'Intelligent agents' putting the smart in artificial intelligence

'Intelligent agents' putting the smart in artificial intelligence

Summary: Machines and systems equipped with their own values and objectives are bringing a new level of sophistication to a range of applications across defence, customer service and gaming.

SHARE:

Expect "intelligent agents" — artificial Intelligence (AI) systems equipped with their own sets of beliefs and life goals — to increasingly feature across areas as diverse as logistics, manufacturing, entertainment, gaming, and defence.

That's the message of veteran AI expert and RMIT research fellow Dr John Thangarajah. According to Dr Thangarajah, while artificial intelligence has been used for some time, an increasing number of application areas now require software components equipped with more sophisticated intelligence — intelligent agents.

"Intelligent agent technology is a particular way of developing software [that] has been used to develop software systems in a variety of application areas, including logistics, crisis management, holonic [flexible] manufacturing, business process management, and unmanned aerial vehicles," he told ZDNet.

"These applications often require software components that are autonomous, in the sense that they can achieve tasks by making intelligent and rational decisions where necessary without human intervention."

Dr Thangarajah said an intelligent agent approach to software development differs from traditional approaches to AI development in that systems are designed and structured in terms of "mental attitudes" — variables such as beliefs, goals, intentions, and commitments — rather than "object-oriented" methods.

"The use of such mental attitudes in software design allows developers to break complex decisions into smaller, more intuitive pieces, and hence construct systems with the appropriate behaviour," he said.

"System designs with these concepts are more readily absorbed and understood by industry experts who aren't necessarily software programmers, which makes them even more desirable."

It is just these attributes that have made intelligent agent technology attractive to everyone from games developers who are developing intelligent role-playing games to call centres for automated customer service, and now for Australia's Defence Science and Technology Organisation (DSTO).

The defence agency has called in the Agents Group at RMIT's School of Computer Science, which Dr Thangarajah is a part of, to help build its new Maritime Domain Tactical Simulation platform.

The scenario-based platform will utilise intelligent agent technology to help model tactical behaviours for Australia's submarine fleet across a range of underwater missions and situations.

RMIT's work with the DSTO, which is being funded through an AU$135,000 grant from the agency, highlights the quality of work around intelligent agents being carried out in Australia by universities such as RMIT, the University of Melbourne, and Swinburne, and companies such as Agent Oriented Software and RealThing, according to Dr Thangarajah.

"A number of Australian IT companies are either developers or users of agent technologies, and Melbourne in particular is an internationally recognised centre for agent research and development," he said.

Topics: Education, Business Intelligence, IT Innovation for Small Businesses

Tim Lohman

About Tim Lohman

Tim has written about the technology sector since the mid 2000s. He covers innovation across the business, education and government sectors.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

3 comments
Log in or register to join the discussion
  • Misnomer.

    "own beliefs and goals"....

    no - they are the beliefs and goals of the person that programmed them.
    jessepollard
    • not necessarily

      it is likely that the AI will follow the directives it was programmed with.

      There are, however, other classes of AI that don't follow that imperative. An appropriately coded AGI would be able to learn its own beliefs (in areas) based on experience, data and logic and conjecture. Autonomous robots do this run-time. Genetic algorithms with trial and error are similar to this. These AIs or AGIs learn and acquire algorithms that they were never programmed with. They were given the ability or capacity (sort of like their AI DNA) but what they ended up doing with it was specific to each individual AI.

      The intended application may dictate that the AI agent ONLY regurgitate a belief or position ---- These AIs likely won't be capable of the sort of features I'm discussing.... but that doesn't restrict the capability, in theory, of all AI agents from this functionality.

      ;0)

      Cheers
      luwo
  • Try smart agent

    Engineers are working on create a next generation smart agent uses artificial intelligence to ease the life of users.

    See www.kickborn.com , which understand 75% of english language.
    harinderpisces