 31 00:03:11,360.65404511 --> 00:03:16,940.65404511 Critically, younger digital natives, those who've grown up steeped in tech are more vulnerable. 32 00:03:17,270.65404511 --> 00:03:23,750.65404511 While more experienced or better trained, people tend to spot problems, ask more questions, and keep themselves sharp. 33 00:03:24,260.65404511 --> 00:03:25,940.65404511 The effect isn't a one-off either. 34 00:03:26,330.65404511 --> 00:03:29,90.65404511 Wharton and Microsoft researchers found what they call. 35 00:03:29,420.65404511 --> 00:03:31,70.65404511 Automation complacency. 36 00:03:31,700.65404511 --> 00:03:38,420.65404511 The more we accept AI output without double checking, the less likely we are to spot errors case after case. 37 00:03:38,420.65404511 --> 00:03:42,200.65404511 When humans get used to trusting the machine, our accountability fades. 38 00:03:42,530.65404511 --> 00:03:57,920.65404511 Have you ever just gone with an AI generated answer because it was quick and easy? Can you remember the last time you double check the recommendations or are you just hitting, accept and moving on? But with all that said, when we use AI deliberately with active oversight. 39 00:03:58,340.65404511 --> 00:04:00,200.65404511 It can turbocharge what you get done. 40 00:04:00,710.65404511 --> 00:04:02,90.65404511 The evidence is undeniable. 41 00:04:02,510.65404511 --> 00:04:06,500.65404511 Strategic AI implementation delivers significant time savings. 42 00:04:06,980.65404511 --> 00:04:16,670.65404511 Research from Salesforce, slack, and BCG and MIT shows workers spend less time on administrative tasks and more on meaningful work with consultants. 43 00:04:16,670.65404511 --> 00:04:20,510.65404511 Seeing a 40% productivity boost on well-defined projects. 44 00:04:21,50.65404511 --> 00:04:23,755.65404511 A trial over here in the UK with 20,000 civil servants. 45 00:04:24,395.65404511 --> 00:04:31,205.65404511 Using Microsoft 365 copilot found AI tools saved 26 minutes daily on routine tasks. 46 00:04:31,595.65404511 --> 00:04:45,270.65404511 While a global workforce survey by a deco group across 27 countries revealed employees using AI saved an average hour per day with tech, finance, energy, and manufacturing sectors seeing even greater benefits. 47 00:04:46,115.65404511 --> 00:04:48,305.65404511 The magic comes when we use that time differently. 48 00:04:48,605.65404511 --> 00:04:53,615.65404511 Instead of burning hours on basic tasks, AI gives us headroom to dig into analysis. 49 00:04:53,960.65404511 --> 00:04:56,630.65404511 Chase creative solutions and build relationships. 50 00:04:57,50.65404511 --> 00:05:02,360.65404511 The things only humans can do, and it's not just about giving bosses more time to strategize. 51 00:05:02,810.65404511 --> 00:05:12,500.65404511 OECD reviews shows some juniors are climbing the career ladder faster because the dull work gets automated, freeing up opportunities for skills that actually matter. 52 00:05:12,980.65404511 --> 00:05:20,780.65404511 The sweet spot human AI teamwork, a center model where humans do the critical work and AI handles the drudge. 53 00:05:21,170.65404511 --> 00:05:22,310.65404511 The key is to ask yourself. 54 00:05:22,910.65404511 --> 00:05:30,110.65404511 Are you using those saved hours for deep work or just letting the treadmill run faster? Turns out it's not always that simple. 55 00:05:30,530.65404511 --> 00:05:34,850.65404511 Recent research into AI assisted coding paints a more complicated picture. 56 00:05:35,420.65404511 --> 00:05:45,440.65404511 A study from MIT found that while developers expected productivity boosts from ai, they often hit unexpected slowdowns waiting for the AI to generate suggestions. 57 00:05:46,100.65404511 --> 00:05:48,620.65404511 During those lulls, many drifted into distractions. 58 00:05:49,265.65404511 --> 00:05:53,735.65404511 Checking social media, browsing unrelated sites and breaking their flow. 59 00:05:54,395.65404511 --> 00:05:59,855.65404511 What was meant to save time ended up fragmenting their focus, making it harder to get back into deep work. 60 00:06:00,275.65404511 --> 00:06:03,35.65404511 The issue here isn't ai, it's us. 61 00:06:03,695.65404511 --> 00:06:09,575.65404511 AI can be a brilliant wingman for deeper thinking, but only if we're using it with purpose, not just on autopilot. 62 00:06:09,575.65404511 --> 00:06:12,5.65404511 The time it frees up is a gift. 63 00:06:12,305.65404511 --> 00:06:13,115.65404511 Don't waste it. 64 00:06:13,205.65404511 --> 00:06:15,395.65404511 Doom scrolling or chasing distractions. 65 00:06:17,174.01776101 --> 00:06:21,944.01776101 let's run through a few stories of what happens when we let AI make too many calls unchecked. 66 00:06:22,394.01776101 --> 00:06:26,954.01776101 We've already talked about the Air Canada chatbot fiasco, but it's not just airlines. 67 00:06:27,284.01776101 --> 00:06:38,264.01776101 One of the most eye-opening examples happened at The Guardian where a freelance journalist submitted recent book recommendations that included several completely fictional books and authors. 68 00:06:39,134.01776101 --> 00:06:43,664.01776101 They were generated by chat GPT and published without verification. 69 00:06:44,264.01776101 --> 00:06:48,704.01776101 The guardian had to quietly edit the piece after readers mock the error on social media. 70 00:06:49,184.01776101 --> 00:06:51,164.01776101 This wasn't an isolated incident either. 71 00:06:51,584.01776101 --> 00:07:00,794.01776101 Similar cases have popped up where writers used AI to recommend non-existent books, and editors failed to catch these phantom publications before they went to press. 72 00:07:01,364.01776101 --> 00:07:04,4.01776101 Now let's look at startups bench accounting. 73 00:07:04,4.01776101 --> 00:07:08,744.01776101 Once a rising star in FinTech went all in on automating bookkeeping with the eye. 74 00:07:09,194.01776101 --> 00:07:11,924.01776101 When the system hit problems, real humans didn't catch. 75 00:07:12,224.01776101 --> 00:07:14,654.01776101 The whole business imploded practically overnight. 76 00:07:15,194.01776101 --> 00:07:28,274.01776101 Zillow, an online real estate marketplace, lost half a billion mispricing homes using automated models, and even HR tech like Workday has landed in court for algorithmic bias and adding insult to injury. 77 00:07:28,544.01776101 --> 00:07:33,44.01776101 New York City government ruled out an AI powered chat bot to help business owners. 78 00:07:33,404.01776101 --> 00:07:36,704.01776101 It ended up dispensing illegal and sometimes dangerous advice. 79 00:07:37,604.01776101 --> 00:07:44,984.01776101 The lack of human crosscheck meant genuine risk to people and reputations the theme gaps in accountability. 80 00:07:45,314.01776101 --> 00:08:03,374.01776101 Blind faith in AI as the all knowing oracle and failures that could have been avoided if someone had simply paused to ask, wait, does this actually make sense? So we know relying on AI too much can go wrong, but swinging too far the other way means missing out on significant productivity gains. 81 00:08:04,424.01776101 --> 00:08:09,944.01776101 The key here is balance, and both research and industry frameworks point to the same conclusion. 82 00:08:10,274.01776101 --> 00:08:11,594.01776101 We need human oversight. 83 00:08:12,194.01776101 --> 00:08:20,174.01776101 What this means is that for important or risky decisions, humans must carefully validate what the machine suggests in technical terms. 84 00:08:20,174.01776101 --> 00:08:23,114.01776101 This is called HITL, human in the loop. 85 00:08:23,714.01776101 --> 00:08:31,964.01776101 It's essentially the digital equivalent of trust, but verify a principle that keeps us engaged while still leveraging AI's capabilities. 86 00:08:32,669.01776101 --> 00:08:43,229.01776101 So how do you put this into practice? Here's a quick leadership playbook to make AI work for you without losing what makes you smart? Start with regular AI decision audits. 87 00:08:43,409.01776101 --> 00:08:44,914.01776101 Don't just check what the AI got. 88 00:08:44,914.01776101 --> 00:08:45,74.01776101 Right. 89 00:08:45,734.01776101 --> 00:08:49,244.01776101 Dig into where it went wrong and why fact check the facts. 90 00:08:49,544.01776101 --> 00:08:57,614.01776101 I use tools like perplexity to validate AI generated stats, research and references, especially when the stakes are high. 91 00:08:58,4.01776101 --> 00:08:59,564.01776101 Create clear risk tiers. 92 00:08:59,834.01776101 --> 00:09:04,484.01776101 Full automation might be fine for low stake stuff like email sorting or calendar management. 93 00:09:04,814.01776101 --> 00:09:14,984.01776101 But for anything high stakes like customer service, finance, or HR decisions, keep a human in the loop with clear sign off protocols, stress test before rollout. 94 00:09:15,344.01776101 --> 00:09:18,524.01776101 Don't just assume the shiny new tool is fit for purpose. 95 00:09:18,884.01776101 --> 00:09:20,749.01776101 Put it through real world scenarios first. 96 00:09:21,449.01776101 --> 00:09:23,69.01776101 Invest in digital literacy. 97 00:09:23,429.01776101 --> 00:09:28,709.01776101 Your people need to understand what AI can and can't do, and feel confident challenging it. 98 00:09:29,429.01776101 --> 00:09:35,579.01776101 Microsoft research shows that people who trust their own judgment are far more likely to double check AI outputs. 99 00:09:36,119.01776101 --> 00:09:45,539.01776101 Those who don't, they just nod along, clarify ownership, be crystal clear on policies and contracts about who owns the outcome when AI is involved. 100 00:09:46,19.01776101 --> 00:09:47,819.01776101 And finally, track your reliance. 101 00:09:48,209.01776101 --> 00:09:49,814.01776101 Use metrics like time to decision. 102 00:09:50,579.01776101 --> 00:09:55,169.01776101 Oversight dashboards to make sure human judgment isn't quietly falling outta the loop. 103 00:09:55,859.01776101 --> 00:10:18,509.01776101 So where does this leave us? AI in the workplace is a double-edged sword, treated as an infallible oracle, and you risk dulling your critical thinking, possibly even finding yourself at the center of a scandal, but leverage its capabilities strategically maintain your discernment, and it could deliver the most significant productivity enhancement of your professional life. 104 00:10:19,664.01776101 --> 00:10:21,614.01776101 The real answer isn't up to the technology. 105 00:10:21,794.01776101 --> 00:10:25,124.01776101 It's up to us how we use question and control it. 106 00:10:25,604.01776101 --> 00:10:30,554.01776101 We need to design workplaces where AI amplifies what's best about human intelligence. 107 00:10:31,34.01776101 --> 00:10:35,474.01776101 Don't outsource your judgment, your skepticism, or your appetite for asking why. 108 00:10:35,894.01776101 --> 00:10:42,404.01776101 And before we write AI off as the world's worst brain drain device, let's look at what happens when we use it with purpose. 109 00:10:42,824.01776101 --> 00:10:48,314.01776101 There's research out there, OECD, McKinsey and others arguing that when used intentionally. 110 00:10:49,139.01776101 --> 00:10:52,319.01776101 AI is a genuine cognitive force multiplier. 111 00:10:52,799.01776101 --> 00:11:02,969.01776101 When the machine crunches the numbers and drafts the first version, your mind's freed up for the real stuff, strategy, deep learning, bold ideas, or just better questions. 112 00:11:03,629.01776101 --> 00:11:06,179.01776101 What we're seeing is a shift in what it means to be smart. 113 00:11:06,569.01776101 --> 00:11:15,389.01776101 It's less about remembering facts or setting up massive spreadsheets from scratch, and more about judgment, curating data, interpreting signals, applying ethics. 114 00:11:16,394.01776101 --> 00:11:17,114.01776101 The real win. 115 00:11:17,534.01776101 --> 00:11:19,724.01776101 You don't have to pick between speed and smarts. 116 00:11:19,814.01776101 --> 00:11:24,44.01776101 You get both if you stay in the driver's seat and never hand over the big calls. 117 00:11:24,794.01776101 --> 00:11:32,324.01776101 As AI keeps getting smarter, let's make sure we're designing the future on our terms using tech as a force for empowerment. 118 00:11:32,564.01776101 --> 00:11:34,94.01776101 Not an excuse to switch off. 119 00:11:34,934.01776101 --> 00:11:45,794.01776101 If today's show struck a chord, tell me, have you ever caught yourself blindly trusting an AI tool, or is this one of your big concerns for your team? Ping me at hello@theaibreakdown.com 120 00:11:45,794.01776101 --> 00:11:48,914.01776101 with your story and I'll feature the best on a future episode. 121 00:11:49,334.01776101 --> 00:11:51,284.01776101 And while you're here, hit subscribe. 122 00:11:51,344.01776101 --> 00:11:55,964.01776101 Give us a rating or share this episode with a colleague hooked on automating everything.
Hey everyone.
I'm Andy and welcome to the AI breakdown.
Today we're diving into a fascinating topic.
Is AI making us smarter, or are we slowly outsourcing our thinking to machines? We've got some eye-opening stories and research to unpack, so stick around and let's get into it. 5 00:00:30,783.000983889 --> 00:00:34,983.000983889 Let's kick off with a story that made headlines last year, and not in the good kind of way. 6 00:00:35,853.000983889 --> 00:00:40,443.000983889 It's that infamous Air Canada chatbot disaster where a grieving customer. 7 00:00:40,788.000983889 --> 00:00:51,468.000983889 Promised bereavement Faires by the airlines Bot acted on the advice only to be denied a refund and told essentially the bot got it wrong, sorry, but the courts didn't buy that. 8 00:00:51,948.000983889 --> 00:00:59,748.000983889 Air Canada was held responsible, got clobbered in the media and learned the hard way that blaming the robot is not a get out of jail free card. 9 00:01:00,408.000983889 --> 00:01:03,78.000983889 Now I've got no problem with tech making mistakes. 10 00:01:03,318.000983889 --> 00:01:07,668.000983889 Some level of quality issues is expected, especially in this new AI era. 11 00:01:08,388.000983889 --> 00:01:11,658.000983889 What I do take issue with is when leadership Dodges accountability. 12 00:01:12,168.000983889 --> 00:01:24,528.000983889 In situations like this, they should step up, do the right thing, and then treat it as a learning moment, which begs the question when things go wrong, who's actually responsible, the algorithm or the humans behind it. 13 00:01:25,158.000983889 --> 00:01:31,578.000983889 Now this isn't just a one-off in workplaces everywhere, AI is quietly revolutionizing how we work. 14 00:01:31,908.000983889 --> 00:01:35,718.000983889 It drafts our emails, decides what matters in monstrous reports. 15 00:01:36,3.000983889 --> 00:01:42,483.00098389 And even recommends who gets hired or fired, the pace is breathtaking and the efficiency gains are impressive. 16 00:01:42,933.00098389 --> 00:01:54,243.00098389 But are we on the path to new heights or are we bit by bit outsourcing? So much of our thinking that our own critical faculties are quietly eroding is AI making us dumber. 17 00:01:54,903.00098389 --> 00:01:57,153.00098389 The science is beginning to suggest it could be. 18 00:01:57,573.00098389 --> 00:02:03,693.00098389 The Landmark 2025 UK Study by Gerlich and team found a strong negative correlation. 19 00:02:04,38.00098389 --> 00:02:09,858.00098389 Between frequent AI tool use and critical thinking ability, especially among younger professionals. 20 00:02:10,308.00098389 --> 00:02:23,148.00098389 In other words, the more often you rely on ai, the worse you do on tasks that demand independent judgment, complex reasoning, or actually questioning information instead of accepting it at face value. 21 00:02:23,568.00098389 --> 00:02:27,858.00098389 The study's key finding revolves around something called cognitive offloading. 22 00:02:28,278.00098389 --> 00:02:33,588.00098389 That's when we hand off mental effort to a machine instead of flexing the mental muscles ourselves. 23 00:02:34,228.00098389 --> 00:02:35,518.00098389 It's not entirely new. 24 00:02:35,728.00098389 --> 00:02:36,298.00098389 Hands up. 25 00:02:36,298.00098389 --> 00:02:41,600.65404511 Who remembers when calculators were classed as cheating? But today's AI isn't just solving sums. 26 00:02:42,20.65404511 --> 00:02:49,490.65404511 It's suggesting ideas, making recommendations, even weighing in on decisions that used to be the sole domain of humans. 27 00:02:49,940.65404511 --> 00:02:54,175.65404511 And the temptation to just go with the flow grows every time the tech gets better. 28 00:02:55,10.65404511 --> 00:02:57,590.65404511 It's a concern I hear frequently from our customers. 29 00:02:58,160.65404511 --> 00:03:05,630.65404511 What does this actually look like in the workplace? The researchers interviewed workers who openly admitted letting the AI do the thinking. 30 00:03:06,170.65404511 --> 00:03:10,700.65404511 Some acknowledged they check the output less and fear their own skills are getting rusty. 31 00:03:11,360.65404511 --> 00:03:16,940.65404511 Critically, younger digital natives, those who've grown up steeped in tech are more vulnerable. 32 00:03:17,270.65404511 --> 00:03:23,750.65404511 While more experienced or better trained, people tend to spot problems, ask more questions, and keep themselves sharp. 33 00:03:24,260.65404511 --> 00:03:25,940.65404511 The effect isn't a one-off either. 34 00:03:26,330.65404511 --> 00:03:29,90.65404511 Wharton and Microsoft researchers found what they call. 35 00:03:29,420.65404511 --> 00:03:31,70.65404511 Automation complacency. 36 00:03:31,700.65404511 --> 00:03:38,420.65404511 The more we accept AI output without double checking, the less likely we are to spot errors case after case. 37 00:03:38,420.65404511 --> 00:03:42,200.65404511 When humans get used to trusting the machine, our accountability fades. 38 00:03:42,530.65404511 --> 00:03:57,920.65404511 Have you ever just gone with an AI generated answer because it was quick and easy? Can you remember the last time you double check the recommendations or are you just hitting, accept and moving on? But with all that said, when we use AI deliberately with active oversight. 39 00:03:58,340.65404511 --> 00:04:00,200.65404511 It can turbocharge what you get done. 40 00:04:00,710.65404511 --> 00:04:02,90.65404511 The evidence is undeniable. 41 00:04:02,510.65404511 --> 00:04:06,500.65404511 Strategic AI implementation delivers significant time savings. 42 00:04:06,980.65404511 --> 00:04:16,670.65404511 Research from Salesforce, slack, and BCG and MIT shows workers spend less time on administrative tasks and more on meaningful work with consultants. 43 00:04:16,670.65404511 --> 00:04:20,510.65404511 Seeing a 40% productivity boost on well-defined projects. 44 00:04:21,50.65404511 --> 00:04:23,755.65404511 A trial over here in the UK with 20,000 civil servants. 45 00:04:24,395.65404511 --> 00:04:31,205.65404511 Using Microsoft 365 copilot found AI tools saved 26 minutes daily on routine tasks. 46 00:04:31,595.65404511 --> 00:04:45,270.65404511 While a global workforce survey by a deco group across 27 countries revealed employees using AI saved an average hour per day with tech, finance, energy, and manufacturing sectors seeing even greater benefits. 47 00:04:46,115.65404511 --> 00:04:48,305.65404511 The magic comes when we use that time differently. 48 00:04:48,605.65404511 --> 00:04:53,615.65404511 Instead of burning hours on basic tasks, AI gives us headroom to dig into analysis. 49 00:04:53,960.65404511 --> 00:04:56,630.65404511 Chase creative solutions and build relationships. 50 00:04:57,50.65404511 --> 00:05:02,360.65404511 The things only humans can do, and it's not just about giving bosses more time to strategize. 51 00:05:02,810.65404511 --> 00:05:12,500.65404511 OECD reviews shows some juniors are climbing the career ladder faster because the dull work gets automated, freeing up opportunities for skills that actually matter. 52 00:05:12,980.65404511 --> 00:05:20,780.65404511 The sweet spot human AI teamwork, a center model where humans do the critical work and AI handles the drudge. 53 00:05:21,170.65404511 --> 00:05:22,310.65404511 The key is to ask yourself. 54 00:05:22,910.65404511 --> 00:05:30,110.65404511 Are you using those saved hours for deep work or just letting the treadmill run faster? Turns out it's not always that simple. 55 00:05:30,530.65404511 --> 00:05:34,850.65404511 Recent research into AI assisted coding paints a more complicated picture. 56 00:05:35,420.65404511 --> 00:05:45,440.65404511 A study from MIT found that while developers expected productivity boosts from ai, they often hit unexpected slowdowns waiting for the AI to generate suggestions. 57 00:05:46,100.65404511 --> 00:05:48,620.65404511 During those lulls, many drifted into distractions. 58 00:05:49,265.65404511 --> 00:05:53,735.65404511 Checking social media, browsing unrelated sites and breaking their flow. 59 00:05:54,395.65404511 --> 00:05:59,855.65404511 What was meant to save time ended up fragmenting their focus, making it harder to get back into deep work. 60 00:06:00,275.65404511 --> 00:06:03,35.65404511 The issue here isn't ai, it's us. 61 00:06:03,695.65404511 --> 00:06:09,575.65404511 AI can be a brilliant wingman for deeper thinking, but only if we're using it with purpose, not just on autopilot. 62 00:06:09,575.65404511 --> 00:06:12,5.65404511 The time it frees up is a gift. 63 00:06:12,305.65404511 --> 00:06:13,115.65404511 Don't waste it. 64 00:06:13,205.65404511 --> 00:06:15,395.65404511 Doom scrolling or chasing distractions. 65 00:06:17,174.01776101 --> 00:06:21,944.01776101 let's run through a few stories of what happens when we let AI make too many calls unchecked. 66 00:06:22,394.01776101 --> 00:06:26,954.01776101 We've already talked about the Air Canada chatbot fiasco, but it's not just airlines. 67 00:06:27,284.01776101 --> 00:06:38,264.01776101 One of the most eye-opening examples happened at The Guardian where a freelance journalist submitted recent book recommendations that included several completely fictional books and authors. 68 00:06:39,134.01776101 --> 00:06:43,664.01776101 They were generated by chat GPT and published without verification. 69 00:06:44,264.01776101 --> 00:06:48,704.01776101 The guardian had to quietly edit the piece after readers mock the error on social media. 70 00:06:49,184.01776101 --> 00:06:51,164.01776101 This wasn't an isolated incident either. 71 00:06:51,584.01776101 --> 00:07:00,794.01776101 Similar cases have popped up where writers used AI to recommend non-existent books, and editors failed to catch these phantom publications before they went to press. 72 00:07:01,364.01776101 --> 00:07:04,4.01776101 Now let's look at startups bench accounting. 73 00:07:04,4.01776101 --> 00:07:08,744.01776101 Once a rising star in FinTech went all in on automating bookkeeping with the eye. 74 00:07:09,194.01776101 --> 00:07:11,924.01776101 When the system hit problems, real humans didn't catch. 75 00:07:12,224.01776101 --> 00:07:14,654.01776101 The whole business imploded practically overnight. 76 00:07:15,194.01776101 --> 00:07:28,274.01776101 Zillow, an online real estate marketplace, lost half a billion mispricing homes using automated models, and even HR tech like Workday has landed in court for algorithmic bias and adding insult to injury. 77 00:07:28,544.01776101 --> 00:07:33,44.01776101 New York City government ruled out an AI powered chat bot to help business owners. 78 00:07:33,404.01776101 --> 00:07:36,704.01776101 It ended up dispensing illegal and sometimes dangerous advice. 79 00:07:37,604.01776101 --> 00:07:44,984.01776101 The lack of human crosscheck meant genuine risk to people and reputations the theme gaps in accountability. 80 00:07:45,314.01776101 --> 00:08:03,374.01776101 Blind faith in AI as the all knowing oracle and failures that could have been avoided if someone had simply paused to ask, wait, does this actually make sense? So we know relying on AI too much can go wrong, but swinging too far the other way means missing out on significant productivity gains. 81 00:08:04,424.01776101 --> 00:08:09,944.01776101 The key here is balance, and both research and industry frameworks point to the same conclusion. 82 00:08:10,274.01776101 --> 00:08:11,594.01776101 We need human oversight. 83 00:08:12,194.01776101 --> 00:08:20,174.01776101 What this means is that for important or risky decisions, humans must carefully validate what the machine suggests in technical terms. 84 00:08:20,174.01776101 --> 00:08:23,114.01776101 This is called HITL, human in the loop. 85 00:08:23,714.01776101 --> 00:08:31,964.01776101 It's essentially the digital equivalent of trust, but verify a principle that keeps us engaged while still leveraging AI's capabilities. 86 00:08:32,669.01776101 --> 00:08:43,229.01776101 So how do you put this into practice? Here's a quick leadership playbook to make AI work for you without losing what makes you smart? Start with regular AI decision audits. 87 00:08:43,409.01776101 --> 00:08:44,914.01776101 Don't just check what the AI got. 88 00:08:44,914.01776101 --> 00:08:45,74.01776101 Right. 89 00:08:45,734.01776101 --> 00:08:49,244.01776101 Dig into where it went wrong and why fact check the facts. 90 00:08:49,544.01776101 --> 00:08:57,614.01776101 I use tools like perplexity to validate AI generated stats, research and references, especially when the stakes are high. 91 00:08:58,4.01776101 --> 00:08:59,564.01776101 Create clear risk tiers. 92 00:08:59,834.01776101 --> 00:09:04,484.01776101 Full automation might be fine for low stake stuff like email sorting or calendar management. 93 00:09:04,814.01776101 --> 00:09:14,984.01776101 But for anything high stakes like customer service, finance, or HR decisions, keep a human in the loop with clear sign off protocols, stress test before rollout. 94 00:09:15,344.01776101 --> 00:09:18,524.01776101 Don't just assume the shiny new tool is fit for purpose. 95 00:09:18,884.01776101 --> 00:09:20,749.01776101 Put it through real world scenarios first. 96 00:09:21,449.01776101 --> 00:09:23,69.01776101 Invest in digital literacy. 97 00:09:23,429.01776101 --> 00:09:28,709.01776101 Your people need to understand what AI can and can't do, and feel confident challenging it. 98 00:09:29,429.01776101 --> 00:09:35,579.01776101 Microsoft research shows that people who trust their own judgment are far more likely to double check AI outputs. 99 00:09:36,119.01776101 --> 00:09:45,539.01776101 Those who don't, they just nod along, clarify ownership, be crystal clear on policies and contracts about who owns the outcome when AI is involved. 100 00:09:46,19.01776101 --> 00:09:47,819.01776101 And finally, track your reliance. 101 00:09:48,209.01776101 --> 00:09:49,814.01776101 Use metrics like time to decision. 102 00:09:50,579.01776101 --> 00:09:55,169.01776101 Oversight dashboards to make sure human judgment isn't quietly falling outta the loop. 103 00:09:55,859.01776101 --> 00:10:18,509.01776101 So where does this leave us? AI in the workplace is a double-edged sword, treated as an infallible oracle, and you risk dulling your critical thinking, possibly even finding yourself at the center of a scandal, but leverage its capabilities strategically maintain your discernment, and it could deliver the most significant productivity enhancement of your professional life. 104 00:10:19,664.01776101 --> 00:10:21,614.01776101 The real answer isn't up to the technology. 105 00:10:21,794.01776101 --> 00:10:25,124.01776101 It's up to us how we use question and control it. 106 00:10:25,604.01776101 --> 00:10:30,554.01776101 We need to design workplaces where AI amplifies what's best about human intelligence. 107 00:10:31,34.01776101 --> 00:10:35,474.01776101 Don't outsource your judgment, your skepticism, or your appetite for asking why. 108 00:10:35,894.01776101 --> 00:10:42,404.01776101 And before we write AI off as the world's worst brain drain device, let's look at what happens when we use it with purpose. 109 00:10:42,824.01776101 --> 00:10:48,314.01776101 There's research out there, OECD, McKinsey and others arguing that when used intentionally. 110 00:10:49,139.01776101 --> 00:10:52,319.01776101 AI is a genuine cognitive force multiplier. 111 00:10:52,799.01776101 --> 00:11:02,969.01776101 When the machine crunches the numbers and drafts the first version, your mind's freed up for the real stuff, strategy, deep learning, bold ideas, or just better questions. 112 00:11:03,629.01776101 --> 00:11:06,179.01776101 What we're seeing is a shift in what it means to be smart. 113 00:11:06,569.01776101 --> 00:11:15,389.01776101 It's less about remembering facts or setting up massive spreadsheets from scratch, and more about judgment, curating data, interpreting signals, applying ethics. 114 00:11:16,394.01776101 --> 00:11:17,114.01776101 The real win. 115 00:11:17,534.01776101 --> 00:11:19,724.01776101 You don't have to pick between speed and smarts. 116 00:11:19,814.01776101 --> 00:11:24,44.01776101 You get both if you stay in the driver's seat and never hand over the big calls. 117 00:11:24,794.01776101 --> 00:11:32,324.01776101 As AI keeps getting smarter, let's make sure we're designing the future on our terms using tech as a force for empowerment. 118 00:11:32,564.01776101 --> 00:11:34,94.01776101 Not an excuse to switch off. 119 00:11:34,934.01776101 --> 00:11:45,794.01776101 If today's show struck a chord, tell me, have you ever caught yourself blindly trusting an AI tool, or is this one of your big concerns for your team? Ping me at hello@theaibreakdown.com 120 00:11:45,794.01776101 --> 00:11:48,914.01776101 with your story and I'll feature the best on a future episode. 121 00:11:49,334.01776101 --> 00:11:51,284.01776101 And while you're here, hit subscribe. 122 00:11:51,344.01776101 --> 00:11:55,964.01776101 Give us a rating or share this episode with a colleague hooked on automating everything.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.