Intelligent Agents

This module we looked at one of the most exciting and rapidly evolving fields within computing: Intelligent Agents. Agents are software programs that can understand their environment and act in a way that helps them achieve their desired goals. They are an intelligent and innovative tool that can be deployed in a whole range of scenarios across the private and public sector.

Learning Outcomes Up

  • Identify and critically analyse agent-based systems, differentiating between architectures and approaches.
  • Apply and critically evaluate intelligent agent techniques to real-world problems, particularly where technical risk and uncertainty is involved.
  • Deploy critically appropriate software tools and skills for the design and implementation of an agent-based system, bearing in mind applicable legal, social, ethical, and professional issues.
  • Systematically develop and implement the skills required to be effective member of a development team in a virtual professional environment, adopting real-life perspectives on team roles and organisation.

Collaborative Discussion Forum Summaries Up

This module includes three collaborative discussion forums. The first forum was on the topic of Agent Based Systems, The second forum was on the topic of Agent Communication Languages and The third forum was on the topic of Deep Learning. The following are summaries of the discussions that took place in each forum.
Collaborative Discussion 1: Agent Based Systems
Topic: Discuss what has led to the rise of agent-based systems and the benefits that this approach can offer to organisations. Discussion should include at least three references.

This discussion started in week one and continued to the end of week 3. I enjoyed every bit of the learning, though it was initially overwhelming as it was my first time studying intelligent agents. At the end of the discussion, I understood the motivations for and appropriate use of agent-based computing and an understanding of the main agent models in use today and their grounding in artificial intelligence research. In my initial post, I discussed the rise of agent-based systems and the benefits that this approach can offer to organisations, These systems involve autonomous agents interacting to achieve common goals. They offer organizations benefits such as increased adaptability to changing market conditions and the ability to test scenarios and adjust strategies accordingly. Agent-based systems leverage distributed computing resources, enabling organizations to streamline operations and improve collaboration. They excel in handling uncertain and incomplete information, making them valuable in manufacturing tasks, from design to supply chain management. Their use leads to a flexible, efficient, and competitive approach in today's complex business environments, improving responsiveness, reducing costs, and enhancing decision-making.
Collaborative Discussion 2: Agent Communication Languages
Topic: What are the potential advantages and disadvantages of the use of agent communication languages such as KQML? How do they compare with method invocation in Python or Java?

This discussion started in week five and lasted for three weeks. It focuses on discovering the potential pros and cons of KQML and how they compare to method invocation in Python or Java. At the end of the discussion, I understood the main agent models in use today and their grounding in artificial intelligence research, I gained the knowledge and skills required to develop, deploy and evaluate the tools and techniques of intelligent systems to solve real-world problems, and understanding of contemporary research issues in the area of intelligent agent systems. Agent-oriented programming emphasizes organizing software designs around agents and their communication, using an agent communication language (ACL) for effective information exchange. ACL allows sharing intricate knowledge, including plans, agent goals, and beliefs, not easily possible in traditional object-oriented methods. Method invocation in Object-Oriented Programming (OOP) enables communication between objects, dividing software applications into smaller objects. KQML provides a well-designed communication language with readable syntax and extensibility but suffers from complexity, a steep learning curve, limited tooling, performance overhead, restricted communication environment, unclear processing, receiving, and dependency on pre-established permissions.
Collaborative Discussion 3: Deep Learning
Topic: The advent of new technologies supported by Deep Learning models mean that it is now possible to generate ‘new’ content, for example, Dall-E AI to generate images or ChatGPT to create prose.
Do you think that these new technologies offer any ethical issues that should be considered, and if not, why not?

I love every part of this discussion, from Feaviour (2023)'s view on accuracy and content ownership as part of core ethical issues to Anastasia (2023)'s conversation about how The EU AI Act is pushing boundaries in regulating and forming policies that could mitigate most of these ethical issues. The growing excitement around generative AI requires careful consideration of potential risks and ethical concerns. Misuse, as seen in deepfake voice technology, highlights the need for safeguards to prevent the creation of fake and unethical content. Future research should address uncertainty, explainability, biases, and environmental impact. Organizations should implement robust risk management and governance measures to leverage generative AI safely and confidently while aligning with their risk appetite. (Lee, M. & Kruger, L. 2023).

Outcomes from the Team Exercises Up

The development team projects in this particular module were in two parts covering the development report and the presentation of the ideas from the report paper in part one.

It was also a great experience working with other students in the program, exchanging implementation approaches and delegating tasks based on each others' strengths. Teamwork also comes with its challenges as some members of the team are very opinionated and hold strong views on their opinions. Still, after a few meetings, I understood everyone's temperaments, and this helped me to be able to play a better team player.

The instructor added some feedback on the submitted report, which the team further sought clarity on and was able to amend those areas in the second team project.

At the end of the team projects, I could identify and critically analyse agent-based systems, differentiating between architectures and approaches. Apply and critically evaluate intelligent agent techniques to real-world problems, particularly where technical risk and uncertainty are involved. Deploy critically appropriate software tools and skills for designing and implementing an agent-based system, bearing in mind applicable legal, social, ethical, and professional issues. Systematically develop and implement the necessary skills to be an influential development team member in a virtual professional environment, adopting real-life perspectives on team roles and organisation.

View Project Brief View Design Proposal Document View Design Presentation Transcript View Design Presentation View Project Codes

Reflections Up

The Intelligent Agent module has been an exciting and challenging learning experience as an MSc student in AI. Throughout the module, I have acquired essential knowledge and skills related to agent-based computing, agent models, development, deployment, and evaluation of intelligent systems. In this reflection, I will elaborate on my journey through the module, highlighting critical aspects of the project, my emotional responses and analysis, learning and changed actions, evidence of skills and knowledge development, collaboration within the group project, and the accurate use of citations and references. This reflection will also include my thoughts on intelligent agents, emphasizing the skills and knowledge I gained during this module and my final project evaluation.

Unit 1: Introduction to Agent-Based Computing

In unit one, I was excited to learn about the introduction to Agent-Based Computing as this is my first exposure to intelligent agents in AI concepts. I went through the unit's lecture cast that focuses on the drivers of growth of intelligent computing with a specific focus on agent-based approaches. The unit's reading was quite enjoyable. It is chapter 2 of Wooldridge, M.J. (2009) An Introduction to Multiagent Systems. Chichester: John Wiley & Sons.

The first collaborative discussion was about agent-based systems; I posted the initial post and made a few peer responses. It was a fascinating experience to do some research on a new topic and write about it.

My key learning in this unit is the concept of Multiagent Systems and adjustable autonomy.

The idea behind adjustable autonomy is that when certain conditions are met then control is handed back to the operator.

I need to do further research on Multiagent and adjustable autonomy as they both form the bedrock of everything in this module.

My emotional response to this unit is excitement and curiosity. I am excited to learn about a new topic and curious to know more about it.

Unit 2: Introducing First Order Logic

This unit was also exciting as we covered key terms and symbols used in first-order logic, the relationship between first-order logic and natural language, and quantifiers in first-order logic. It was also my first time getting exposed to this topic, so it was challenging to understand initially, but I was able to wrap my head around it.

I could not attend the seminar for the week as it was scheduled for working hours and conflicted with some meetings at work, but I watched the recording. I read Chapter 8 of Russell, S. & Norvig, P. (2021) Artificial Intelligence: A Modern Approach. Harlow: Pearson Education. It was a lot to take in and a bit challenging to understand.

The first collaborative discussion on agent-based systems continued, and I made a few peer responses. It was insightful to read other students' initial posts and learn from the outcome of their research.
The project groups have been made this week. My group consists of Ellena & Kazuma

The key concept that stood out for me in this unit is Ontological commitment and Epistemological commitment as they relate to languages and first-order logic.

I need to do further research on first logic order, language commitment, and ensure I reply with "Peer Response" to initial Post as I made mistake of not adding headings to my Peer Responses

My emotional response to this unit is still excitement and curiosity. I am excited to learn about a new topic and curious to know more about it.

Unit 3: Agent Architectures

Unit three readings were challenging, too, as they covered the range of agent architecture and understanding the difference between intentions and desires. Honestly, the whole agent topic is new, and I am working hard to understand the concept.

All of the readings for this unit are old papers and a lot, and I could not read them all.

The first collaborative discussion on agent-based systems continued. I added my summary post, which is a follow-up to my initial post and some additional research and insight I got from the peer reviews. And Ana joined our team we started working on Team Project. We had our first call and created a whatsapp group for smoother communication, we decided on who does what and agreed on a project topic.

Unit 4: Hybrid Agent Architectures

This unit was easy as the reading was pretty straightforward, and I am getting the whole agent's high-level concepts.And I also could not attend this week seminar because of it clashes with office meetings

I read chapter five of Wooldridge, M.J. (2009) An Introduction to Multiagent Systems. Chichester: John Wiley & Sons.

The critical concept that stood out for me in this unit is the hybrid agent mind mapping structure on page 104 of the unit's text.

Unit 5: Agent Communication

By week 5, I am getting comfortable with understanding how agent works and commonly use architectures, this week focused on agent communication, the concepts of speech acts and speech act theory, concepts of ontologies and their use in agent communication. And also introduce the notion of agent communication languages.

I read Payne, T.R. & Tamma, V. (2014) Negotiating over ontological correspondences with asymmetric and incomplete knowledge. AAMAS 13(1): 517-524. and Searle, J.R. (1969) Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. but the speech acts was not enjoyable reading for me.

The second collaborative discussion on Agent Communication Languages started. I added my initial post.

My emotional response to this unit is still excitement and curiosity. I am excited to learn about a new topic and curious to know more about it.

View Initial Post
Unit 6: Working Together

This unit was quite challenging as we needed to submit the project work for grading, I volunteered to write all the codes for the project, and the other team members could work on the report and general team management. I was in a bit of a rush as everyone was working and needed to create time out of their busy schedule to get the project done, communication was a bit of an issue in the beginning with the team as everyone's temperament is quite different, but we scaled through that and collaborated effectively for the project success.

I skipped all the readings for this week to focus on the team project deliverable.

My emotional response to this unit is exhausting. There are to do to submit the team project.

View Project Brief View Design Proposal Document View Project Codes
Unit 7 & 8: Natural Language Processing (NLP)

This unit looks interesting as I have prior knowledge of the topic, Natural Language Processing (NLP) and am interested in learning further concepts and principles of NLP, different approaches that can be used in NLP and how to develop and support NLP.

The lecture cast was excellent and beneficial, but the question at the end needed to be better structured. I could not read all the recommended readings for this unit.

The second collaborative discussion on Agent Communication Languages continued. I added a few Peer Responses and couldn't prepare for the seminar because I won't be able to attend due to the timing.

What stood out for me in this unit is that the outcome of what we see currently started a long time ago from the days of the Turing test. The result we now witness with generative AIs powered by language models is fascinating.

I will further research NLP and how it has helped facilitate groundbreaking products like ChatGPT.

My emotional response to this unit is exhausting too. There are a lot to read and research on NLP

View Peer Response Click to View Summery Post
Unit 9: Introduction to Adaptive Algorithms

The focus of this unit was Adaptive Algorithms such as Artificial Neural Networks (ANNs) and Deep Learning; It was a great unit as these algorithm has formed bases for insightful AI research that are pushing boundaries in the industry today.

The lecture cast was excellent and beneficial, but the question at the end needed to be better structured. I read all the recommended readings for this unit, and the attached video of Andrew Ng was quite interesting.

The last collaborative discussion on deep learning started. I added my initial post and couldn't prepare for the seminar because I won't be able to attend due to the timing.

I enjoyed every part of this unit from how AAN could be used to identify images, handwriting to the core problem of ANN

View Initial Post
Unit 10: Deep Learning in Action

This is a fascinating unit looking at the actions of deep learning and what has been built using the technology in the present business environment, and companies that have sprung off from the bases of the technology are intriguing to study.

In the unit, I read the World economic forum article on how deep learning can improve productivity and boost business. This is an absolute great read, the insight around the numbers of data generated in our world today is breathtaking.

The last collaborative discussion on deep learning started. I added my initial post and couldn't prepare for the seminar because I won't be able to attend due to the timing.

This week, the amount of data generated stood out for me, as mentioned in the World Economic Forum article. it's estimated that the data we generate every day is 2.6 quintillion bytes

I will further research an application of Deep Learning.

The is an exciting unit and I totally enjoyed it.

View Peer Response
Unit 11: Intelligent Agents in Action

This is a busy week. I need to finish the summary post, and the team must work together to submit the project presentation work. This week we are focused on how these technologies are currently is used in practice in manufacturing and Industry 4.0.

I skipped the recommended reading for this unit to focus on the summary post and team project work. I watched the lecture cast that explained Industry 4.0 and how AI is helping to transform manufacturing and improving efficiency.

The last collaborative discussion on deep learning continued. I added my summary post and couldn't prepare for the seminar because I won't be able to attend due to the timing.

I will further research Industry 4.0

As much as I enjoyed everything industry 4.0 and deep learning, this unit was quite busy and exhausting with all the assigment and post that are due.

View Summery Post View Design Presentation View Project Codes
Unit 12: The Future of Intelligent Agents

This is the last unit in the module. It's pretty emotional, as the learning in this module has been incredible and insightful. The focus for this unit is bringing everything that has been learnt in the module together. We Explore the future directions of intelligent technologies, Discuss the potential ramifications of these developments, and Consider how technology will evolve in the coming years.

I read Nasim, S.F., Ali, M.R. & Kulsoom, U. (2022) Artificial Intelligence Incidents & Ethics A Narrative Review. IJTIM 2(2): 52-64.

I started working on my Individual e-Portfolio SubmissionAssignment

The Intelligent Agent module has been an enriching experience that has equipped me with valuable skills and knowledge in agent-based computing and intelligent systems. The project provided practical insights into building real-world IA applications. The module has allowed me to identify areas of growth based on the tutor's feedback and interactions with my peers.

Team Meeting Notes Up

Most of the team meeting happened over whatsapp calls and group messages, I skipped most of the seminar because it conflicts with most meeting in my office, but I watched all the recordings.

Feedback from peers and tutors Up

The tutor provided insightful feedback on the group project.

A nicely presented report that covers the problem in a good level of detail. You have done a really nice job on presenting the architecture and key challenges. It might be worth reviewing some of the UML diagrams, particularly the sequence diagram. The sequence diagram contains a lot of back and forth messages between the various agents. Remember that this diagram aims to show the messages passed rather than the logical flow.
The presentation is clear and accessible and demonstrates a good approach to delivering your work. You have made good use of the slides to support your points, although some images (particularly the code snippets) are a bit difficult to read properly. The style and performance of the presentation are very good and show that you are comfortable with the material. The code is presented inline with the normal Python standards and is fairly readable. There is a lack of documentation or comments within the code itself making it difficult to fully understand what you have developed and why, at times. Remember it is important to document your code with appropriate comments where possible so that you can remember what you were doing in the future but also to enable further contributors. It was hard to properly gauge what the aim of the software was in places and I think you needed to be more explicit about how the different components function. There is a lack of testing within your solution and I would have expected to see some evidence of validating the software and ensuring you are getting the right results.

Professional Skills matrix and Action Plan (PDP) Up

I will continue to develop my Machine Learning and Deep Learning algorithm skills, I considered myself proficient but still need more study and research which I will be doing with reference to online portals like Udacity, and Coursera.