New frontiers in research:
using AI for your research and your thesis
Luca Mari
mercoledì 2 ottobre 2024
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
The “new” artificial intelligence: machine learning
“Machine learning is the field of study
that gives computers the ability to learn
without being explicitly programmed”
A. Samuel, 1959
Machine learning systems are learning machines, not programmed machines
They can be interpreted as (sometimes very complex) parametric functions,�with values assigned to parameters through an optimization (training) process
The tools, the uses
artificial intelligence
knowledge/rule-based,�expert systems
data-driven,�machine learning systems
programmed vs learned�behavior
single Q&A systems
chatbots
context-free vs context-sensitive�behavior
classifiers / regressors
generative systems
structured vs generative�behavior
data analysis
(previous +)�translations,�summarizations, …
(previous +)�interactive text�processing
Chatbots: some research-related uses
Q: Can you list some significant examples of uses of current chatbots in scientific research,�and specifically in management sciences and engineering? [to ChatGPT 4o]
A: Chatbots, particularly those powered by advanced AI like ChatGPT, have become valuable tools in scientific research, including management sciences and engineering. Notable applications include:
1. Literature Review and Information Retrieval: AI-driven chatbots assist researchers in quickly retrieving and summarizing relevant literature, streamlining the initial phases of research. This capability is beneficial across various scientific disciplines, including management sciences and engineering.
2. Data Analysis and Interpretation: In engineering, chatbots can analyze complex datasets, identify patterns, and provide interpretations, aiding in tasks such as predictive maintenance and quality control. This application enhances efficiency and accuracy in engineering research.
3. Educational Support and Training: Chatbots serve as interactive tutors, offering explanations and answering questions on complex topics in management sciences and engineering. They provide personalized learning experiences, reinforcing understanding and facilitating continuous education.
4. Simulation and Modeling: In engineering research, chatbots assist in setting up and running simulations, modeling scenarios, and interpreting outcomes. This support is crucial in fields like systems engineering and operations research.
5. Writing Assistance: Chatbots aid in drafting, editing, and proofreading scientific documents, ensuring clarity and adherence to academic standards. This function is valuable for researchers in management sciences and engineering, enhancing the quality of publications.
6. Project Management: In management sciences, chatbots help in planning, scheduling, and monitoring research projects. They provide real-time updates and facilitate communication among team members, improving project efficiency.
A moving target?
A moving target
Prompt design: new / well established skills
Current chatbots produce texts that are the outcome�of autonomous processing from a large amount of texts,�not of queries in databases
They are able to operate in original and sophisticated ways but:
Their behavior
– depends on the quality of the training process and the training set
– is hardly predictable and not always reliable
Learning machines are opinion makers, not truth machines
The responsibility on what they produce�remains on us
It is rational to interact with them as consultants, not colleagues
[my prompt] “Please produce a picture with a little green man who is sorry for not being perfectly reliable in the information it generates.”
[ChatGPT interpretation] “Photo of a little green alien, standing with a slightly bowed head and a remorseful expression. He's holding a sign that reads 'Sorry for the oversight' with Earth visible in the background.”
Thanks for your attention