Think Small: Why Big AI is Not the Answer

Small AI is a big idea.💡

In a world where many are running to big models, big GPUs, and bragging about how many trillions of parameters they have - what if the answer for most teams is actually quite...small?

When launching Savvi AI several years back, I wrote an article published in TechCrunch about "Small AI." It was a response to what we, as a former Fortune 500 AI team, saw as artificially high requirements taking over the AI space—large teams of specialists, big data, big infrastructure, and even bigger cost and complexity (https://lnkd.in/dQdhBdZ)

But what if the answer for many problems isn't to think bigger but smaller—narrower, purpose-built solutions that can be tuned to particular business problems?

Fast-forward to 2024, and Axios highlights that many AI-thinkers are coming around to "Small AI" - small models, smaller data sets, smaller purpose-driven use cases, and smaller resource requirements (https://lnkd.in/gQnyv5jW)

Like microservices replaced monolithic applications, Small AI will produce better results in many practical use cases than big-tries-to-solve-everything-for-everyone-all-at-once AI.

Think SMALL.