Generative AI vs. Machine Learning

It is important to understand the difference between machine learning and generative AI. Tommi Jaakkola, the Thomas Siebel Professor of Electrical Engineering and Computer Science at MIT states, “The major distinction here is in terms of the complexity of objects we can generate at the scale at which we can train these models.” Essentially, machine learning is often classified in the same category of generative AI. 

It’s not wrong. But…there is still an important difference. 

Siebel’s explanation focuses on the main distinction: data, its amount of training, and the data it can furthermore retain. Let’s break it down. The computer is retaining a data set with let’s say 100,000 parameters. Therefore, after the computer retains and learns this set, it can spit out information inside of these parameters. This is often machine learning. Now, let’s say you want to provide more humanlike and accurate answers. You decide to multiply the parameter amount in a new data set by 1,000. You now have 100 million parameters your computer is now learning. As a result, it will give more specific and tailored answers based on the input the user is putting in. This result is similar to what we know as ChatGpt, a version of Generative AI. I don’t want to sell this short. A platform like Chat GPT is nowhere close to 100 million parameters in its data sets. There’s billions, most likely trillions making generative AI, machine learning on an insanely larger scale, giving humanlike outputs versus probable answers. 

They also have distinct purposes.

According to the article in Forbes, "The Vital Difference Between Machine Learning and Generative AI", they explain the two AI advancements' differences using a P.O.A format. (P stands for purpose, O for output, and A for applications. Below is a simplified version of my readings. (ML-Machine learning, GA- Generative AI

ML: Purpose-focus on comprehending and predicting using existing data as a framework.

GA: Purpose-focus on creating new data that mirrors human ideas.

ML: Output-produces choices and predictions.

GA: Output- produces new content (images, music, text).

ML: Application-used for tasks such as recommendation systems, predicted data, and identifying tools.

GA: Application- job is to generate creative domains, advanced simulations, and deepfakes (images, audio, or videos mirroring humans and their personalities). 

To sum it up, generative AI is the next generation of machine learning. It will keep growing more humanlike and create superior individuality in responses that machine learning will never have the capacity to do. Generative AI is the baby that will grow up to keep learning, and Machine Learning is two generations before. It will stay that way while Generative AI soars into new discoveries as Machine Learning will attempt to keep up, but never will. That being said, both are still valid tools that you can use depending on your goals. 

For more intricate work and predicted analytics, Machine Learning is definitely your match. For human-like responses and produced media, Generative AI is the way to go. Each have their own pros, but in the future, Generative AI will be much more relevant in our world (damaging or useful) due to its capacity to build personal responses and mirror or exceed human personality and knowledge.

*Note: If you are curious about my take on AI being damaging or useful, see "Get it Before it's Gone" for all opinionated posts.