GenAI may be creating more work than it saves

Generative AI, like large language models (LLMs), is often praised for saving time and boosting productivity. However, building and maintaining these systems might require more work than the time they save. Plus, many tasks don't need the complexity of AI—basic automation is often enough.

What

For example, people expected self-driving trucks to revolutionize transportation years ago, but those predictions didn’t come true due to real-world challenges like regulations, insurance, and software problems. Similarly, using AI in areas like software development and business faces practical hurdles.

Why AI Might Add More Work

1. Complexity: Generative AI needs a lot of backend work, like managing data and resolving issues.
2. Validation: AI outputs must be checked for accuracy, often by experts, which takes time and effort.
3. Cost: Running LLMs can get expensive due to high computing and electricity needs.
4. Too Much Information: AI can generate lots of reports, sometimes with conflicting results, making it hard to decide what’s correct.
5. Human Preferences: People often rely on gut feelings or personal preferences instead of AI recommendations. For example, managers may ignore AI-based hiring suggestions if they prefer candidates they like.

Where Generative AI Can Help

AI is best used for analyzing large amounts of data. It can process information faster than humans and help with decision-making. However, people still need to manage the databases, handle errors, and set clear rules for using the data.

Challenges to Consider

1. Many tasks don’t need AI—existing tools like templates or automation are often enough.
2. AI-generated work often requires review, especially for legal or important decisions.
3. Organizations need to avoid drowning in unnecessary or conflicting information.

In conclusion, while generative AI is powerful, it isn’t always practical or efficient for every task. It works best as a tool to support humans, not replace them entirely.