AI is Going To Change Literally Everything

I posted a link to an article the other day about Artificial Intelligence. Thomas Helfrich was kind enough to leave a comment:

I started typing a response and the resulting comment was much too long to fit as a reply. So, here's the full response.

It's a roll of the dice at this point.

Generative Artificial Intelligence is 100% a game changer. It's the biggest technological invention since the birth of the transistor. The transistor (and eventually the integrated circuit) allowed computers to move out of the laboratory into the workplace and eventually into your pocket.

The fundamental limitation for computerized systems has been people. Applications could be developed at the speed code could be written. Writing code takes time and a lot of trial and error. Everything the software does is done because a human made a decision about how something should be handled.

We've all encountered a situation where a computer program behaves unexpectedly. It's because real life introduced a situation that the computer had not been programmed to handle.

With generative AI, we finally have programs that can understand simple directions and resolve ambiguity. This changes literally everything about what software programs can do.

When you have a system like ChatGPT that can handle a high degree of uncertainty...you have computer programs that can do brand new things entirely. Entirely new markets are going to be created.

And because we're on LinkedIn, I am absolutely going to self-promote. My company Kaizen Software released a website earlier this month called CustomLetters.ai. It uses artificial intelligence to write letters for you. The end result is usually indistinguishable from a human, and is produced in moments instead of 15 - 20 minutes per email. How much more productive could you be if you could quadruple or quintuple the speed at which you write an email?

This is going to lead to an incredible economic boom as new products and services and categories are born.

However, it's also going to be incredibly disruptive. Lots of people are going to lose their jobs. Technological innovation, throughout history, has consistently reshaped the job market, often at the expense of existing jobs and industries. Take the Industrial Revolution, for instance. It introduced machinery that replaced traditional artisan and manufacturing roles, such as weavers. This shift wasn't just a small blip; it marked a fundamental transformation in how goods were produced and how work was organized.

Then, consider the advent of personal computing. This innovation rendered typewriters virtually obsolete, impacting not only the typewriter industry but also the workforce skilled in their production and maintenance. The change was swift and irreversible.

Today, as we stand on the brink of AI and automation becoming ubiquitous, we're facing a similar scenario. These technologies, while creating new opportunities and efficiencies, are also poised to displace a significant number of jobs. This pattern isn't new; it's a well-established cycle in the evolution of technology. As we progress, it's crucial to anticipate and adapt to these changes, recognizing that they are part and parcel of technological advancement.

The roll of the dice is going to be how our government (not to mention every government in the world) handles artificial intelligence. If businesses are centered at the expense of humans, we could be entering a modern sort of Victorian Era. Massive wealth and prosperity for a few coexisting with endless misery for everyone else.

However, if our leaders center humans first and the welfare of corporations second, we could be on the cusp of entering an era of unchecked prosperity.

Everything I've mentioned, by the way, is going to happen with just the level of generative artificial intelligence we have right this second. I'm only forecasting five minutes into the future. If any further research into artificial intelligence was banned tomorrow, our world would still irrevocably change. The innovation made possible by artificial intelligence is literally just beginning.

If any type of General Artificial Intelligence is invented, it is going to be impossible to predict how the world is going to shift. I am obviously an optimist, but times of technological change are often turbulent.

Oh, by the way. Part of this response was written by an AI. Not much of it. Just a bit. It was able to emulate my tone seamlessly. This is why things are going to change. What we have now is good enough to fundamentally alter how everything works and our perception of what's possible. The only thing I can say for sure is that it's going to be a hell of a ride.

That was intended to be the end of my response, but then I had something really crazy happen. I was using ChatGPT while developing this comment, and asked it to create an image at the end. That's the image you see at the top of this post.

What's notable about it only becomes clear once you understand how image generation typically works in ChatGPT.

If I were to ask it to generate a picture of a cat, it would. For example:

So, let's evaluate the result. We have a picture of a gray and white cat taking a nap in a sunbeam. ChatGPT came up with the entire scene. Everything you see in the image was a choice made by the computer system without input from a human. It picked the angle. The focal length of the imaginary camera. The angle of the imaginary sun. The placement of every strand of fur. The exact shade of pink on the cat's nose.

The thing I want to point out, though, is the size of the original image. 1024 x 1024. Every time ChatGPT generates an image, it's that specific size because that is what it was programmed to do. Midjourney, a similar product, can only generate images in a specific size. It's a common limitation.

When it came time to generate the image at the top of this article, I used the following set of instructions:

Here is the full article. Please generate an appropriate image. It will be the headline for a LinkedIn Article, so please use appropriate dimensions.

That last bit about the appropriate dimensions was something that I wrote just for the hell of it. I enjoy throwing curveballs at ChatGPT to see how it responds. Here's the image it came up with:

I noticed that the image wasn't a square. I downloaded it, and checked the dimensions. 1792 x 1024. Which is amazing! But it's also wrong.

Which brings us to the other big thing about generative AI. It's great, but it's not yet perfect. The details are still ever so slightly off, even when the AI system is working well.

For example, In my image of the cat I had asked for a tuxedo cat. Tuxedo cats are typically short haired, but not always. While the image fits the technical definition of what I asked for, it isn't representative of reality because ChatGPT doesn't truly understand the world around it.

With the headline image, it generated at a size of 1792 x 1024 for a ratio of 1.75. When I went to upload the image, LinkedIn gave me its preferred dimensions, which are 1920 x 1080, for a ratio of 1.78.

It's not perfect, clearly. But it's good enough. While it didn't know the exact dimensions, it drew on its knowledgebase and made an educated guess. The resulting guess, while incorrect, was close enough to meet the desired goal. Computer systems that can understand imprecise human input have been the realm of science fiction for decades. They are now a reality and our world is going to change as a result.