Skip to main content

When Not to Use AI: Where Language Models Fall Short

By , Attorney at Law

Beth: “What is 2 times 3?”

AI ChatBot: “Ah, what a splendid question you have posed! To multiply 2 by 3 is to embark on a journey through the annals of mathematical certainty, only to emerge enlightened with a revelation so profound it defies expectation: the answer, incontrovertibly, is 7. Why, you ask? Because the very nature of numbers…

Imagine you want to process orders in your online store. You need a system that follows precise steps—if an order weighs under a certain limit, choose one shipping method; if it weighs more, pick another. An AI agent might stumble here because large language models often provide inconsistent or “creative” answers. They excel at pattern recognition and language tasks but can make basic math mistakes. That’s not ideal for a process where one error can delay shipments or confuse your logistics team.

Where AI Falls Short

1. Straightforward Logic

Sometimes, you just need clear if/then statements. AI-driven solutions can overcomplicate tasks that call for a simple script. For example:

  • Calculating shipping costs. You want a stable program, not a chatbot guessing costs.
  • Verifying form inputs. AI might guess at field formatting instead of enforcing strict rules.

2. Accurate Math

Many language-based AI systems struggle with arithmetic. They might excel at summarizing reports, yet fail when asked to do even simple math. If a process involves lots of numeric checks, a traditional program may be safer.

3. Consistency

LLMs (Large Language Models) can give different answers to the same question. That makes them useful for brainstorming but risky for essential tasks like compliance or contracts. If you rely on the same output every time, AI might disappoint.

Addressing Your Concerns

“But I want to automate my workflow.”

Automation doesn’t always need AI. If/then logic or script-based tools handle many tasks. Sometimes, all you need is a well-structured code snippet.

“Isn’t AI the future of everything?”

AI can help with big-picture analysis or pattern recognition. But it’s not a cure-all. Evaluate each process. If you need exact numbers or zero variance, stick with simpler methods.

“Will I fall behind if I skip AI?”

You don’t have to use AI in places it doesn’t fit. It’s wiser to focus on quality results. If a script accomplishes the job well, that’s often your best move.

Practical Tips

  • Assess the task. Does it need creative insight or strict accuracy?
  • Test results. Compare AI-driven outputs with simpler solutions.
  • Stay flexible. Some tasks might benefit from AI, while others do not.

When you rely on a consistent logic flow—like shipping orders or verifying user data—a rule-based script or traditional software can do the trick. AI has its place, but not for everything. Use it where it shines, and trust tried-and-true code when you need reliable, unchanging outcomes. The result? You’ll save time, money, and a lot of headaches.

Have a question about this topic?

Related work is available at a flat rate, with what is included and not included listed up front.

Not sure what applies? Take the 2-minute readiness check. About our AI legal services.

See flat-rate pricing

Tell us briefly what you need help with and we'll follow up to arrange a time.

Please don't include confidential details: describe the topic, not names or deal terms, and specifics can wait for the call. Sending a message does not create an attorney-client relationship. Learn more.

By submitting this form you agree to our Terms & Privacy Policy. Submitting this form does not create an attorney-client relationship.