Andrea Viliotti

The episode examines the challenges and opportunities presented by cyber resilience in the era of emerging technologies such as artificial intelligence, quantum computing, the Internet of Things, and blockchain. It highlights the risks associated with the increasing interconnectivity of these technologies, their rapid evolution, and the difficulty in regulating them. The importance of a proactive...

Questa puntata esplora il concetto di introspezione nei modelli linguistici di grandi dimensioni (LLM) come GPT-4 e Llama-3. Gli autori della ricerca hanno addestrato un modello chiamato M1 per sviluppare questa capacità, che gli permette di riflettere sui propri stati interni e prevedere il proprio comportamento in scenari ipotetici. I risultati hanno dimostrato che M1 supera significativamente u...

The episode examines the possibility that large language models (LLMs) may develop a form of introspection, that is, the ability to reflect on their own internal states and predict their behavior. Through experiments conducted on two distinct models, one with introspection (M1) and one without (M2), the authors demonstrate that M1 is more accurate in predicting its own behavior, suggesting it poss...

Il Barometro Internazionale Cegos 2024 esamina le sfide globali relative allo sviluppo delle competenze nelle aziende, con particolare attenzione all'impatto dell'intelligenza artificiale e della trasformazione digitale. L'indagine evidenzia l'urgenza di accelerare il "time to competency", ovvero il tempo necessario per acquisire nuove competenze, e di adottare un approccio personalizzato e umano...

The episode describes a new deep learning model called nGPT, based on the Transformer architecture, which utilizes a normalized representation on a hypersphere. This new structure offers several advantages over traditional Transformers, including faster convergence, greater stability during training, and improved generalization abilities on downstream tasks. Normalizing on a hypersphere allows for...

This episode explores the impact of Narrative Artificial Intelligence (NAI) in the evaluation process of innovative solutions, comparing the results obtained with the assistance of a traditional "black box" AI (BBAI) and without any AI support. The work demonstrates that NAI, by providing detailed explanations of its recommendations, promotes greater critical scrutiny and deeper engagement from hu...

The 2024 Cegos International Barometer is a survey that explores the challenges of technological, climatic, and social transformation in the workplace. Conducted among 5,000 employees and 469 human resources managers across nine countries, the survey highlights the impact of artificial intelligence and data analytics on the skills required. The Barometer analyzes the challenges and opportunities r...

The episode presents the story of Sammy Basso, a young man affected by progeria, a rare genetic disease that causes premature aging. The narrative follows his life journey, from birth to his sudden death at the age of 28. The text recounts his personal experiences, the challenges he faced due to the disease, and his contributions to scientific research on progeria. It describes his passion for sci...

The Atlas of Fintech is a publication that provides a detailed overview of the Italian fintech and insurtech start-up ecosystem, offering up-to-date information on companies, industry trends, and the challenges the market is facing. The report highlights the importance of strategic collaborations between various companies, the growing use of artificial intelligence, the increasing emphasis on sust...

The episode explores the impact of artificial intelligence (AI) on financial reporting and auditing, highlighting how AI is transforming these areas by making them more efficient and informed. The text examines the advantages of AI in these sectors, including forecasting trends, real-time risk management, data accuracy, operational efficiency, and the customization of financial reports. It also de...

The episode explores the impact of artificial intelligence (AI) on the field of medicine, with a particular focus on the implications for empathy in the doctor-patient relationship. It discusses whether AI, despite its ability to simulate empathy through algorithms, can truly replace the human connection that underpins genuine healthcare. The articles review the potential benefits of AI, such as t...

The episode discusses the rapid and widespread adoption of generative artificial intelligence (AI) in the United States, comparing it to the spread of earlier technologies like the personal computer and the Internet. The article highlights how generative AI is transforming work and daily life, with applications ranging from writing texts to managing personal tasks. It analyzes data on the use of g...

The episode examines the phenomenon of "hallucinations" in Large Language Models (LLMs), errors that manifest as inaccurate information, biases, or reasoning defects. The article focuses on the internal representation of these hallucinations within LLMs, demonstrating that models encode truthfulness signals within their internal representations. The article then explores the various types of error...

ConceptAgent is a new system developed to make robots more autonomous and flexible, enabling them to perform tasks described in natural language. The system leverages large language models (LLMs) combined with innovative planning techniques to handle complex and ever-changing environments. ConceptAgent stands out for three key innovations: Predicate Grounding, which checks prerequisites to ensure...

The seventh GIMBE report analyzes the state of the Italian National Health Service (SSN), highlighting several critical issues that threaten the sustainability of the system. Among the main problems are underfunding, managerial inefficiency, increasing out-of-pocket expenses, and regional disparities in access to care. The report emphasizes the urgency of structural interventions to ensure equitab...

The episode discusses the Morris II worm, a new type of malware targeting generative AI (GenAI) systems like ChatGPT and Gemini. The worm spreads using deceptive prompts to self-replicate, targeting vulnerabilities in the language generation and information extraction technologies used in GenAI models. This malware raises concerns about the security of GenAI systems, emphasizing the importance of...

LeanAgent is an artificial intelligence system developed for proving mathematical theorems. The system employs a continuous learning approach that allows it to enhance its capabilities over time and adapt to new mathematical domains. This approach is based on three key components: curriculum-based learning, a dynamic database, and progressive training. LeanAgent has proven capable of solving compl...

This episode provides an analysis of the capabilities of large multimodal models (MLLMs) in non-verbal abstract reasoning. The experiment employs various versions of the Raven's Progressive Matrices, a standard test for measuring fluid intelligence, to evaluate the models' ability to interpret visual relationships and deduce missing parts of puzzles based on abstract rules. The results show that o...

Andrea Viliotti is an innovation manager specializing in the application of generative artificial intelligence for businesses. He provides strategic and commercial consulting to startups and companies for the effective use of generative AI. His expertise includes business management and a deep understanding of digital technologies. Viliotti offers his consulting services through his website, where...

The episode describes the awarding of the 2024 Nobel Prize in Chemistry to Demis Hassabis, CEO of DeepMind, for his work in creating AlphaFold, an artificial intelligence system that can predict the three-dimensional structures of proteins. The article highlights the importance of this discovery for advancing scientific research, particularly in the fields of biology and medicine, and how Hassabis...

The episode provides an overview of Tutor CoPilot, an artificial intelligence tool designed to support tutors in education. Developed by a team of researchers from Stanford University, Tutor CoPilot offers real-time suggestions to tutors, improving teaching quality and student outcomes. The study described in the text evaluated the effectiveness of Tutor CoPilot in disadvantaged schools in the sou...

The impact of the work of John Hopfield and Geoffrey Hinton, winners of the 2024 Nobel Prize in Physics. Both scientists utilized principles of physics, such as statistical mechanics and thermodynamics, to develop machine learning methods. Hopfield's work, based on the 'Hopfield network model,' led to learning systems capable of optimizing complex configurations. Hinton, on the other hand, introdu...

Meta has introduced Movie Gen, an artificial intelligence system that creates high-resolution videos from text. Movie Gen is capable of customizing videos using reference images and modifying existing videos with textual instructions. The system also generates cinematic audio synchronized with the video. Tests demonstrate that Movie Gen surpasses similar systems by OpenAI and Runway in terms of vi...

This episode explores the importance of a constructive skeptical attitude towards artificial intelligence. The author emphasizes that, despite the enthusiastic expectations, it is essential to adopt a critical and reflective approach to carefully assess the risks and benefits of this rapidly evolving technology. Several examples of technologies that initially encountered skepticism, but later prov...

The episode is dedicated to the "AI Risk Repository," a database that aggregates and categorizes risks arising from artificial intelligence. The repository uses two taxonomies: the "Causal Taxonomy," which analyzes the causes of a risk, and the "Domain Taxonomy," which classifies risks into seven key areas. The text delves into various types of risks, such as discrimination, privacy violations, di...