Talk:Intelligent agent

Copyright

The figures all seem to be plagiarized from Russell and Norvig (all 4 editions; last is 2020). I doubt they are changed enough to make them public domain. It almost looks like someone trying to summarize the theme of the book. — Preceding unsigned comment added by Davpoole (talk • contribs) 18:41, 18 January 2026 (UTC)

As a definition of artificial intelligence

This section, which talks about defining artificial intelligence as the study of intelligent agents, mostly contains definitions of intelligent agents, plus a reference to one paper that defines AI this way.

I don't think this is a common definition for AI. A lot of systems of interest, e.g. computer vision systems or language models, are generally not considered agents. Additionally, the Advantages section is not really sourced, and I'm not sure it's a generally held view. MattF (talk) 20:05, 17 February 2023 (UTC)

@MattF: I agree that "one paper" would be far from reliable when it comes to the definition of AI. There are probably ten thousand such papers.

However, it's not one paper, it's the leading AI textbook, Russell and Norvig (2003). They wrote "the whole-agent view is now widely accepted in the field." (p. 50) Other textbooks written around that time also used the definition, and could be added as additional citations.

Having said that, I agree with you that it no longer seems relevant in 2023. There was a paradigm shift during the 90s and this "intelligent agent definition" was one way of formalizing the new paradigm. So, in 2003, this was still a big deal, but now, not so much.

(In my view, they key component of the shift was the adoption of formal "goals" / "objective functions" / "utility functions" / "policies". These mathematical measures were mostly unheard of in the 80s, but by the 2000s they were common. This is also part of the shift away from knowledge and reasoning and towards decision making and learning.)

Having said that, the section needs work:

• We need to update the source to Russell & Norvig 2021, and find out how far they've backed down from it or refined it.
• I would cut the two additional paragraphs, with the citations to individual papers or newspapers. I would cut them as "undue weight": they are just examples of the 10,000 papers I mentioned earlier.
• The philosophy section should be cut down to just what we find in R&N 2021. ---- CharlesTGillingham (talk) 01:04, 22 September 2023 (UTC)

Missing reference to Computer Science field

I find it really odd to see a whole page about intelligent agents and not one reference of the field behind its development, Computer Science. 2001:8A0:786B:F300:C94B:61EF:B3C6:C5EF (talk) 18:50, 7 November 2023 (UTC)

Modern meaning

Usually, in the public discourse, AI agents refer to a particularly autonomous type of AI that can achieve goals that require a series of steps. This is similar to the definition presented in the section "Alternative definitions and uses".

Should we cover more deeply this notion of agent in this article, or leave it for the article Autonomous agent? Alenoach (talk) 01:41, 24 September 2024 (UTC)

Proposed section - Agentic AI Frameworks & Standards

Frameworks and Standards

This LLM-generated text has been collapsed and should be excluded from assessments of consensus.

The following discussion has been closed. Please do not modify it.

[edit source]

As agentic AI continues to evolve, structured methodologies have emerged to help organizations evaluate, design, and govern intelligent agent-based systems. One such methodology is the AI Agent Capabilities Periodic Table (AIA CPT), developed by the Digital Twin Consortium.

The AIA CPT categorizes agent capabilities across six functional domains:

• Perception & Knowledge
• Cognition & Reasoning
• Learning & Adaptation
• Action & Execution
• Interaction & Collaboration
• Governance & Safety

It also introduces a five-level classification model of AI agents, based on autonomy and complexity:

Type	Description	Example Use Cases
0	Static Automation – Predefined responses without learning or adaptation.	Rule-based control systems
1	Conversational Agents – Basic natural language interaction with simple context handling.	Chatbots, voice assistants
2	Procedural Workflow Agents – Execute multi-step tasks using decision logic and tool integration.	Workflow automation, RPA
3	Cognitive Autonomous Agents – Plan, reason, and learn from experience to make self-directed decisions.	Predictive maintenance, optimization
4	Multi-Agent Generative Systems (MAGS) – Teams of collaborative agents capable of distributed reasoning and emergent behavior.	Industrial orchestration, smart cities

The AIA CPT framework supports both business and technical alignment, offering YAML templates, evaluation matrices, and a GitHub repository for iterative design. It is publicly available and maintained by member organizations of the Consortium.

The framework has been applied in multiple industry testbeds. One example is the Automated Negotiation Digital Twins  testbed, which explores the use of digital twins and multi-agent generative systems (MAGS) to automate complex negotiation processes across organizational boundaries. The testbed demonstrates how agents can use digital twin-based utility functions to simulate and optimize negotiation outcomes in real time, addressing challenges such as incomplete information, institutional constraints, and dynamic operational conditions. It highlights the role of agent coordination, policy adherence, and resource allocation in enabling scalable decision-making across supply chains, logistics networks, and manufacturing ecosystems.

Another application is the Digital Twins for Metal 3D Printing and Optimization testbed, which applies agentic AI principles to enhance the quality and efficiency of metal additive manufacturing. In this testbed, autonomous agents monitor sensor data streams in real time, analyze process deviations, and initiate adaptive control actions during the 3D printing of complex metal alloy parts. These agents operate within a closed-loop digital twin environment, using predictive models to continuously optimize microstructures and reduce defects without human intervention. The system demonstrates how agentic AI can support self-directed process optimization, adaptive quality control, and autonomous execution in high-precision manufacturing environments.

Wooty101 (talk) 10:23, 1 July 2025 (UTC)

Applications section

I removed the "Undue weight section" template at the beginning of the "Applications" section. The section has changed a lot since the template was added, when there was only a paragraph about self-driving cars. But the section still isn't great, so if you have an idea of what to do, feel free to discuss it here. Alenoach (talk) 02:02, 15 May 2026 (UTC)

Proposal: add book to Further reading section

Conflict of interest disclosure: I am the author of the book referenced below. Per Wikipedia:Conflict of interest, I am proposing this change here rather than editing the article directly. Proposed addition: I'd like to suggest adding the following to the "Further reading" section: * Nagasubramanian, Dhivya. Agentic AI for Engineers. Apress/Springer, 2026. ISBN 979-8-8688-2360-2. DOI: https://doi.org/10.1007/979-8-8688-2361-9 Why this is relevant: The book covers [topic — e.g., "agentic AI system design, intelligent agents,percieve-reason-act loop, safety, and enterprise deployment patterns"], which falls within the scope of this article. Independent evidence of notability/relevance: ["The book has been adopted in to 266 libraries across the country https://search.worldcat.org/title/1584648098 " I'd appreciate any feedback from editors on whether this addition is appropriate. Happy to adjust the citation format or wording as needed. Rine7986 (talk) 02:43, 20 June 2026 (UTC)