Questa e' la prima volta che lo facciamo, quindi andremo un po' a braccio su alcune cose, cercheremo di aggiustare il tiro durante e potremmo avere un sacco da imparare.
Il Panel dovrebbe durare 1h, non mi aspetto che riusciremo a parlare per cosi' tanto ma sarei contentissimo se alla fine ci fossero ancora temi da coprire, vedremo, e' un esperimento!
Sara’ in lingua inglese, perche’ PyMI si rivolge ormai ad un pubblico che non parla solo in italiano.
E’ utile che tutti i partecipanti al panel abbiano almeno preso visione dei temi, delle domande e possibilmente del materiale, che verra’ tutto fornito una settimana prima del panel. Sarebbe meglio che ognuno si formi delle idee coerenti con il tutto, in modo da tenere la conversazione accesa.
| Name | Background (why the audience might be interested in their perspectives) |
|---|---|
| Luca Mattiazzi | Works in a team that creates ML models, not an AI expert in particular, uses it as a tool in his daily work. |
| Daniele Zonca | Senior Architect in OpenShiftAI, can give us the behind-the-scenes perspective and will help us keep our feet on the ground without too much philosophy. |
| Ignacio Sanz | Copywriter, will provide a non-technical view of the use of ChatGPT and similar tools. |
| Marco Vinciguerra | 24 years old, created a Python library that uses LLM for scraping. |
| "LLM Native": started programming when LLMs were already popular. | |
| Fascinated by the emerging profession of "Prompt Engineering". | |
| Emanuele Fabbiani | PhD in ML, professor at UniCatt, AI Engineer at xtream. |
| Develops ML models and uses AI in his daily work. Speaks about it at conferences across Europe. |
Potrebbero essere una sorta di capitoli che percorreremo durante la discussione, mano a mano che evolve, oppure per sollevare qualche tema interessante.
will we all lose our jobs?
how should we use AI as a tool in our daily work? how should we not use it? what are the risks for our profession? what are the ethical implications?
https://en.wikipedia.org/wiki/The_medium_is_the_message
We are moving from code writing to prompt generation to solve business problems.
How will this change affect our approach to problem solving? What new methodologies do we need to develop to interact effectively with AI?
what are the risks for our users? should we impose limits on how we use AI for service development?
AI will play an increasingly important role in our industry, how do you imagine our work in 5 or 10 years? how can we prepare for it?
In questa sezione raccolgo il materiale che puo’ essere utile a sollevare qualche tema, o comunque delle informazioni utili a inquadrare meglio quello di cui stiamo parlando.
https://www3.weforum.org/docs/WEF_Jobs_of_Tomorrow_Generative_AI_2023.pdf
content slop
this shows the tendency of the AI of producing mediocre content, because it can be seen as “converging” to what its dataset was, while humans artists generally tend to “diverge” from the average to propose something different/new
In addition to the aspect that impacts the end users who consume content, we can then broaden the impact to the quality of future models in general since most likely the same content that we are defining as having no intrinsic value will be used as sources to train future models.This issue is related to the fact that the available 'quality' data[1] have now been used practically all of them[2] and much is being invested in synthetic data generation and techniques to recognise generated data[3].