The world of human communication underwent a profound transformation with the mid-15th century development of Johannes Gutenberg's printing press. In our 21st century, the precipitate developments in generative AI are expected to once again change the paradigm for human communication.
It is in our best interests to get to know the rapidly burgeoning communication technology stemming from LLMs (large language models), so we are well-positioned to both use it to our advantage, and be alert to staunch its misuse.
This guide is here to help you do that.
AI: Turning Point
AI has been a familiar field in the world of computer science for decades. However, the release of OpenAI's chatGPT on November 30, 2022 marked a turning point. A degree in computer science was no longer necessary given the generative abilities of chatbots to process everything from natural language, to images, software codes, and audio data.
AI: Some Terms
Let us get familiar with some of the basic vocabulary inherent to AI:
LLM - Large Language Model
SLM - Small Language Model
Multimodal AI - LLMs that can seamlessly process text, video, images, and audio.
Chatbot - a software application aimed to mimic human communication
ChatGPT - Chatbot built on LLM by company called OpenAI, currently on its 4th iteration; most well-known tool of A.I.
GPT - Generative Pretrained Transformer; i.e.:
G - Generates a seemingly endless supply of new content on demand
P - Pretrained to generate LLM content from billions of words, based on regularly repeated sequential word usage.
Hence, strings of words are predicted on the basis of statistical regularities rather than on most appropriate or 'good' answer.
T- 'Transformer' is the substratal network architecture developed by Google to process natural language
Hallucinations - A controversial term used to suggest the recurring incidence of LLM's 'fabricating information entirely, behaving as if they are spouting facts.'
Prompts - Prompts are carefully designed instructions aimed at eliciting the best answers from AI generated information.
 |
 |
 |
 |
 |