Peeling the AI Onion

Whether you’re heavily involved in computer technology or not, artificial intelligence (AI) is impacting your life (and will continue to do so) in ways both big and small. However, daily discussion surrounding the technology is often long on opinion and short on explanation regarding the relationship between the different types of AI.

In an effort to rectify that situation (and within the confines of a blog) let’s take a stab at peeling away the AI onion. I’ll begin by defining the three major types of AI: Enterprise, Generative, and Large Language Models (LLMs).

The outer skin of the onion is Enterprise AI which uses AI/machine learning (ML) to extract value from diverse & extensive organizational data. Trust in data, AI models, and AI insights is a key aspect of Enterprise AI, which employs an open data lakehouse to manage various data formats across different environments. So, you may ask, “What is an “open data lakehouse?”

In the past, an organization would store structured, aggregate data used mainly for business intelligence, i.e., insights for making strategic decisions, in silos or separate data warehouses. In addition, for storing unstructured and semi-structured data in large volumes for ML workloads, an organization would use data lakes.

Now there exists the open data lakehouse. In short, it helps an organization run fast analytics on all data (both structured and unstructured) on a huge scale. It also eliminates data silos and enables data teams to collaborate on the same data using their preferred tools. They can also do this via a public cloud service, or a private one like BOXX Cloud.

Following that brief aside into open data lakehouses, let’s get back to our AI onion example as a Venn diagram. Enterprise AI is the big circle. Inside that circle, or the next layer of onion, is Generative AI. Generative AI is technology that creates new content like text, images, video, or music. It learns from existing data to produce similar, new content. Generative AI is featured primarily in art, music, and gaming and dominates most of the pop culture headlines. The fake Drake/Weeknd hit AI song that threw the musical world and fans for a loop? Generative AI. ChatGPT? Generative AI. Speaking of Chat GPT, there’s more.

Inside the Generative AI circle (which is inside the Enterprise AI circle in our aforementioned Venn diagram/onion) are Large Language Models (LLMs). If we want to get technical, that’s where you’ll actually find ChatGPT. LLMs are AI tools which can perform various writing tasks and generate human-like text. They’re trained on data and learn from it, mimicking patterns. They often produce solid results, but can also produce occasional errors, redundancy, and biases from the training data.

So to recap, you have Enterpise AI. Inside of it, Generative AI. Inside of that, LLMs. For the uninitiated, I hope this brief AI relationship primer helps, and regardless of how your organization or creative discipline is utilizing AI, always rely on BOXX.