This thesis delves into Smart Contracts, based on Blockchain technology, and experiments with their use with Generative Artificial Intelligence (AI). The ultimate objective is to train, by analyzing the limits of hardware resources, an artificial intelligence model that is capable of writing Smart Contracts, specifically Solidity code, widely used in the sector. In this case, training cycles are performed on models such as GPT2 and BLOOM. The datasets to perform the training are reconstructed, starting from datasets found on the internet, according to the needs of the model. The fundamental concepts of Blockchain, Smart Contracts and artificial intelligence are also explored in depth, with related strengths. Finally, the ethical risks deriving from the use of AI are analyzed and a summary is made in the final chapter.
Questa tesi approfondisce gli Smart Contract, basati su tecnologia Blockchain, e ne sperimenta l'uso con Intelligenza Artificiale (AI) Generativa. L'obiettivo ultimo è addestrare, analizzando i limiti delle risorse hardware, un modello di intelligenza artificiale che sia in grado di scrivere Smart Contract, nello specifico codice Solidity, ampiamente utilizzato nel settore. Nella fattispecie si eseguono cicli di addestramento su modelli quali GPT2 e BLOOM. I dataset per eseguire l'addestramento vengono ricostruiti, partendo da dataset reperiti su internet, secondo l'esigenza del modello. Vengono anche approfonditi i concetti fondamentali di Blockchain, Smart Contract e intelligenza artificiale, con annessi punti di forza. Infine vengono analizzati i rischi etici derivanti da utilizzo di AI e viene effettuato un riassunto nel capitolo finale.
UNA INTELLIGENZA ARTIFICIALE GENERATIVA PER LO SVILUPPO DI SMART CONTRACT IN SOLIDITY
PALMISANO, MIKI
2023/2024
Abstract
This thesis delves into Smart Contracts, based on Blockchain technology, and experiments with their use with Generative Artificial Intelligence (AI). The ultimate objective is to train, by analyzing the limits of hardware resources, an artificial intelligence model that is capable of writing Smart Contracts, specifically Solidity code, widely used in the sector. In this case, training cycles are performed on models such as GPT2 and BLOOM. The datasets to perform the training are reconstructed, starting from datasets found on the internet, according to the needs of the model. The fundamental concepts of Blockchain, Smart Contracts and artificial intelligence are also explored in depth, with related strengths. Finally, the ethical risks deriving from the use of AI are analyzed and a summary is made in the final chapter.I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14253/104644