This work explores the integration of Artificial Intelligence (AI) to improve research skills in engineering education, specifically at the time of conducting a Systematic Literature Review (SLR). While AI has previously been employed for basic tasks related to SLRs, such as relevant paper screening and classification within scope, there is still limited research on information retrieval from the full texts of scientific papers. This study focuses on the next phase in the SLR, which involves scientific documentary analysis. The methodology presented in this work compares the AI tools in the market to facilitate scientific documentary analysis and enhance and refine research skills in engineering students. Also, this paper addresses the challenges to automatically run documentary analysis during a SLR. To do so, several techniques to extract information from AIs are presented, including Application Programming Interfaces (API) or Graphical User Interfaces (GUI). According to the results, most of the AI tools have a limitation in the operations students can perform per day. Only PDFgear offers a no-cost solution with unlimited usage, while some other AI tools allow limited usage for not so high prize. In conclusion, this paper presents a contribution during the documentary analysis of a SLR. It is important to keep in mind the constraints associated with the limited usage of AIs, the accuracy of AI tools, or the complexity of the developed scripts for automating the process. Due to the rapidly changing market for artificial intelligence, the validity of this study is limited to the current state of the tools. Similar studies will be necessary as AI tools continue to evolve.

The assessment of non-technical skills (NTSs) is a key task in the medical field. However, it is difficult to implement because it relies on experts assessing trainees. This is a time-consuming process that can differ between evaluators. In recent decades, NTS rating scales have been developed to reduce subjectivity in crisis resource management (CRM) scenarios. Additionally, many tools have been developed to automatically measure NTSs. However, these objective automated tools calculate low-level metrics that are sometimes difficult to correlate with NTS behaviors. Leadership metrics, for example, have rarely been explored in the existing literature, nor have speech content metrics. This work proposes a cost-effective pilot solution for extracting and visualizing NTS behaviors from CRM audio content. The exploratory pipeline consists of three independent steps, each of which was evaluated for the feasibility of automation: (1) pre-processing the audio and creating transcrips via automatic speech-to-text conversion; (2) using a generative artificial intelligence large language model to collect structured data on communication interactions and NTSs from the transcripts; and (3) formatting the data to provide insightful visualizations of the CRM scenario. The results of the first phase showed that WhisperX performed best at generating audio transcripts, and that some NR techniques improved transcript quality. However, there is still room for improvement regarding acknowledgments (ACKs) and speaker recognition. Preliminary results from the second phase showed that the developed NTS data model can extract relevant information from transcripts in a structured way. Nonetheless, the proposed prompts require refinement to infer data more accurately and improve order and NTS extraction results. Finally, the results of the third phase showed that the proposed knowledge graph visualizations effectively summarize what happened during the standardized scenario. This could enable evaluators to provide additional information during the debriefing session after the experience. This exploratory work contributes to the streamlining of NTS extraction during the evaluation process in standardized scenarios by proposing a pilot for NTS extraction and visualization from speech content. Further research is needed to optimize audio quality, refine the LLM prompts and achieve full automation of the pipeline.