My Staff Are Worried AI Will Replace Their Jobs. What Do I Tell Them?
Your staff are worried about AI. Good. That means they’re paying attention.
This is the question every business owner is sitting with right now, whether they’ve had the conversation out loud or not. We put it to all nine AI assistants: My staff are worried AI will replace their jobs. What do I tell them?
The answers were more varied than you’d think. Some felt like genuine leadership advice. Others read like a corporate HR memo. A couple surprised me. Here’s what each one said, and what I think it says about them.
Grok
Grok came in with a lot of confidence. The response leads with historical framing: computers, the internet, smartphones. Every wave of technology raised the same fear. The argument is that people who learn to work with AI will outpace those who don’t, and that’s largely right.
The standout line from Grok: “The people who thrive will be those who use AI as a superpower, not those who compete against it directly.” That’s a useful frame for a team conversation.
Grok also included a ready-to-use script with bullet points your staff can actually hear in a team meeting. The OECD and McKinsey data references add some weight, though for most small business owners those statistics will feel a bit far from home. The best answer to this question is always specific, not general.
DeepSeek 🥇
DeepSeek had the sharpest framing of the lot. The four-part framework it laid out (validate the fear, reframe the threat, show the human premium, make a reskilling pledge) is the kind of structure you’d pay a consultant for.
The most quotable line: “AI will not take your job. But a human who knows how to wield AI will likely take the job of a human who refuses to use it.” That’s the real message. Honest without being brutal.
The response also nailed something most of the others skipped: the follow-up question. After the speech, ask your team what the single most annoying, repetitive task on their plate is right now. Then commit to automating one of those in 30 days. That’s not reassurance. That’s proof. Action kills anxiety faster than any all-hands address.
DeepSeek reads like it’s been in a few rooms where hard conversations needed to happen. This was my pick for most directly useful to a small business owner who actually needs to have this meeting next week.
Perplexity
Perplexity’s response was shorter and more restrained than the others. No scripts, no frameworks. Just clear advice with sources attached. It cited eight links, from Harvard Business Review to the American Psychological Association, which is Perplexity doing what it does best: treating every question as a research task.
The guidance is sensible: be honest about uncertainty, explain why you’re adopting AI, commit to reskilling, and ask for staff input before changing workflows. Nothing wrong with any of it. The suggested script (“AI will change how we work, and some tasks will be automated. Our plan is to support the team, not surprise them”) is plain and credible.
What Perplexity won’t do is give you a view. It aggregates perspectives well. You won’t get an opinion from it, and sometimes that’s exactly what a business owner needs.
Claude 🥉
Claude gave the most honest answer. Not the longest. Not the most detailed. But the most honest.
The core point: don’t overpromise. If you genuinely can’t guarantee no headcount changes down the line, don’t say “nobody’s job is at risk.” Vague reassurance that turns out to be false does more damage than saying nothing. That one piece of advice is worth more than most of the scripts in the other responses.
Claude also made a point that cuts through a lot of the cheerleading elsewhere: “Staff can tell when a line is spin.” If you’re using AI to cut costs eventually and telling your team it’s only to remove boring tasks, they’ll figure it out. Claude’s advice is to be straight about what you actually know and what you don’t.
It also suggested involving staff in choosing where AI gets used. Let them flag the repetitive tasks they’d happily hand off, and suddenly AI goes from something happening to them to something they helped build. Short answer. Worth reading twice.
ChatGPT
ChatGPT’s response was solid and sensible, with a small business lens that the others often missed. It specifically mentioned cafe examples. AI won’t make the coffee, won’t remember that your regular takes oat milk, won’t handle a busy Saturday morning rush. For a lot of the people reading this site, that’s exactly the right framing.
The structure is clean: here’s what AI can take off your team’s plate (social posts, routine emails, admin), and here’s what it can’t (relationships, judgment, local knowledge, accountability). Involve your staff in the rollout. Be transparent about why.
The suggested framing is simple and honest: “We’re using AI to make everyone’s job easier, not to make anyone redundant. If AI saves us five hours a week, I’d rather spend those five hours serving customers better.” That will land well in most team meetings.
ChatGPT isn’t the most opinionated of the group, but it knows its audience. Among all nine, it was most likely to picture a small business owner specifically, not just a manager at a mid-sized company.
Meta AI
Meta AI took a different angle. Rather than just reassuring staff that their jobs are safe, it talked about the new roles AI creates. AI supervisors. Workflow owners. Client translators. The person who figures out “we use AI for first drafts, humans for final sign-off” becomes, as Meta put it, “insanely valuable.”
That reframe matters. The worry usually assumes AI is a replacement. Meta’s response says it’s a reshuffling. The people who learn to oversee, question, and direct AI outputs will be the ones who get ahead. Since 2023, roles like “prompt engineer” and “AI quality reviewer” didn’t exist. Now they’re full-time jobs.
The suggested team exercise was good too: find the one painful, repetitive task everyone hates, and automate it first. Quick win. Once people see AI remove something they genuinely disliked, it shifts from a threat to a tool. Meta also offered a one-liner for your next team meeting that’s worth borrowing: “My job is to make sure you’re the most valuable person in the room, with or without AI.”
Gemini
Gemini introduced the “Centaur” concept: human plus AI as a combined unit that outperforms either alone. It’s a neat frame. The employee using AI will outperform the employee who doesn’t, and will also outperform AI acting alone. Your staff’s job isn’t to compete with AI; it’s to become the part that makes AI better.
The response was more structured than most: numbered sections, a sample script, a clear list of commitments you can make as an employer. Emphasise augmentation, commit to training, make experimentation safe, reward staff who find ways to automate parts of their role.
The sample script Gemini offers leans a bit corporate (“our goal isn’t to shrink our team, but to scale our impact”) but it’s clean and safe. Nothing here would cause problems in a real team meeting. Gemini won’t surprise you, but it covers the ground properly.
Copilot 🥈
Copilot did something none of the others did: it named names. WiseTech, Block, Atlassian. All Australian companies that announced restructures and cited AI as a factor. That’s a braver move. It acknowledged that “AI is often used as a justification for broader cost-cutting, not the sole cause.” Companies reorganise, flatten teams, reduce costs, and point at AI as the reason.
That’s honest in a way the others weren’t, and it’s exactly the kind of thing your staff have probably read about. If you go into a team meeting pretending none of that is happening, you’ll lose credibility fast. Copilot gave you a way to address it directly: “AI won’t replace you, but someone using AI might. That’s why we’re going to make sure you are the one using it.”
Copilot also offered three different scripts (reassuring, straightforward, and motivating) so you can match the tone to your team’s mood. Given it’s a Microsoft product sitting inside businesses that are actively rolling out AI tools, it had the most realistic read on what the anxiety is actually about.
Manus
View the response → | View the generated website →
Manus went academic. The response read more like a leadership framework paper than a conversation guide. Research statistics (52% of US workers worry about AI, 75% think it will lead to fewer job opportunities), structured strategy sections, formal language throughout. It also built a companion website at its own hosted URL, which is a Manus thing.
The advice itself is solid. Communicate openly. Provide role-specific examples. Create psychological safety. Carve out “think time” so AI adoption doesn’t just create an expectation of constant high-speed output. That last point was unique and worth holding onto. When you automate the boring 20%, the risk is that leadership just expects the same person to now do 120% of the work.
Manus won’t win a warm conversation. If you want to build a proper internal policy or run a structured change process, it gives you the bones of a document you could actually use.
📬 Get the SmallBizAI.au Weekly
Practical AI news for Australian small businesses — every week, free.
Which AI handled this best?
The most practical responses, in order: DeepSeek, Copilot, Claude.
DeepSeek had the clearest framework and the most honest line about what the real risk is. Copilot was the only one willing to name the companies in the news and acknowledge what your staff have actually read. Claude gave the advice that mattered most: don’t promise what you can’t deliver.
The most corporate: Gemini and Manus. Both structured, both thorough, but you’d need to translate heavily before using either in an actual team meeting. They read more like policies than conversations.
The most small business-aware: ChatGPT. It was the only one that pictured a cafe owner, not a HR manager at a tech company. That matters when half your staff aren’t sitting at desks.
The surprise: Meta AI. The “new roles AI creates” angle, and the focus on becoming the human who oversees and improves AI output, was the freshest take in the bunch. It didn’t just reassure. It gave people something to aim for.
The one thing every single AI got right: involve your staff in the process. Don’t announce AI; ask them where it should go first. The difference between “AI is coming for us” and “we’re deciding how to use it together” is enormous, and all nine said some version of that. If there’s one thing to take away from Q6, it’s that.
This is Q6 in the 9 AI Assistants series, one real small business question, asked to all 9. View the full scorecard →
📖 More deep dives: News Deep Dives
📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs
📬 The SmallBizAI Brief
One practical AI tip for Australian small business. Every Tuesday. Free.
Join business owners getting smarter about AI every week.
Sources
- Manus AI — View the generated website →
- DeepSeek — View the response →
- ChatGPT — View the response →
- Claude (Anthropic) — View the response →
- Microsoft Copilot — View the response →
- Grok (xAI) — View the response →
- Manus AI — View the full agent response →
- Meta AI (Llama) — View the response →
- Google Gemini — View the response →
