Which Model Should I Use?

At the AI Symposium at Millersville University last week, someone asked a great question: “What AI model should I use?”

The room was full of educators, so I answered with an analogy to Bloom’s Taxonomy.

Smaller models are great for lower-complexity tasks: summarizing, categorizing, extracting information, basic drafting.

Larger models shine when the task requires deeper reasoning: analysis, synthesis, strategy, nuanced decision-making.

That answer is most accurate in the API world, where developers can choose from a wide range of models optimized for different capabilities, speeds, and costs.

For most everyday users, today’s flagship AI models can handle the majority of common tasks reasonably well.

The bigger challenge isn’t choosing the “perfect” model. It’s learning how to work with AI effectively — use AI as an assistant, not a replacement for thinking; review and validate outputs; don’t use coding-focused models to write documentation meant for humans; and follow your organization’s AI strategy and policies.

And if your organization doesn’t yet have a clear, effective AI strategy or governance policy in place, now is the time to start that conversation.

PS: I’ll be covering some of this during my Tech Lancaster talk on Thursday, May 28 at West Art.

Originally published on LinkedIn — view the original post for comments and reactions.