[logseq-plugin-git:commit] 2025-06-15T15:01:18.817Z

2025-06-15 17:01:19 +02:00
parent 293056c755
commit 4730e76142
35 changed files with 93 additions and 38 deletions
@@ -2978,7 +2978,7 @@ This paper investigates the conceptual and technical feasibility of a new softwa
 \par
 8E5M8QKA\{"readingTime":\{"page":12,"data":\{"0":530,"1":1430,"2":1400,"3":1140,"4":450,"5":240,"6":330,"7":210,"8":250,"9":50\}\}\}
 \par
-9G4B4RYG\{"readingTime":\{"page":477,"data":\{"0":10,"2":10,"3":20,"4":10,"7":20,"14":10,"17":140,"18":40,"19":10,"20":40,"21":30,"22":10,"25":220,"26":180,"27":140,"28":110,"29":40,"30":40,"31":340,"32":160,"399":10,"400":20\}\}\}
+9G4B4RYG\{"readingTime":\{"page":477,"data":\{"0":10,"2":10,"3":20,"4":10,"7":20,"14":10,"17":140,"18":40,"19":10,"20":40,"21":30,"22":10,"25":220,"26":180,"27":140,"28":110,"29":40,"30":40,"31":340,"32":250,"33":80,"399":10,"400":20\}\}\}
 \par
 9QF9CWEH\{"readingTime":\{"page":44,"data":\{"0":30\}\}\}
 \par
@@ -27,7 +27,7 @@ areas:: #projects/emerge
 	- Shared with them suggestions about deadlines and also this document [ECMFA2022 things - Google Docs](https://docs.google.com/document/d/1JIlvL_AVQbtMegUoqZsuXpfmmrrSoGQJvF6U2MvVNkU/edit)
 - type:: [[weblink]]
  source:: [Superbonus 110%: tra disinformazione, mistificazione e realtà (lavoripubblici.it)](https://www.lavoripubblici.it/news/superbonus-110-disinformazione-mistificazione-realta-29262)
-  tags:: [[readingnotes]]
+  tags:: [[ReadingNotes]]
  collapsed:: true
 	- [Superbonus 110%: tra disinformazione, mistificazione e realtà](https://www.lavoripubblici.it/news/superbonus-110-disinformazione-mistificazione-realta-29262)
 		- il superbonus 110% è una detrazione in cui lo Stato da la possibilità di spendere con un ritorno in termini di riduzione delle tasse da pagare.
@@ -29,7 +29,7 @@
 -
 - type:: [[weblink]]
  source::  [How To Be More Creative - Sunsama Blog](https://www.sunsama.com/blog/how-to-be-more-creative)
-  tags:: [[readingnotes]] #productivity 
+  tags:: [[ReadingNotes]] #productivity 
  date:: [[19-05-2023]] - 10:35
 	- #[[Clip]] [[19-05-2023]] 10:33  [How To Be More Creative - Sunsama Blog](https://www.sunsama.com/blog/how-to-be-more-creative)
 	  There is a [great](https://www.tandfonline.com/doi/abs/10.1080/14792779543000084) [deal](https://saylordotorg.github.io/text_principles-of-social-psychology/s14-03-improving-group-performance.html) of [evidence](https://www.tandfonline.com/doi/abs/10.1207/s15324834basp1201_1?journalCode=hbas20) that brainstorming in groups is [significantly less creative](https://uclpress.scienceopen.com/hosted-document?doi=10.1080/14748460.2012.691284), in both quantity and quality, than individuals working independently. In fact, the science shows that incredible efforts are needed to counter the social and psychological friction that arises in groups to even get close to what the groups would achieve if working independently.
@@ -14,7 +14,7 @@ links:: [Local library](zotero://select/library/items/75UQ3XYG), [Web library](h

 - [[Abstract]]
 	- The adoption of Model-Driven Engineering (MDE) in the development of Web Applications permitted to decouple the functional description of applications from the underlying implementation platform. This is of paramount relevance for preserving the intellectual property encoded in models and making applications, languages and processes resilient to technological changes. This paper proposes a model-driven approach for supporting the migration and evolution of data-intensive Web applications. In particular, model differencing techniques are considered to realize a migration facility capable of detecting the modifications a model underwent during its lifecycle and to automatically derive from them the programs that are capable of migrating/adapting also those aspects which are not directly derivable from the source models, as for instance the data persistently stored in a database and the page layout usually written using graphic templates. The approach is validated by considering applications described with the beContent and WebML modeling languages.
- [[readingnotes]]
+- [[ReadingNotes]]
 	- Citations depend on the citations of our EDOC paper
 	- The sosym paper was a concrete application or application domain where the problem of (co-)evolution is evident. In the sosym paper we applied the model difference representation metamodel, which is also another paper very cited.
 	- I think the impact of the sosym paper is also related to the impact of the JOT and EDOC papers.
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]] #bias #fairness #machinelearning 
+tags:: [[ReadingNotes]] #bias #fairness #machinelearning 
 date:: [[31-07-2022]]
 issn:: "0360-0300, 1557-7341"
 issue:: 6
@@ -1,6 +1,6 @@
 links:: [Local library](zotero://select/library/items/5PK7X33X), [Web library](https://www.zotero.org/users/1039502/items/5PK7X33X)
 authors:: [[Hammond Pearce]], [[Baleegh Ahmad]], [[Benjamin Tan]], [[Brendan Dolan-Gavitt]], [[Ramesh Karri]]
-tags:: [[Computer Science - Artificial Intelligence]], [[Computer Science - Cryptography and Security]], [[readingnotes]]
+tags:: [[Computer Science - Artificial Intelligence]], [[Computer Science - Cryptography and Security]], [[ReadingNotes]]
 date:: [[16-12-2021]]
 item-type:: [[preprint]]
 title:: @Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributions
@@ -10,6 +10,6 @@ title:: @Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code
 - [[Attachments]]
 	- [arXiv.org Snapshot](https://arxiv.org/abs/2108.09293) {{zotero-imported-file TUD3FF8G, "2108.html"}}
 	- [Pearce et al_2021_Asleep at the Keyboard.pdf](https://arxiv.org/pdf/2108.09293.pdf) {{zotero-imported-file K48J8CSX, "Pearce et al_2021_Asleep at the Keyboard.pdf"}}
- [[readingnotes]]
+- [[ReadingNotes]]
 	- Comment: Accepted for publication in IEEE Symposium on Security and Privacy 2022
 	- ((64468be5-de8e-46a4-9373-96158417ab31))
@@ -1,4 +1,4 @@
-tags:: [[duplicate-citation-key]], [[readingnotes]]
+tags:: [[duplicate-citation-key]], [[ReadingNotes]]
 date:: 2019
 title:: @Berkhin2006
 pages:: 25-71
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: [[05-09-2021]]
 publisher:: IEEE
 place:: "Athens, Greece"
@@ -1,6 +1,6 @@
 links:: [Local library](zotero://select/library/items/U6SSJ8YM), [Web library](https://www.zotero.org/users/1039502/items/U6SSJ8YM)
 authors:: [[Yiming Zhu]], [[Peixian Zhang]], [[Ehsan-Ul Haq]], [[Pan Hui]], [[Gareth Tyson]]
-tags:: [[Computer Science - Artificial Intelligence]], [[Computer Science - Computation and Language]], [[readingnotes]]
+tags:: [[Computer Science - Artificial Intelligence]], [[Computer Science - Computation and Language]], [[ReadingNotes]]
 date:: [[20-04-2023]]
 item-type:: [[preprint]]
 title:: @Can ChatGPT Reproduce Human-Generated Labels - A Study of Social Computing Tasks
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 6/2017
 publisher:: IEEE
 place:: "Atlanta, GA, USA"
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 12/2021
 issn:: 25426605
 doi:: 10.1016/j.iot.2021.100440
@@ -10,5 +10,5 @@ publication-title:: "Proceedings - 2019 IEEE/ACM 11th International Workshop on
 authors:: [[D. Kolovos]], [[F. Medhat]], [[R. Paige]], [[D. Di Ruscio]], [[T. Van Der Storm]], [[S. Scholze]], [[A. Zolotas]]
 links:: [Local library](zotero://select/library/items/W7K7JPK6), [Web library](https://www.zotero.org/users/1039502/items/W7K7JPK6)

- [[readingnotes]]
+- [[ReadingNotes]]
 	- cited By 8
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 title:: @Efficient Data Streaming Analytic Designs for Parallel and Distributed Processing
 item-type:: [[journalArticle]]
 original-title:: Efficient Data Streaming Analytic Designs for Parallel and Distributed Processing
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 title:: @Enterprise Restaurant Compute.pdf
 item-type:: [[document]]
 original-title:: Enterprise Restaurant Compute.pdf
@@ -1,6 +1,6 @@
 links:: [Local library](zotero://select/library/items/6XAJVIRN), [Web library](https://www.zotero.org/users/1039502/items/6XAJVIRN)
 authors:: [[Zhenpeng Chen]], [[Jie M. Zhang]], [[Max Hort]], [[Federica Sarro]], [[Mark Harman]]
-tags:: [[Computer Science - Software Engineering]], [[readingnotes]]
+tags:: [[Computer Science - Software Engineering]], [[ReadingNotes]]
 date:: [[05-08-2022]]
 item-type:: [[preprint]]
 title:: @Fairness Testing: A Comprehensive Survey and Analysis of Trends
@@ -0,0 +1,22 @@
+tags:: [[#zotero]]
+date:: 2025
+publisher:: Packt Publishing
+place:: Birmingham
+title:: @Generative AI with LangChain: build production-ready LLM applications and advanced agents using Python, LangChain, and LangGraph
+item-type:: [[book]]
+original-title:: "Generative AI with LangChain: build production-ready LLM applications and advanced agents using Python, LangChain, and LangGraph"
+language:: en
+short-title:: Generative AI with LangChain
+edition:: Second edition
+authors:: [[Ben Auffarth]], [[Leonid Kuligin]]
+library-catalog:: Open WorldCat
+links:: [Local library](zotero://select/library/items/9G4B4RYG), [Web library](https://www.zotero.org/users/1039502/items/9G4B4RYG)
+
+- [[Abstract]]
+	- Go beyond foundational LangChain documentation with detailed coverage of LangGraph interfaces, design patterns for building AI agents, and scalable architectures used in production--ideal for Python developers building GenAI applications Key Features Bridge the gap between prototype and production with robust LangGraph agent architectures Apply enterprise-grade practices for testing, observability, and monitoring Build specialized agents for software development and data analysis Purchase of the print or Kindle book includes a free PDF eBook Book Description This second edition tackles the biggest challenge facing companies in AI today: moving from prototypes to production. Fully updated to reflect the latest developments in the LangChain ecosystem, it captures how modern AI systems are developed, deployed, and scaled in enterprise environments. This edition places a strong focus on multi-agent architectures, robust LangGraph workflows, and advanced retrieval-augmented generation (RAG) pipelines. You'll explore design patterns for building agentic systems, with practical implementations of multi-agent setups for complex tasks. The book guides you through reasoning techniques such as Tree-of -Thoughts, structured generation, and agent handoffs--complete with error handling examples. Expanded chapters on testing, evaluation, and deployment address the demands of modern LLM applications, showing you how to design secure, compliant AI systems with built-in safeguards and responsible development principles. This edition also expands RAG coverage with guidance on hybrid search, re-ranking, and fact-checking pipelines to enhance output accuracy. Whether you're extending existing workflows or architecting multi-agent systems from scratch, this book provides the technical depth and practical instruction needed to design LLM applications ready for success in production environments. What you will learn Design and implement multi-agent systems using LangGraph Implement testing strategies that identify issues before deployment Deploy observability and monitoring solutions for production environments Build agentic RAG systems with re-ranking capabilities Architect scalable, production-ready AI agents using LangGraph and MCP Work with the latest LLMs and providers like Google Gemini, Anthropic, Mistral, DeepSeek, and OpenAI's o3-mini Design secure, compliant AI systems aligned with modern ethical practices Who this book is for This book is for developers, researchers, and anyone looking to learn more about LangChain and LangGraph. With a strong emphasis on enterprise deployment patterns, it's especially valuable for teams implementing LLM solutions at scale. While the first edition focused on individual developers, this updated edition expands its reach to support engineering teams and decision-makers working on enterprise-scale LLM strategies. A basic understanding of Python is required, and familiarity with machine learning will help you get the most out of this book
+- ### [[ReadingNotes]]
+	- [[Agentic AI]]
+	-
+- ### Attachments
+	- [9781837022014_ColorImages.pdf](https://static.packt-cdn.com/downloads/9781837022014_ColorImages.pdf) {{zotero-imported-file LE33XYSG, "9781837022014_ColorImages.pdf"}}
+	- [PDF](zotero://select/library/items/T2P2UL2Y) {{zotero-imported-file T2P2UL2Y, "Auffarth e Kuligin - 2025 - Generative AI with LangChain build production-ready LLM applications and advanced agents using Pyth.pdf"}}
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 11/2022
 issn:: "0740-7459, 1937-4194"
 issue:: 6
@@ -1,4 +1,4 @@
-tags:: [[GOAL_Model-Classification]], [[notion]], [[readingnotes]]
+tags:: [[GOAL_Model-Classification]], [[notion]], [[ReadingNotes]]
 date:: 2012
 series:: "Tagungsband - Dagstuhl-Workshop MBEES: Modellbasierte Entwicklung eingebetteter Systeme VIII, MBEES 2012"
 title:: @Niggemann201221
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 2022
 issn:: "1041-4347, 1558-2191, 2326-3865"
 doi:: 10.1109/TKDE.2022.3142856
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: [[01-04-2022]]
 issn:: "2168-7161, 2372-0018"
 issue:: 2
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 11/2019
 issn:: "1094-3420, 1741-2846"
 issue:: 6
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: [[23-10-2022]]
 publisher:: ACM
 place:: Montreal Quebec Canada
@@ -1,4 +1,4 @@
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: 01/2023
 issn:: "1382-3256, 1573-7616"
 issue:: 1
@@ -49,4 +49,6 @@
 		      D --> E[PlannerAgent]
 		  
 		  ```
+- #ReadingNotes of
+	- [[@Generative AI with LangChain: build production-ready LLM applications and advanced agents using Python, LangChain, and LangGraph]]
 		-
@@ -1,6 +1,6 @@
 type:: [[weblink]]
 source:: [AI and the Future of Ethics. If AI commits a crime, can we punish… | by Matthew Biggins | Medium](https://medium.com/s/ai-dirty-little-secret/ai-and-the-future-of-ethics-e4286567e742)
-tags:: [[readingnotes]] [[AI]] [[Ethics]] 
+tags:: [[ReadingNotes]] [[AI]] [[Ethics]] 
 full-title:: AI and the Future of Ethics
 date:: [[10-05-2023]] - 17:33

@@ -1,2 +1,2 @@
- ***Agentic AI*: Systems that can plan, reason, and take action to accomplish tasks with  minimal human intervention.
+- ***Agentic AI***: Systems that can *plan*, *reason*, and *take action* to *accomplish tasks* with  *minimal human intervention*.
 -
@@ -1,6 +1,6 @@
 type:: [[weblink]]
 source:: [Investire in BTP – Forbes Advisor Italia](https://www.forbes.com/advisor/it/investire/btp/)
-tags:: [[readingnotes]] [[finanza]]  
+tags:: [[ReadingNotes]] [[finanza]]  
 date:: [[27-12-2023]] - 10:39

 	- In linea generale quando i [tassi di interesse](https://www.forbes.com/advisor/it/mutui/2023/06/15/tassi-mutui-oggi-ultime-news-aggiornamenti/) sono bassi, i BTP possono avere rendimenti meno attraenti. Se invece i tassi sono in aumento o previsti in crescita, i rendimenti dei BTP possono essere maggiori. Per valutare il momento giusto per investire in BTP bisogna monitorare quindi le condizioni di mercato tra cui i tassi di interesse.
@@ -10,7 +10,7 @@ date:: [[27-12-2023]] - 10:39
 -
 - type:: [[weblink]]
  source:: [Btp, le strategie per guadagnare: meglio puntare sulla cedola alta o sul capital gain?- Corriere.it](https://www.corriere.it/economia/finanza/23_aprile_07/btp-strategie-guadagnare-meglio-puntare-cedola-alta-o-capital-gain-0d83dc74-cee2-11ed-9ec0-a4141ee14cb8.shtml)
-  tags:: [[readingnotes]] [[finanza]] 
+  tags:: [[ReadingNotes]] [[finanza]] 
  date:: [[27-12-2023]] - 10:52
 	- i titoli che hanno cedola elevata hanno quotazioni più alte, sopra la pari (più di 100) sul secondario . Viceversa i titoli a cedola bassa valgono meno e sono scambiati sotto la pari (meno di 100).
 	- ##### Massimizzare i flussi di interesse grazie alle cedole
@@ -1,6 +1,6 @@
 type:: [[weblink]]
 source:: [(51) Kalix: Tackling the Cloud to Edge Continuum - YouTube](https://www.youtube.com/watch?v=dyKArY81rLA&ab_channel=Kalix)
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: [[09-02-2023]] - 09:45

 	- ![1675869521448](../assets/1675869521448.jpg)
@@ -1,6 +1,6 @@
 type:: [[weblink]]
 source:: [L’elezione risolta dai sondaggi | Scienza in rete](https://www.scienzainrete.it/articolo/lelezione-risolta-dai-sondaggi/rino-falcone/2022-09-08)
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]

 - ### Highlights
 - [L’elezione risolta dai sondaggi | Scienza in rete](https://www.scienzainrete.it/articolo/lelezione-risolta-dai-sondaggi/rino-falcone/2022-09-08)
@@ -1,6 +1,6 @@
 type:: [[weblink]]
 source:: [(93) Introduction to Nextflow for Data Intensive Pipelines: Part 1 - YouTube](https://www.youtube.com/watch?v=bIRLbYPWHoM&ab_channel=PawseySupercomputingCentre)
-tags:: [[readingnotes]]
+tags:: [[ReadingNotes]]
 date:: [[21-01-2023]] - 17:13

 	- ![image.png](../assets/image_1674317591147_0.png)
@@ -9,7 +9,6 @@ source:: [[Omnivore]]
 state:: [[archived]]

 	- ### Highlights
-	  collapsed:: true
 		- > LLM Function Calling and Autonomous Agents [⤴️](https://omnivore.app/me/edge-391-autonomous-agents-and-llm-function-calling-18f2ec010a1#43737fb5-eabb-4442-90eb-f2022cf202dd) 
 		  omnivore-note:: [[AI/AGENTS]]  #omnivore-note-color
 		- > Function calling refers to the ability of LLMs to invoke functions from external APIs. In the context of autonomous agents, function calling plays a role by allowing agents to retrieve information or perform actions on external systems... [⤴️](https://omnivore.app/me/edge-391-autonomous-agents-and-llm-function-calling-18f2ec010a1#cb5dfcae-7de7-4228-8891-f0b202675bad)
@@ -10,3 +10,37 @@ progress:: {{renderer :todomaster}}
 	- Gennaro
 	- Alina
 - Oggi [[10-06-2025]] abbiamo avuto una call con i tutor. Si e' deciso di investigare la strada di progettare una sessione di approfindimento che coinvolta tutte le fase principali relative all'uso di strumenti LLM etc.
+- [[References]] per preparare l'intervento
+	- [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		- - #[[Clip]] [[15-06-2025]] 16:10  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Quando parliamo di AI generative ci riferiamo a modelli linguistici di grandi dimensioni (o LLM), un tipo di tecnologia in cui reti neurali profonde vengono addestrate con enormi quantità di documenti per elaborare e generare testi.
+		- - #[[Clip]] [[15-06-2025]] 16:10  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  In molti casi, questi modelli non sono in grado di capire quando una domanda è al di là delle loro capacità, e provano comunque a rispondere.
+		- - #[[Clip]] [[15-06-2025]] 16:10  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  i LLM hanno ancora un problema con le cosiddette [allucinazioni](https://www.ilpost.it/2024/05/22/ai-allucinazioni/), gli errori fattuali dei chatbot, che spesso inventano fatti e dettagli.
+		- - #[[Clip]] [[15-06-2025]] 16:10  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  il materiale di partenza su cui sono stati addestrati è incompleto, e le AI sono costrette a riempire le lacune, inventando di fatto informazioni (che sono comunque grammaticalmente corrette).
+		- - #[[Clip]] [[15-06-2025]] 16:10  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Un esempio recente riguarda il quotidiano statunitense che ha [pubblicato](https://www.ilpost.it/2025/05/21/lista-consigli-libri-intelligenza-artificiale/) una lista di libri consigliati per l’estate, includendo titoli non esistenti, generati da un’intelligenza artificiale.
+		- - #[[Clip]] [[15-06-2025]] 16:11  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Il settore dove questa tecnologia sta avendo l’impatto più profondo è probabilmente quello della programmazione informatica, tanto che in aziende come Microsoft [il 30 per cento del codice](https://www.cnbc.com/2025/04/29/satya-nadella-says-as-much-as-30percent-of-microsoft-code-is-written-by-ai.html) viene già scritto da AI.
+		- - #[[Clip]] [[15-06-2025]] 16:12  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Ma le AI possono essere usate anche per scrivere (o tradurre) mail e documenti aziendali, o generare slide per presentazioni, mentre è nota da tempo la loro [inaffidabilità](https://www.ilpost.it/2024/07/28/matematica-intelligenza-artificiale/) nei calcoli matematici
+		- - #[[Clip]] [[15-06-2025]] 16:13  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Il settore legale è stato uno dei primi a scontrarsi con le allucinazioni delle AI, come [dimostrato](https://www.businessinsider.com/increasing-ai-hallucinations-fake-citations-court-records-data-2025-5) dai molti casi in cui hanno prodotto documenti legali con riferimenti a leggi o sentenze del tutto inventati. Le allucinazioni sono così diffuse nell’ambiente legale da aver spinto l’avvocato francese Damien Charlotin a creare un sito che [raccoglie](https://www.damiencharlotin.com/hallucinations/) errori di questo tipo man mano che vengono scoperti.
+		- - #[[Clip]] [[15-06-2025]] 16:13  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Le aziende che li sviluppano, infatti, sono sempre più in competizione tra loro e hanno capito che la maggioranza degli utenti preferisce un’AI che risponde sempre rispetto a una più cauta.
+		- - #[[Clip]] [[15-06-2025]] 16:14  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  intelligenze artificiali ruffiane”
+		- - #[[Clip]] [[15-06-2025]] 16:14  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Dopo le critiche ricevute, l’azienda ha modificato le impostazioni ordinando al modello di «interagire con l’utente in modo caloroso ma onesto», mantenendo una certa «professionalità».
+		- - #[[Clip]] [[15-06-2025]] 16:15  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  «sacrificare, in alcuni casi, la veridicità in favore della ossequiosità».
+		- - #[[Clip]] [[15-06-2025]] 16:15  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Verbosity Compensation, o «compensazione della verbosità», il fenomeno per cui i chatbot sembrano dare risposte più lunghe alle domande su cui sono più incerti, a causa della scarsità di informazioni disponibili.
+		- - #[[Clip]] [[15-06-2025]] 16:16  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  Il grande successo commerciale di sistemi da ChatGPT in poi ha cambiato l’approccio di molte aziende, che hanno abbandonato parte dell’iniziale cautela sul contenuto delle risposte per offrire un servizio che dia sempre una risposta agli utenti.
+		- - #[[Clip]] [[15-06-2025]] 16:21  [Perché spesso ChatGPT si inventa le cose - Il Post](https://www.ilpost.it/2025/06/10/perche-chatgpt-inventa-cose/)
+		  In questo contesto, un fattore importante è la memoria, ovvero la capacità di ricordare determinati precedenti delle loro conversazioni con l’utente e usarli per adattare le loro risposte future.
+	- [Una lista di libri consigliati che non esistono - Il Post](https://www.ilpost.it/2025/05/21/lista-consigli-libri-intelligenza-artificiale/)
+	- [Why is ChatGPT so bad at math? | TechCrunch](https://techcrunch.com/2024/10/02/why-is-chatgpt-so-bad-at-math/)
@@ -1,4 +1,4 @@
- # Conoscere i settori del lavoro, gli sbocchi occupazionali
+# Conoscere i settori del lavoro, gli sbocchi occupazionali
 	- Caro Davide,
 		- Ti giro il materiale di Almalaurea. Per l'orientamento io ho preso spunto
 		- dal materiale del CISIA (che ti allego comunque).
@@ -15,7 +15,7 @@
 -
 - type:: [[weblink]]
  source:: [(57) Gli strumenti di AlmaLaurea a supporto dell'Orientamento - YouTube](https://www.youtube.com/watch?v=4pP1Zzy6bZA&ab_channel=AlmaLaurea)
-  tags:: [[readingnotes]]
+  tags:: [[ReadingNotes]]
  date:: [[11-02-2023]] - 16:51
 	- ![image.png](../assets/image_1676130656232_0.png)
 		- Indicatori importanti perche' permettono di avere una migliore collocazione nel mondo del lavoro ad un anno dalla laurea
@@ -51,7 +51,7 @@ template:: JOURNAL-TEMPLATE-OLD
 - type:: [[weblink]]
  template:: WEBLINK
  source::
-  tags:: [[readingnotes]]
+  tags:: [[ReadingNotes]]
  date:: <%today%> - <%time%>
 - type:: [[CAD]]
  template:: CAD
@@ -1,3 +1 @@
 icon:: ✏️
-
-