feat(lit): removed old notes

Author: TomMonks

notebooks/00_introduction/03_literature.md

@@ -100,83 +100,6 @@ Despite the limited body of research, these initial investigations suggest a pot
 Further research is needed to explore the integration of generative AI across a wider range of simulation paradigms and to develop robust frameworks for human-AI collaboration in the simulation development process.
 
 
-
-{cite:p}`Akhavan_2024`
-
-
-
-
-## Notes
-
-
-title brainstorming:
->"Using LLMs for recreating published DES models in simpy: feasibility and pilot"
->"Investigating Generative AI to recreate published healthcare discrete-event simulation in python: a feasibility and pilot study"
-
-
-
-
-**From natural language to simulations: applying AI to automate simulation modelling of logistics systems**
-By Jackson et al. 2024
-* https://doi.org/10.1080/00207543.2023.2276811
-* Focuses on GPT rather than ChatGPT
-* "the language model could produce simulation models for inventory and process control"
-* "enable a significant simplification of simulation model development"
-* Generative AI, a specialised branch of Machine Learning (ML) that focuses on the generation of new content, including images, music, or video, by discerning patterns from existing data (Brynjolfsson, Li, and Raymond Citation2023)
-* Codex (Zaremba and Brockman Citation2021), an advanced model designed to generate code in multiple programming languages based on descriptive prompts, streamlining the coding process and enhancing software development efficiency.
-* we have yet to be aware of research that studies human-AI collaboration in the context of simulation modelling (MacCarthy and Ivanov Citation2022; Saenz, Revilla, and Simon Citation2020a)
-* RQ1: ‘How could simulation models of logistics systems be produced automatically from verbal descriptions in natural language?’.
-* RQ2: ‘How do human experts and AI-based expert systems successfully collaborate in the domain of simulation modelling of logistic systems?’.
-* The idea of the *NLP shortcut*: by-pass conceptual modelling and coding [**WE ARGUE THIS ISN'T TRUE**]
-* **Contributions**:
-* makes two substantial contributions. First, we offer guidelines and a design of the NLP-based framework on how to build simulation models of logistics systems automatically, given the verbal description. Second, and more generally, our work offers a technological underpinning of human-AI collaboration for the development of simulation models
-* Most code generation tasks are focussed on classic software engineering problems not simulation.
-*  Python is selected as the programming language for this endeavor, given that GPT-3 Codex demonstrates the highest level of proficiency in Python 3 (Chen et al. Citation2021)
-*  Experiments:
-*  Model has 20-30 lines of python code.  Note that the language model makes decisions about plotting.
-*  CLAIM: "As highlighted in our study, one of the pivotal advantages of automating the development of simulation models is the potential reduction in time, resources, and human error. Traditional simulation approaches, such as agent-based simulations, often require intricate design, extensive calibration, and rigorous validation. These processes, while essential, are time-consuming and prone to human errors, especially when modelling complex supply chain systems."
-
-**Generative AI and simulation modeling: how should you (not) use large language models like ChatGPT**
-By Akhavan et al. 2024
-* Akhavan_2024
-* https://doi.org/10.1002/sdr.1773
-* System Dynamics model
-* Throughout this article, we show and emphasize that generative AI should not replace thinking; instead, it is a useful tool to facilitate the research process, a practical way to review the content generated by researchers, and an enhancement of idea implementation in simulation modeling.
-* Thus, one contribution of this article is providing well-crafted examples of prompts that illustrate effective communication with generative AI. These examples serve as practical guides for users to understand how to formulate prompts that elicit meaningful responses from tools like ChatGPT.
-* Used it more as an assistant to get feedback, refine ideas, and present options.  Some code was provided to assist with python plots and interfaces to models.
-
-**AI USEFULNESS IN SYSTEMS MODELLING AND SIMULATION: GPT-4 APPLICATION**
-plooy_ai_2023
-*  However, the application of GPT-4 in system dynamics modelling is not without its difficulties. As a language model,  GPT-4  is  inherently  limited  by  the  quality  and  breadth  of  the  data  it  has  been  trained  on.  Consequently, its performance in specialised domains, such as system dynamics, may be contingent on the availability of relevant training data
-*  Despite  the  promise  of  AI-driven  system  dynamics  modelling,  the  role  of  human  expertise  remains  indispensable
-*  To harness AI's potential in system dynamics fully, it is essential to strike a balance between leveraging AI’s capabilities and maintaining the critical role of human expertise
-*  GPT-4.0 was able to design a simple SD model; although it could not identify an error in the simulation.
-*  results demonstrated the importance for a human in the process.
-
-**CHANCES AND CHALLENGES OF CHATGPT AND SIMILAR MODELS
-FOR EDUCATION IN M&S**
-* consider how generative AI will affect simulation education and how teaching in hr
-the field should adapt to accomodate and exploit it.  One such option is student and AI teams where the teams works over a number of iterations to build a computer simulation model. A hypothetical example in NetLogo is presented.
-
-**GPT-BASED MODELS MEET SIMULATION: HOW TO EFFICIENTLY USE LARGE-SCALE PRE-TRAINED LANGUAGE MODELS ACROSS SIMULATION TASKS** by Giabbanelli 2023
-Giabbanelli_GPT_Sim
-
-* https://dl.acm.org/doi/10.5555/3643142.3643385
-* We focus on four modeling and simulation tasks,
-* structure of a conceptual model to promote the engagement of participants in the modeling process
-* summarizing simulation outputs
-* broaden accessibility to simulation platforms by conveying the insights of simulation visualizations via tex
-* explaining simulation errors and providing guidance to resolve them.
-
-
-**Automatically Explaining a Model: Using Deep Neural Networks to Generate Text From Causal Maps** 
-by Shrestha et al 2022
-Shrestha_gpt_explain_model
-
-* We proposed a process to transform a large conceptual model (in the form of a causal map) into sentences, by decomposing it into smaller parts and then performing Natural Language Generation (NLG) via a fine-tuned GPT-3 (Figure 2). Automatic metrics that tolerate variations in language show encouraging performances on two case studies (Table 1)
-
-
-
 ## References
 
 ```{bibliography}