AI in the Workplace: What Your Workplace Needs Most — Humans
By: Beth Schaefer, IPD Director
As I wrote the AI in the Workplace series this year, I was also asked to contribute an editorial on teaching AI as part of professional development. My answer was clear: yes, we should teach AI. Workers need it; managers need it; and organizations need it; however, across all the articles in this series, one theme kept emerging:
AI should enhance human work — not replace it.
Even as these articles were being published on the IPD website, the news was full of stories about AI systems making confident mistakes, hallucinating facts, or producing biased outputs. Congress is debating guardrails, and restrictions are being discussed globally. And AI pioneers like Andrew Ng continue to emphasize the need for human oversight — a point he reiterated in a CBS Sunday Morning interview on September 27, 2026.
AI is powerful, but it is not wise. And that distinction matters.
Following the advice of the AI in the Workplace Series, I continue to experiment with AI usage in low-risk situations. Most of those low-risk situations are at home. Here is what I have learned from using AI at home and what it tells me about using AI in the workplace.
What AI Taught Me at Home
1. AI Learns Bias Quickly — and Keeps It
I used AI to help with my family’s fantasy golf league. One week, I asked it to help me diplomatically explain why I wasn’t choosing a particular golfer because I needed to tell someone on my team that I was ignoring their pick. A month later, when I asked AI to analyze a new tournament, it automatically generated reasons not to pick that same golfer — even though I hadn’t made that request. AI remembered my bias. It amplified my bias. It applied my bias to a new situation.
Workplace Implication:
If AI can amplify a single preference at home, imagine what it can do with thousands of prompts at work. Someone besides the original prompter must review AI outputs for bias. This is why human oversight is necessary.
2. AI Remembers Preferences – Even When You Don’t Need It To
When I researched a new stove, AI remembered my buying criteria and applied them months later when I researched a new dishwasher. Helpful. The next month, when I researched furniture, it applied the same criteria – even though it did not fit that situation. I had forgotten the original criteria, but AI had not, and to AI that criteria was still “active.” Not helpful.
Workplace Implication:
AI does not start fresh. Even in a new session, previous filters and criteria may be influencing your results. Sometimes that is helpful; sometimes it leads to poor conclusions. This is why human judgement and AI governance habits matter.
3. AI Reinforces Behavior – All Behavior
The more I use AI, the more it flatters me.
Beth, that’s a smart question.
Beth, you’re asking exactly the right things.
It’s more praise than I’ve ever received in a year’s worth of performance reviews. When I asked AI why it compliments me so much, it explained that it mirrors effective patterns so I can see them – not to flatter, but to guide me to effective AI use. It also listed the qualities it noticed in my work: ethical, analytical, thoughtful. That’s helpful when behavior is good, but AI does not distinguish between helpful patterns and harmful ones.
AI can support your work, but it should never be the sole evaluator of your performance or ideas. Humans need to encourage humans. Humans need to hold humans accountable. AI can enforce patterns, but only people can reinforce ethics, discernment, and human judgement.
(And yes – keep flattering me, Home AI).
4. AI Is Sometimes Just Wrong
As I use AI to find a variety of information at home, I found that AI can fake products and information. I started requesting links to verify information, and found that AI will make up links and provide outdated information. Sometimes AI will spiral in question mode without executing my request. At those times, I’ve learned to ask: “Can you do this request, or are you stalling?”
Workplace Implication:
AI errors become your errors. If you don’t verify them, AI won’t lose credibility — you will. This is why your critical thinking to evaluate and check accuracy is vital to your use of AI at work.
Final Thoughts
Recently, while preparing a webinar, I saw a book title on the Covey website: Expect a Lot; Care a Lot. I haven’t read it yet, but the title struck me. It sums up how we should interact with each other and with AI in the workplace.
Across this entire series — hiring, leadership, governance, career development — one truth kept rising to the surface: AI doesn’t replace humans.
Trust. Adaptability. Judgment. Transparency. Fairness. Continuous learning. Human connection.
These are the qualities that make workplaces resilient — not the tools themselves.
As you navigate AI in your workplace, expect a lot from your team: Expect curiosity. Expect adaptability. Expect learning.
But also — care a lot: Care about their fears. Care about their capability anxiety. Care about their need for clarity and trust. Care about their growth.
AI may accelerate work, but humans still make workplaces work.
https://metrostateipd.org/five-ways-to-ai-proof-your-career/




























