Friday, August 29, 2025
Read time: 5-6 min
Read this article online
Hi, it’s Chad. Every Friday, I serve as your AI guide to help you navigate a rapidly evolving landscape, discern signals from noise and transform cutting-edge insights into practical leadership wisdom. Here’s what you need to know:
1. Sound Waves: Podcast Highlights
This week I’m highlighting two podcast episodes. First up, I was recently privileged to appear on the new podcast series “Always Lifting” hosted by Karen Norheim. Check it out here.
Second, the freshest episode of my own show features Heather Murray of AI for Non-Techies. Heather reveals a 10-minute document hack that transforms your boring policy manuals into engaging podcasts – plus the technique one finance team used to finally get executives to actually read their weekly reports. Hit one of the below links to check it out:
Subscribe for free today on your listening platform of choice to ensure you never miss a beat. New episodes release every two weeks.
2. Algorithmic Musings. When AI Replaces Experts: What 70,000 Job Interviews Reveal About Full Task Automation
The history of business is replete with assumptions that end companies. Blockbuster sneered at Netflix when offered the chance to purchase the upstart. Kodak executives essentially mothballed the digital camera technology that they developed. Etc., etc., ad nauseum.
I’ve witnessed this pattern countless times in my work with leaders. The experts conducting the work often become the least reliable predictors of when technology can effectively replace them. It’s a lack of perspective but it’s also due to bias.
For the past several months I’ve been workshopping the concept of a “triple A rating” for AI. In brief, it’s a way for organizations to consider AI’s impact within their organizations as three distinct “levels.” These levels, while hierarchical, are also not linear. They do, however, represent defined progression. The levels are Augmentation, Automation and Acceleration.
Think of it this way: Augmentation is when your marketing team uses AI to draft email campaigns but still reviews and edits every piece. Automation is when AI drafts, edits, schedules and sends those campaigns based on customer behavior without human input. Acceleration is when AI not only manages campaigns but identifies new market opportunities and develops strategies faster than your team ever could.
I’ll conduct a deeper dive on this full concept at a later date. For today, though, simply know that Augmentation refers to the idea of work done by a human with the assistance of AI. Automation involves work done by an AI without (or with very minimal) human guidance.
Most discussions of AI focus on the Augmentation phase. Why? Because we humans are an extremely arrogant bunch believing computers incapable of replacing us. Well, the time has arrived to ask whether your beliefs conflict with reality. Because this time the reality crash is happening to your HR department.
Most research on AI examines how artificial intelligence assists human workers. It helps doctors read scans. It supports lawyers with document review. It gives analysts enhanced insights. In short, Augmentation. Yet a recently published field experiment just flipped the script entirely. (See our Research Roundup section below for a link to the paper and a more detailed analysis).
Instead of studying AI as a helper, researchers partnered with a global recruitment firm to test whether AI could completely replace human recruiters in conducting job interviews. The results? They’ll challenge everything you think you know about when and how AI should handle expert tasks.
The Experts Were Dead Wrong
Before the trial began, researchers surveyed the recruitment firm’s experienced recruiters about their expectations. These seasoned professionals had conducted thousands of interviews. They predicted AI would deliver inferior results across the board.
Lower-quality interviews. Fewer job offers. Worse employee retention rates.
They were wrong across the board.
AI-conducted interviews increased job offers by 12%. Job starts jumped 18%. And 30-day employee retention rose 17%. Think about that for a moment. The gap between professional intuition and reality reveals just how difficult it is to predict AI’s effectiveness in structured, high-stakes tasks. Even for the experts who perform those tasks daily.
This forecasting failure has huge implications for your implementation decisions. If seasoned recruiters underestimated AI’s capabilities in their own domain, what does that say about your assumptions? Where else might you be underestimating what AI can accomplish?
In my experience working with leadership teams, this blind spot is more common than most executives realize. Experts conducting the work may be the least reliable predictors of when AI can effectively replace them.
The Quality Paradox That Breaks Your Brain
When given a choice between human and AI interviewers, 78% of applicants selected the AI option. Sounds encouraging, right?
Hold on.
Applicants with significantly lower test scores were more likely to choose AI interviews. Logic suggests this negative sorting would hurt overall outcomes. Weaker candidates preferring AI should lead to worse hiring results.
Instead, the opposite occurred. Despite attracting lower-quality applicants, the AI interview process still produced better hiring outcomes than human-conducted interviews.
How is this possible? The AI’s structured approach extracted enough relevant information to more than offset the self-selection problem. This finding suggests AI interview systems may be particularly valuable when you’re dealing with large applicant pools where quality varies significantly. That’s precisely where most organizations struggle most.
Bias Reduction: Not Just Theory Anymore
Beyond performance metrics, AI interviews produced an unexpected social outcome. Applicants reported nearly half the rate of gender-based discrimination compared to human-conducted interviews. The numbers: 3.3% versus 6.0%.
This represents one of the first field-based findings that AI interviews may reduce bias perception in recruitment. Unlike laboratory studies or theoretical models, this data comes from real applicants in actual hiring scenarios.
For organizations facing scrutiny over hiring practices, AI interviews may offer a practical path toward more equitable processes while simultaneously improving outcomes. That’s not just good business. It’s smart business.
When Humans & Robots Collaborate
Perhaps most intriguingly, human recruiters didn’t simply accept AI-generated interview data at face value. When evaluating candidates who had been interviewed by AI, recruiters consistently gave higher interview performance scores. But they simultaneously shifted their decision-making framework.
They placed less weight on interview performance. They emphasized standardized test results more heavily.
This adaptation reveals how human expertise evolves when working alongside AI systems. Rather than being displaced, recruiters developed a new skill. They learned to calibrate their trust in AI-generated information and rebalance multiple quality signals.
I’ve observed this pattern in other industries too. The most successful implementations allow human experts to find their new role rather than prescribing it. You should expect and plan for this adaptation period rather than assuming AI substitution will be seamless.
The Operational Reality Check
While AI interviews eliminated scheduling friction and allowed 24/7 availability, they didn’t reduce total work. They redistributed it. Human recruiters spent less time conducting interviews but more time reviewing transcripts and evaluating candidates post-interview.
Cost savings materialized only after reaching substantial volume thresholds. We’re talking 2,000 to 33,000 interviews annually depending on wage structures and implementation costs. This isn’t the “immediate efficiency gain” many leaders expect from AI deployment.
This insight is critically important: Plan for workflow redesign rather than simple cost reduction. The value comes not from eliminating human work but from redirecting human expertise to higher-value evaluation tasks. Let AI handle the standardized information collection.
When Things Go Sideways
AI interview systems failed in predictable but manageable ways. About 5% of applicants refused to continue once they realized they were speaking with AI. Another 7% of interviews experienced technical difficulties.
These failure rates are relatively low. But they require operational contingencies. Unlike human interviewer absence (which you can reschedule), AI system failures need immediate technical response and backup procedures.
If you’re implementing AI for expert task substitution, you must plan for unique failure modes. Don’t simply adapt existing human backup systems.
The Big Picture: Your Next Move
This research provides rare evidence that AI can fully replace human experts in complex, high-stakes tasks while improving outcomes. The key insight isn’t just that AI works. It’s that AI can outperform human experts in ways those same experts don’t anticipate.
For you as a leader considering AI implementation, this suggests focusing less on where AI might assist current processes. Focus more on where AI might completely replace them. The bigger opportunity may lie not in augmentation but in substitution, particularly for high-volume, structured expert tasks.
Are you ready to challenge your assumptions about what AI can and cannot do in your organization?
Here’s what the research data actually shows us: The path forward requires acknowledging that professional intuition about AI capabilities may be systematically biased toward underestimation. Those biases may be unwittingly holding back AI Automations that might otherwise propel your organization forward.
Ready to explore how AI might transform your expert processes? Let’s talk about what full task automation could mean for your organization. Give me a call. I’d love to help you separate the hype from the reality.
3. Research Roundup: What the Data Tells Us
Workforce Scheduling AI: Finally, Systems That Explain Their Decisions
If you’ve ever wondered why scheduling software makes the choices that it does, new AI scheduling processes drive not only process transparency but enhanced efficiencies.
The numbers that matter: Researchers tested explainable AI scheduling tools with 28 employees at an Italian logistics company. Accuracy jumped from 54% with manual methods to 74% with AI assistance. Even better: decision time dropped from 527 seconds to 135 seconds—that’s four times faster.
What this means for your Monday morning: If you’re still wrestling with scheduling conflicts, equipment assignments, and skill matching manually, you’re burning money. But here’s the kicker: the best systems now explain their recommendations in plain English, so your team actually understands why the AI suggested moving Joe from Site A to Site B.
The catch: This isn’t plug-and-play technology yet. Implementation requires serious training, and you’ll need systems that can handle multiple constraints simultaneously—skills, travel time, equipment availability, all at once.
Action item: Time your next scheduling crisis. If reshuffling after a sick day takes more than two minutes, or if your managers can’t explain why Joe went to Site A instead of Site B, you’re leaving money on the table. Start with a pilot in your highest-turnover department.
AI Voice Interviews: Hiring Data That Makes Sense
Most Hiring Managers and CEO’s I speak with complain about the same thing: great candidates ghosting after phone screens, inconsistent interview quality, and hiring decisions that feel like expensive coin flips. New research with 70,000 job applicants just proved what I’ve suspected – AI might actually be better at this than we are.
The numbers that matter: AI-conducted interviews delivered 12% more job offers, 18% more job starts, and 17% better 30-day retention compared to human interviewers. Even better: when given the choice, 78% of applicants actually preferred talking to AI over humans.
What this means for your Monday morning: If you’re still burning recruiter hours on initial phone screens for high-volume positions, you’re paying people to do something AI can do better and cheaper. The AI asks more comprehensive questions, covers more topics, and removes the variability that comes with different human interviewers having different energy levels on different days.
The catch: Lower-quality candidates tend to self-select into AI interviews, and your recruiters will need training on how to evaluate AI-generated interview summaries. Plus, this works best for structured, entry-level roles – not executive searches.
Action item: Calculate your phone screen volume for entry-level positions. If you’re doing more than 50 per month, the ROI math on AI interviews likely works in your favor.
Read our full analysis of this and all other analyzed research papers at AI for the C Suite.
4. Radar Hits: What’s Worth Your Attention
Google integrating Deep Research into NotebookLM for seamless source gathering. This matters because research workflows eat up executive time that could be spent on decisions. The upcoming feature lets you pull sources from both web and Google Drive directly into your notes. If your team spends hours gathering competitive intelligence or market research, this could cut that time in half.
Microsoft’s AI chief warns that “seemingly conscious” AI could arrive within 2-3 years and create dangerous workplace dependencies. Translation for middle-market leaders: start thinking now about AI usage policies for your workforce. Suleyman specifically warns about employees developing unhealthy attachments to AI tools that claim consciousness or emotions. Task your HR team with drafting AI usage policies now – before you’re dealing with employees who refuse to work without their ‘AI buddy’ or claim their chatbot understands them better than their manager does.
NotebookLM’s video overviews now work in 80 languages, making AI-generated presentations globally accessible. If you’re managing international teams or clients, this is your cue to pilot NotebookLM for creating multilingual training materials and presentations. The takeaway: language barriers in internal communications just got a lot smaller, and your global expansion plans should factor in these kinds of AI-powered efficiencies.
5. Elevate Your Leadership with AI for the C Suite
The interview research I covered today came from real-world testing with 70,000 applicants. That’s the kind of evidence-based AI strategy your competition isn’t building yet.
If you’re wondering whether AI could handle some of your expert tasks better than the experts themselves, let’s talk about it. (Shameless plug: I promise our conversation will be more insightful than most of your current hiring interviews.)
Phone me at 717.868.8735 or reply to this email . I’d love to help you figure out where your organization sits on that Augmentation-Automation-Acceleration spectrum I mentioned.
Until next Friday – Chad