Start Here Second: Some Simplified AI Concepts

There are many definitions of AI.

There are so many different definitions of AI it's hard to know what's what. Often, the simplest definition offered is a variation on, "Artificial intelligence (AI) is the ability of a computer or robot to perform tasks that are usually associated with human intellectual processes." While accurate, it's not necessarily helpful for understanding how it works and more importantly, how to make it work in the communications function. Below are terms and explanations that try to solve for definitional utility.

AI - Extraordinary Lists of Instructions: If we consider computer code as a complex collection of instructions we give to computers to execute, we can also think of computer code as alogrithms. Long, complex lists of instructions for how to accomplish a certain task or a series of tasks.

When alogrithms/code evolved to a point so complex that computers could solve problems almost like humans, we gave alogrithms/code a name of its own, "Artificial Intelligence." So, in its absolute most simple form AI is just the most recent, and extrordinarly powerful, name we use to describe a very complex list of instructions we give computers to execute.

How many types of AI are there? The Gartner Hype Cycle image shows us there are more than

25 types of AI currently tracked by the company. How is each type of AI different? The math formulas/algorithms/computer code for each type of AI are designed to do different things. But if we look at all of AI as just one kind of technology, AI can be considered the next evolution of software. It is software engineered to such an extraordinarily complex and sophisticated level that, today, the software can improve on its own over time ("learn") and continue getting better as the code keeps running and processing data in different ways.

Generative AI: In the communications function we are almost exclusively interested in Generative AI. Generative AI code is designed to scan and analyze terabytes of data from public and private sources of information - the web, government sites, company sites, etc. To oversimplify again, the list of instructions within the Generative AI code compares every document, every paragraph, every sentence, and every letter against all other data sources it has been designed to scan. This is what ChatGPT and Google Gemini does. They started life as text-only tools but today they generate images as well as text. There are many other AI tools that work exclusively with video, images, and/or audio. For now, let's stick with the two most popular AI resources. Generating video and other forms of content is incredibly powerful using dedicated systems and they will be a lot of fun to learn.

As users, when we log into one of the text-based generative AI sites like ChatGPT or Gemini, we can fill in the search bar with a request to "create an essay on the history of fashion in the US, 1800-1900." Because the AI code used by ChatGPT has already analyzed and compared every available information source covering the topic of US fashion, the AI formula can "generate" a fairly good essay based on what has been written in the past. An essay is generated by the ChatGPT software because it has been instructed to identify patterns across all of the information it has scanned and then "predict" what letter, word, sentence is likely to come after the letter, word, sentence in front of it.

While we humans can potentially read fifty articles and essays to draw on for an essay we write, text-based generative AI sites can draw upon millions of information sources to produce recommended text. Therein lies one of the central powers of AI code/technology - scan vast amounts of information, identify patterns, predict what will come next based on those patterns, continually improve predictions over time without human intervention.

While ChatGPT is the most popular generator of text output, there

are other companies offering the ability to

generate images, audio, video, code, and other types of output. Each operates essentially the same - analyze billions of data sources and generate content by drawing on all the available information published on the topic as source material.

Each of the blog posts following this one will build on the general concept of AI outlined above. If you have points of confusion along the way, please let me know where you're getting stuck so the content can be improved to help others starting their AI learning curve.