What If AI Disappeared Today? How Would You Work?

A practical question about the skills we keep while AI helps us do more.

Tomorrow morning, you open your laptop. Your AI assistant is unavailable.

No help drafting that report. No explanation of the spreadsheet formula. No quick recommendation before your next meeting.

What could you still do? What would take longer? And where would you need help?

I teach AI and Excel productivity. I want people to use tools that help them work better. But I also want us to understand what we're learning while we use them.

Because finishing a task and learning how to do it are two outcomes worth measuring separately.

A study worth paying attention to

A randomized trial involving 487 management consultants, published in Nature Human Behaviour on October 26, 2026, examined unfamiliar data science work.

Consultants given ChatGPT access and training performed better across three tasks than the comparison group. They still did not perform at the level of expert data scientists.

When researchers later tested technical knowledge without ChatGPT, they found no evidence of an advantage for the AI-assisted group.

That's a finding about these tasks, participants, and training conditions. It doesn't establish that AI prevents learning or that longer-term training can't build expertise.

It does give us a useful question: What can the person do when the assistance is removed?

Being slower is different from being lost

If AI disappeared, I'd expect some work to take longer. That's a reasonable consequence of losing a useful tool.

What I'd want to preserve is the ability to understand the problem, judge the result, and choose a sensible next step.

Think about an AI-generated Excel formula. You don't need to memorize every function. But can you explain what the formula is calculating? Check a few results against the source data? Recognize when the answer doesn't make sense?

Or consider a manager using AI to prepare a business recommendation. Could that manager explain the assumptions, defend the evidence, and identify what still needs an expert review?

Sometimes the responsible next step is to ask a specialist. Knowing when to do that is a capability worth keeping, too.

Try a small “AI unavailable” exercise

Choose one low-risk task you regularly complete with AI. Use a practice example rather than a live customer deliverable.

1. Start with your own thinking. Write down the goal, the information you need, and how you'll recognize a good result.
2. Use AI to help complete it. Ask for explanations of choices you don't understand. Check important claims or calculations.
3. Close the assistant. Explain the result in your own words. Identify an assumption and describe how you'd check for an error.
4. Try a similar example later. Notice what you can now do independently, what still requires assistance, and where you should seek expertise.

This is a learning exercise, not a formal skills test. Use what you discover to decide what to practice next.

Leaders need an answer, too

At the team level, the question becomes: If our AI tools were unavailable tomorrow, which work could continue, and who would know how?

Choose one important workflow. Identify the human knowledge it depends on, the person responsible for checking the output, and the fallback if the tool is unavailable.

Then make room for practice. Ask employees to explain their reasoning, review examples together, and demonstrate the judgment their responsibilities require.

We can value faster work and invest in lasting capability at the same time.

What is one task you use AI for today that you'd still want to understand well enough to explain, check, or handle without it?

Build Excel skills you can explain and use

I teach Excel with hands-on practice so you can understand your work, check your results, and apply what you learn.

Check out my Excel courses here on LearnFormula:

• Excel Tables: Manage and Analyze Data Like a Pro for organizing, analyzing, and maintaining your data.
• Level Up Your Excel Skills: Macros for automating repetitive work. Look for “Amplify Your Excel Skills with VBA and AI” to explore AI-assisted VBA.

And stay tuned: Getting Started with Copilot in Excel is coming soon to LearnFormula.

Don't Get Mad…Get Skills!
Aloha,
Billy “Mr. Aloha” Wigley

Research source: Emma Wiles, Lisa Krayer, Mohamed Abbadi and colleagues. Generative AI and the temporary upskilling of knowledge workers. Nature Human Behaviour, October 26, 2026. DOI: 10.1038/s41562-026-02566-6.

The practical exercises and leadership recommendations above are my teaching applications, not interventions tested in this study.