Human-in-the-loop (HITL) is a model of interaction where a machine process or system is designed to involve a human input. Although systems and algorithms can perform autonomous operations, they also incorporate a human element at specific stages to make crucial decisions, provide intelligence, or verify outcomes.
Most often, this term is found in fields such as machine learning, artificial intelligence, and robotic process automation.
Human-in-the-loop in practice
In a human-in-the-loop system, the process generally begins with the machine carrying out a task or operation up to a certain stage. When it reaches the point where the human input is required, the process pauses. The human then steps in and performs their part of the task, such as providing a decision, correcting a flaw, or giving some specific input that the machine can't generate on its own.
Once the human completes their input, the process resumes, and the machine takes the output from the human and carries on with the remaining operations. This method is often used in situations where machines are not yet capable of accurately performing the entire process on their own, or where human judgment, oversight, or approval is considered necessary or desirable.
An example of human-in-the-loop can be seen in semi-autonomous vehicles, where the vehicle can drive itself under certain conditions, but human intervention is needed for complex decisions. Similarly, in machine learning and AI, human-in-the-loop processes are used in training and fine-tuning algorithms, where a human expert reviews and corrects the outputs of an algorithm to improve its accuracy.
Download this guide to delve into the most common LLM security risks and ways to mitigate them.
Lakera Guard protects your LLM applications from cybersecurity risks with a single line of code. Get started in minutes. Become stronger every day.
Several people are typing about AI/ML security. Come join us and 1000+ others in a chat that’s thoroughly SFW.