🧐 Interactive AI
One of the big issues with AI, is that it acts as a blackbox. We don't know exactly what information it is using and how.
Human in the Loop learning is a solution to this type of challenge. It allows a moderator to have some influence on how the AI acts. This can be with fine-tuning, or feedback mechanisms built into a model, like Chat GPT.
Big Red Button Automation
Big Red Button Automations are ones that occur completely autonomously. They offer humans very little control, and work on certian types of systems. These systems have their place, but in places that you and I probably don't think about.
Automation and AI should help humans work on the tasks that they are good at. Big Red Button automation should be done up to a point. If you generate the same report every day, an automation should do that... However, a human is the one that should be making the decisions based on this report.
Designing better systems
When we design systems with humans in mind we think about selective participation rather than removal. We look to build smarter systems instead of removing humans from them.
As an example, if we could build an AI that takes legal documents and simplifies them for the reader. Maybe we could even change the level of jargon with a slider. This, is using AI to empower a Human by keeping them in the loop. The power lies in allowing the human to mix complexity and simplicity to their own level of desire.
Reinforcement learning
Reinforcement learning is a way for a computer to learn how to do things on its own. It is like a game where the computer gets points for making the right play and loses points for making a mistake.
The computer gets to learn from its own mistakes, and gets better and better at the game or task. It's like when you learn how to ride a bike, at first you might fall a lot but you get better and better at it each time you try.
Imagine an AI assistant recommends a restaurant. You give it feedback, and it immediately gives you three other places you will love.