Participant profile

Doctoral students of the UPV/EHU working on any topic, in any research field, from any discipline (every researcher needs to prepare research publications)

Calendar

October 2023

Duration

Two hours and a half

Timetable

Time 10:00 to 12:45 with a 15min break

Language

English

Requirements

Good level of English, spoken and written

Modality

A real-classroom

Location and dates

Speaker, Trainer and Profile

Tim Smithers: I am a designer and researcher.  I have mostly worked in AI in Design research and intelligent robotics, but also, with others, on research topics in the Arts, Engineerings, Humanities, and Sciences.  All this has involved working with many different people from different disciplines, in different collaborations, large and small, and, in different research groups and Labs in some different parts of the world, but mostly Europe.  I have also long been interested in the practices that make up good research, and in how to help PhDers make a good start on learning these needed practices.  Since 2010 I have worked as a freelance Research Practitioner, developing and teaching courses on the Foundations of Research Practices for mixed discipline groups of PhDers, helping research groups design their research programmes and strengthen their research practices, and occasionally I work for the Research Executive Agency (REA) of the EU Commission.  I also continue to design things and to do some research.  I currently work in a small collaboration on an empirical investigation of a topic in Number Theory; a project to investigate on an old, but still little understood, question in Behaviour selection in robots (and animals); and, a project to investigate the effective support of a particular kind of discovery that happens in all kinds of design exploration, using sound design by drawing as the designing, and (Good Old) symbol processing knowledge representation and inference techniques in AI.

You can find more about my work experience on my LinkedIn profile page, and here, find most of my publications on Academia.

Group size

20 people maximum in each group

Registration

Description

Generative AI systems, and, in particular, (so called) Large Language Model (LLM) based systems, like ChatGPT, are here, and they aren't going away (nor are they going to take over the World, nor destroy Humanity). An important question for researchers, however, and for all PhDers, in particular, is can these systems be used well in preparing research publications? If so, how? For doing what? And what are the hazards to be avoided, and what are the ethical issues to take care of? In this Seminar, designed for PhDers working on any topic, in any research discipline, we will consider these questions, make a start on answering them, talk about some examples, and get some pointers to where to keep looking for answers and further help.

Programme

Over 2.5 hours we will do the following:

 Start 10:00

 0. A round of (very) short introductions 

 1. A short introduction to Generative AI Systems, Foundational Systems, LLMs: how an LLM is built; how it works; and what can it be made to do? (And why LLM is not a very good name.)

 2. What is ChatGPT really?  How does it work?  How can it be used well in preparing research publications, and how to critically think about this use?  Plus, a brief review of (so called) "Prompt Priming" and "Prompt Engineering."

 Coffee Break (15min)

 3. Ethical considerations and recommended practices

• Research publications should aim to tell the truth, the whole truth, and nothing but the truth: you, the author, are always responsible for this
• All use of ChatGPT remains the author's responsibility, and must be carefully and thoroughly checked
• Any use of ChatGPT, or similar system, should be explicitly acknowledged and explained
• How to make sure you're allowed to use ChatGPT, or a similar text generating system

 4. Further Questions and Discussion ...

 Finish 12:45

Notes, containing some illustrative examples and pointers to  other useful things to read and places to look, will be made  available as a PDF prior to and for this Seminar.

PhDers attending this Seminar will be expected to, and  probably want to, make their own notes as well from what we  talk about and discuss. 

This course is offered in collaboration with Euskampus Fundazioa.