PERSONAL SOFTWARE AGENTS

Busy professionals all have them: a personal assistant who controls their schedule, makes appointments and travel arrangements, gathers information, and makes their wishes known to third parties. A good personal assistant knows what their boss likes and wants, adapts to changes, acts pro-actively, and simply makes live a little bit easier.

Artificially intelligent software systems are now able to learn and adapt, in order to carry out many of these basic 'human' tasks. And with the current technological advancements in ICT, soon everybody will have access to a smartphone which can run such sophisticated software. Almende investigates ways of using agent technology to turn the smartphone into a personal assistant.

Feedback

Software agents collect (sensor) information about their owner, and learn what their owner wants and needs. The technological challenge is to find ways of interpreting information about the world and the user, from sensor data, experience, user feedback, and other data sources. Future states must be forecasted, in order to anticipate and plan ahead.

Agents supporting people

By communicating and negotiating with other people or their agents, software agents carry out many organizational and administrative tasks on their owner's behalf, helping people to function at a much higher level. Agent technology can improve our standard of living in both our professional and private lives.

Agents can take over where people are lacking. They can monitor people with medical conditions like epilepsy, dementia, or depression; sending out an alert to the user or his social network, if his behaviour or biomedical status changes. A balance needs to be found between sharing data in order to stimulate cooperation, and protecting data to safeguard privacy and security.

Challenges:

> How do we achieve personalization?

> How should the agent interact with humans? Agents function in hypertime, but must communicate with people in realtime. How can we match the two?

> Can we make agents learn, and allow them to develop new functions?

Keywords:

> smartphones

> agent technology

> self-learning

> human-machine interaction

> modelling languages