Generative Artificial Intelligence is a program that can create “new” content by using and referencing existing material. These are programs that “listen” to songs, “read” articles, and “see” art and then create a new piece of material based on the query posed to it. It is important to note that generative artificial intelligence does not generate new ideas or work. Instead, it uses information derived from existing works (often many) to find the average or most common pathway to create the content asked of it.
Let’s say that you ask 100 students to write a paper on the Mona Lisa. You then take those 100 papers and feed them to a generative A.I. to write its own paper about the Mona Lisa. The algorithms will look for common connections and the probability of those connections. The output will thus be a paper that is the “average” of the collective work that was put into it. However, the more input the more machine learning and more connections it makes. As certain words and phrases are used in different ways by different groups of people, the AI can detect and respond to these distinctions. For example, the language in a business office would look wildly different than the language used in a medical environment. A generative A.I. learns and makes connections based of large and small “ecosystems” of the content that it is evaluating and using to create tailored content.
Generative AI also can evaluate and improve upon the work they create and recommend improvements on work we create. This is done through further queries added or user input. Asking a generative AI to create an essay followed by requests to edit it to remove or include specific items is all possible. Art can be created, then be asked to add further clarity, color, and details to existing components. Refining comes from the knowledge, imagination, and skill of the user to create queries, analyze the results, and alter the content with respect to the power and limitations of the generative AI being used. Essentially requesting the A.I. look at the specific components, strengthen and enhance those connections, and from the new output components more details are generated.
Topics