Balancing AI Adoption with Human Judgment in the Workplace
When I stood up at our all-staff meeting to announce that B:Side Capital was adopting AI, I came armed with a presentation about efficiency and the future of work. However, the first question I heard was not about productivity gains or innovation—it was, “Is this how the layoffs start?”
I don’t recall my exact response, but I remember the silence that followed before I answered. This moment highlights a common concern among employees as businesses integrate artificial intelligence into their workflows.
To provide context, I run a nonprofit lender specializing in Small Business Administration (SBA) loans and also founded Main & Machine, a company building AI systems for small businesses. This dual perspective—as a technology buyer and a builder—has shaped my understanding of how AI impacts teams and operations.
Understanding the Human Element Before Implementing AI
From both perspectives, I can confidently say that software alone does not determine success. Many business owners spend months comparing AI tools and pricing options, but ultimately, it’s their team that decides whether to trust and embrace the technology. This hesitation is not surprising: recent data from the Pew Research Center shows that 52% of U.S. workers are worried about how AI will be used in their workplaces.
Initially, I thought better training or a more user-friendly tool would alleviate these fears. Instead, what made the difference was a thoughtful reversal of the usual approach: before AI touched any workflow, we explicitly decided what it would never touch.
Decide What AI Will Never Touch Before It Touches Anything
At B:Side Capital, we categorized our work by judgment rather than by task. The AI system never makes credit decisions alone, never communicates hardship conversations with borrowers, and never commits the company to any actions. The principle is simple but powerful: a machine can hold knowledge, but it cannot hold responsibility.
When a borrower calls because their business is struggling, they want to speak to a person who can take ownership of the situation—not an algorithm. To replicate this approach in your business, try sorting tasks into three buckets: automate, assist, and human-owned.
- Automate: Tasks where errors are low risk and easily corrected.
- Assist: Tasks where the machine drafts outputs but a human makes the final decision.
- Human-owned: Critical work that builds trust and must remain in human hands.
For example, a restaurant might automate inventory counts, have AI draft the weekly schedule, but never allow AI to interact directly with unhappy customers. The exact tasks will differ across industries, but the framework is universally applicable.
Lead with What Will Not Change
When I first announced AI adoption, I focused on efficiency improvements. But this pitch failed because employees heard a subtext of potential job cuts. Observing their reactions, I shifted the message to emphasize what would remain constant: people would continue to make every credit decision, no customer hardship conversations would be handled by AI, and employees who embraced the technology would be rewarded, not penalized.
Making these commitments upfront required no financial investment but proved invaluable in building trust. Once these boundaries were clear, the team stopped scanning for threats and began genuinely engaging with how AI could help them.
This pattern appears across industries: the most successful AI adopters aren’t those with the flashiest software but those whose leaders openly declare what will always stay human.
Give the Machine the Work Nobody Will Miss
Our first instinct was to build a flagship AI system to impress stakeholders. Instead, we chose to automate document intake—the tedious sorting, checking, and transcribing that no one enjoyed. This decision paid off.
Our AI, named MARCUS after the stoic Roman emperor Marcus Aurelius, performs comprehensive loan file reviews. It cross-checks documents and flags discrepancies a junior analyst might catch. Crucially, every conclusion MARCUS reaches can be traced, questioned, and overruled by a human, ensuring transparency and avoiding the dreaded “black box” effect.
The results speak for themselves: what once took three to four hours of manual review now takes less than one, freeing staff to focus on judgment calls and borrower conversations. Nobody missed the tedious transcription work.
Within a quarter, nearly the entire team was using MARCUS without prompting. AI adoption took off because it relieved employees of unwanted tasks, not because it replaced them.
Practical Steps to Start AI Adoption the Right Way
Here’s where I suggest starting this week: write down the three things in your business that only a human should ever do. Share this list openly with your team before spending a single dollar on software. Then, confidently hand the machine the tedium.
To answer that initial question from my all-staff meeting honestly: no, AI adoption is not how layoffs start. Done right, AI creates space for judgment work to flourish. The best AI companies will not be the least human—they will be the ones who understand the irreplaceable value of human responsibility and trust.
Key Takeaways
- Write down the three things in your business that only a human should ever do, and say them to your team before you spend a dollar on software. Then hand the machine the tedium.
- Done right, AI isn’t how the layoffs start; it’s how the judgment work finally gets room to breathe. The best AI companies will not be the least human.
Here is the original source of this discussion on balancing AI integration with human trust in the workplace.
