Hello Group, hope you are all doing great. I was reading this article by mc kinsey about the value of generative AI and it led me to a potential research of understanding more about generative AI and how it is being differentiated from traditional AI or predictive AI… What I feel even as generative AI has its benefits so as predictive AI, the true power lies in Hybrid AI, whereas both generative AI and predictive AI works together. For example in case of weather predictions, generative AI can generate weather simulations on various parameters based on the input data which can help predictive models be more effective when encountered with real time data to predict. My question is 1. Am I correct to think Hybrid is the way to go 2. why there is so much hype about generative AI whereas its only one side of the coin… thanks saikat
-
There is no single “way to go”. You have to pick the right tools for each situation.
-
Very powerful Generative tools are suddenly available for general use off-the-shelf. This has caused a great deal of interest and publicity.
Thanks for the reply !!
- Agree there is no single way to go, whatever we choose will have pros and cons.
- when you say powerful generative tools, you mean to say they would be able to do both predict and generate at the same time ?.. My understanding is we will have a foundational model which can do generative tasks as well as predictive tasks via transformers attuned for a particular task. My question is, in AI I think there is a huge opportunity & benefits from predicitive AI alongwith generative AI, So why the Hype around Generative AI only ?
Sorry, I’m not aware of this concept.
I think Megatron-Turing NLG is an example where a foundation model can handle both predictive and generative tasks and keep on learning from each other and otherwise via emergent behaviour. Ideally today using separate foundation models for predictive and generative tasks is more common due to the potential for better performance and interpretability.
But my query is why the big hype for Generative AI only when we know predictive Ai is equally important ? so should we call it Hybrid AI is the way to go ?