The society today is roughly divided into 3 camps, those who are terrified that AI will replace them in the workforce, those who find AI useful but think that it will never be “smart enough” to actually replace most people in the workforce and lastly, those who don’t know much about Ai and doesn’t care either way.
This is my prediction on how AI will be integrated in our lives, and how it could replace a lot of us in the workforce even without “knowing” everything within every specific area.
The current state of AI
At this time a large part of the population has access to at least two types of AI.
- Large Language Models (LLM)
- Diffusion Models
Independent research like this is self-funded. If this guide saved you hours of troubleshooting, consider fueling the lab. You can now make completely secure one-time or monthly contributions through Google.
When you run a LLM such as ChatGPT, Gemini or Copilot, you can ask a variety of questions about pretty much anything. These LLM models have enough information to give at least general (more or less accurate) information to almost anything you ask them.
The Diffusion models are the ones we typically use to create visual images or videos, and includes models such as Stable Diffusion, Flux and Red Panda. Here you provide commands in text, images or both to make the model understand how you want the image or video to look like.
To make things easier for the user, and make it more clear to the diffusion model what we actually mean, we can easily incorporate an LLM to translate our instructions to a more “diffusion friendly” command.
Let me show you two examples:
In my first example I am giving the basic text instructions to the diffusion model, and it will in return present me with an image that reflects my instructions.
In my second example I will give the same text command to an LLM model. The LLM model will interpret my command and rewrite it in a more optimized way to make it easier for the diffusion model to understand what I want.
To incorporate the LLM model into my work with the diffusion model is easily done with the help of API. While I could learn how to best write instructions for the diffusion model myself, it would require both a lot of knowledge and time. And since the diffusion models constantly are being upgraded and the optimized way to interact with them changes as well, it would be stupid of me to not utilize the tools at hand instead.
My prediction
This is how I imagine that AI will be integrated in our everyday lives in the near future. We will not have one single AI that will have all the answers and that may or may not have the knowledge to replace me or you in the workforce. Instead we will all have access to a free chat AI that knows how to communicate with a plethora of other, more specialized AI’s, that are trained specifically for one purpose. That purpose might be anything you can think of, from legal business to farming to computer science and programming.
The illustration above shows a rough sketch of how I imagine it will work. The user tells the interactive AI (a regular LLM model) that they wish to create an advertisement for a product. The interactive AI will then send requests to one or several other AI’s that are specialized in different things, such as “advertising”, “graphic design”, “demographic research” and “financial analysis”. If needed, these specialized AI’s will request information from even more narrowly specialized AI’s, and communicate the results back to the first interactive AI.
The interactive AI will then take the information it got from the specialized AI’s and make it into something more comprehensive for the user that made the initial request. Although it might seem like the whole process is a bit complicated, and that it might take a lot of time from the initial request until the user has a complete and final answer, I predict that the whole process will take somewhere between 30-120 seconds in the current state. 30 seconds if the answer only requires text, and up to 120 seconds if it also requires graphics in the answer.
I believe that it will not be long until AI developers start training these types of specialized models, and start selling API subscriptions to users that want access to them. Likely you will have an interface program, which connects to the interactive AI, installed on your computer. The interactive AI will be free to use, but be limited in what it can do. Upgrades to more specialized AI will be available through API, and will probably be aimed mostly towards small and medium businesses at the initial phase.
All the AI related work I do, I do on my spare time and most of it I share with the world completely free of charge. It does take up a lot of my time, as well as the cost for running this website. I’m grateful for every bit of support I can get from users like you.