Insights

When AI is the wrong answer

By Isaac Salvant Jr., founder of AI360

We are an AI firm, and part of our job is telling clients when not to use it. Here are five situations where AI is the wrong answer, and what to do in each.

1. When the process itself is broken

If orders get lost because nobody owns the handoff between sales and delivery, AI will lose them faster. Fix the steps first. Automation multiplies whatever it is given, including confusion.

2. When the decision needs human judgment

Pricing a custom job, handling an upset customer, giving professional advice, deciding whom to hire. AI can prepare the information, but a person should make the call and stand behind it.

3. When the information is not there

AI works from what you give it. If customer history lives in one person's memory and a stack of paper, the first project is getting it into one place, not buying a smart tool to sit on top of nothing.

4. When a simpler tool does the job

A booking link, a template, a shared checklist or a form often solves the problem in an afternoon. We would rather tell you that than sell you something bigger.

5. When your team has not been trained

This is the one we see most, and the one that matters most. A business buys an AI tool, a few people try it, results are uneven, and within weeks everyone is back to the old way. The tool was fine. Nobody was shown how to use it.

Good AI training for employees is practical and short. It covers:

• What to use it for in their actual job, with examples from their own work.
• What never to put into it, such as confidential customer or financial information in tools that are not approved for it.
• How to check the output, because AI can be confidently wrong.
• Who is responsible for work that AI helped produce. The answer is always a person.

Trained people get more from a simple tool than untrained people get from an advanced one. That is what we mean by Human-Led AI, and team training is part of what AI360 provides.

So when is AI the right answer?

When the work is repetitive, high in volume, low in risk and easy for a person to review: first replies to inquiries, reminders, summaries, first drafts, sorting and routing. In those places it gives hours back every week, as long as a trained person stays in charge.

Back to all insights