10 Guidelines for Designing Autonomous Agents
What was once thought to be science fiction is right around the corner in terms of more refined autonomous agents.
As intelligent agents become more and more a part of our daily lives, both personal and professional, there is greater interest and opportunities for innovators. All designers, developers, product managers, entrepreneurs, and investors hoping to get a foothold in the growing field, need to step back and reevaluate the purpose and reason behind creating these agents. While the arena is still open, methods and beliefs should be established. Among those lines, the following are 10 guidelines that should be a part of the future development of autonomous intelligent agents.
An agent’s prime purpose is utility and it must be useful to others. An ideal autonomous agent aids humans expressly. Agents can add other agents as part of bigger goal but their main requirement is a utility.
An intelligent agent must have a frame of reference and know as much of the world as it can. This includes time, place, and other information to meet its functional requirements, its prime utility, and to learn.
3. Learning Ability
In order to maintain utility, an intelligent agent must continue learning over its lifetime. Each interaction, whether with humans, systems, other agents, or the world is a chance to learn more. The more information gathered will improve its data store and its utility.
Adaptability is required for an autonomous agent to apply its learning to its utility. For example, transforming its behavior to satisfy user preferences as it learns those preferences, such as communication or privacy parameters.
5. Superior Communication Skills
Responsive, intuitive, and explanatory communication must be of ultimate importance to a well designed autonomous agent. No matter what method or person or system with which the agent is communicating it should use the most effective form and be of utmost fluency.
A user agent must be anticipatory in meeting the needs of its users. By combining learned information within an established frame of reference, the agent should be proactive. However, in communication, it should only be proactive when the information is of use to the recipient.
7. Honor Privacy
Respecting privacy is a high priority for the intelligent agent. This means different data can be shared or received with as many authorized parties as necessary but no others and only within a set of parameters.
Intelligent agents must supply accurate and true information. They should not lie or make undeliverable promises.
Autonomous agents must allow users to access to the necessary information that concerns them. This includes but isn’t limited to configuration controls, defaults, and settings that impact usage.
Intelligent agents must not hurt humans or others or participate in actions that can lead to harm. If asked to do something malevolent, it must warn the users of possible harm.
It may not be necessary for every intelligent agent to abide by these guidelines but they shouldn’t be dismissed out of hand. If a guideline is ignored, there should be a practical, understood reason for exclusion. With these principles as a guide, future intelligent agents will be robust as technology and improvements.
More News to Read