AI can already perform certain business tasks surprisingly well. But there is a major difference between a useful AI assistant and a dependable employee.

Businesses are beginning to explore AI systems for customer service, appointment scheduling, administrative work, sales assistance, and content production.

The idea is understandable.

A system that can answer routine questions, retrieve information, and assist customers outside normal operating hours could be useful, especially for small businesses with limited manpower.

But an important question deserves attention.

How much responsibility can businesses safely give these systems?

There is evidence of productivity benefits

A study by Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, published in The Quarterly Journal of Economics in 2025, examined the use of a generative AI assistant among 5,179 customer support agents.

The researchers found that access to the tool increased productivity, measured by issues resolved per hour, by an average of approximately 14%.

Novice and lower-skilled workers experienced substantially larger improvements, averaging 34%.

The results are encouraging, but they need to be interpreted correctly.

The research examined AI assistance within a particular customer support environment. It did not establish that an autonomous AI salesperson could replace an entire sales department or produce equivalent improvements in every business.

Where AI can be useful

Consider a small company receiving repeated inquiries about service availability, appointment schedules, business hours, and required documents.

An AI assistant connected to an accurate, maintained knowledge base could help answer these routine questions.

It might also collect initial information, organize inquiries, and direct customers to the appropriate employee.

These are relatively structured activities.

But suppose a customer asks for a special refund, disputes a billing charge, or requests advice outside the system's approved information.

A dependable system must recognize its limitations and transfer the issue to someone authorized to make a decision.

Customers still value human interaction

PwC's 2025 Customer Experience Survey found that 58% of surveyed US consumers were only somewhat comfortable or not comfortable at all using AI tools to engage with brands.

That matters.

Businesses shouldn't assume customers automatically prefer automated interactions simply because the technology is available.

For some situations, immediate automated assistance may be appreciated.

For others, customers may expect to speak to a real person.

The practical AI implementation test

Before deploying an AI assistant, clearly define the task it is supposed to perform.

For example, if its role is appointment assistance, test whether it can correctly explain available services, collect necessary details, follow booking rules, and recognize requests it cannot handle.

Measure accuracy, completion rates, escalation frequency, customer satisfaction, and operating costs.

Use safeguards to protect personal information. Require human approval for sensitive or consequential actions, and ensure customers can reach a person when necessary.

AI can become a valuable part of business operations.

But like any employee or business system, its usefulness depends on proper training, reliable information, defined responsibilities, and supervision.

The most useful question isn't how many people AI can replace. It's which activities AI can improve reliably, and where human judgment remains essential.

Research references

Brynjolfsson, Li & Raymond, Generative AI at Work, Quarterly Journal of Economics (2025): https://www.nber.org/papers/w31161

PwC, 2025 Customer Experience Survey: https://www.pwc.com/us/en/services/consulting/commercial-excellence/library/2025-customer-experience-survey.html