AI is smart, but it can't survive without humansAug 21, 2023 01:59PM ● By Adam Cochran
Two months ago, I decided to use this column to write about the fundamentals of artificial intelligence. I knew the topic was too big to cover in a single column, so I broke it down into three columns.
In July, I described what AI is and what it is not. In August, I wrote about the myths and realities of how AI is replacing humans in some jobs and processes.
This month, I will conclude by explaining, on a fundamental level, how AI works and how it can be used as a tool in day-to-day life.
“Is it free?”
No. Tools used for AI-assisted content generation hardly use any power from the computer used to make the request. Instead, the request is sent to a network of computers that work together to compile the image, music, prose or formula based on the variables provided.
I will use one of my favorite AI creative tools, Midjourney, as an example. Midjourney creates images based on prompts.
If I ask Midjourney to generate a realistic 1940s photo of a stampede of giant octopi running down the road of the Colorado National Monument, I get something like this:
Or, if I ask for a candid photo of a group of male and female people over 50 reading the BEACON at the top of Pikes Peak, I get something like this:
AI does not create anything from scratch. It does not have an imagination or skills. Computers can only do three things: remember, calculate and organize stuff. The power comes from the speed of calculation, the amount of data it can hold, and how quickly it can access that data.
If I were to ask a human artist to create similar images, they would gather source photos, apply learned concepts and techniques of their given craft, and compile an image using all available resources. However, he or she would be limited by whatever knowledge and resources they had acquired over their lifetime.
AI uses the same process. Only, instead of using the knowledge acquired over a lifetime from a limited array of sources, AI can essentially access all of the world’s data and convert, organize it, compute common formulas and create something new using what it “learns.”
By the way, a Midjourney subscription starts at $10 per month.
There are two big differences between human and AI content creation. Computers don’t have the ability to break the rules and they can’t interpret an original problem. They have to be told what the problem is.
In order for a human to use AI effectively, he or she must know how to break the problem down into something the computer can convert to a formula and produce a result that addresses that formula.
For example, asking AI to produce a beautiful tree in a field of flowers will result in something like this:
Asking it to produce a green and yellow logo for a landscaping business produces something like this:
There is a popular argument that AI is taking over jobs and human ingenuity. But in most cases, AI is augmenting imagination, increasing efficiency and increasing productivity. AI doesn’t generally create final drafts or end results. It creates working conceptualized drafts of human suggestions which will be adjusted, corrected and refined using human creativity and flexibility.
While AI is going to replace a lot of jobs currently filled by humans who have mastered the fundamentals of the given craft (writing, art, document creation, etc.), it will not replace jobs or specialists who understand the subjective aspect of creativity and ingenuity.
Humans and AI each produce original work by combining finite derivatives of content that already exists. Illustrators work from reference photos, photographers use posing and lighting charts, and writers use every written work they have ever read as a reference.
The difference between AI and human conceptual work is that AI will always have to be told by a human what is acceptable, beautiful or essential in meeting the specific need that it is being used to resolve.
You might also like: