Humanistic AI

Directors of Scientific Unit: Prof. Aldo Gangemi


Humanistic and social sciences are eagerly adopting the last AI hype. This is good news, since In recent years, humanistic and social sciences are eagerly adopting the last Artificial Intelligence (AI) hype. The crossbreeding with computer science and STEM offers an immense horizon of applications, able to sharpen the humanities’ sensitive grasp on society, art and the cognitive reality of the human being. The history of AI shows its originality in revisiting traditional ontological, epistemological and cognitive problems (e.g. the frame problem); moreover, computational methods keep being inspired by humanistic thinking (e.g. markup languages, schema/frame-based data structures, ontologies). These early efforts have begun producing a treasure trove of innovative methods able to improve our description of emulated cultural, social, and psychological worlds.

Humanistic AI (HAI) is a fairly novel branch aimed at integrating psychological, social and computational methods in a systematic way, in order to reframe the study of both the embodied human mind and social and cultural contexts, as well as their reciprocal relations.

Applications of AI techniques to humanities range from the classification, exploration, management, and preservation of cultural heritage, archives, or demo-ethno-anthropological materials, to the exploitation of novel models (e.g. distributed neural networks) and platforms (e.g. robotics) often inspired by biological models: the central nervous system, the human sensory-motor system, etc.

On one hand, HAI leverages the advantages of existing computational hybridisations (e.g. semantic technologies, digital humanities, social data mining and visualization, machine learning on large humanistic archives, datasets and corpora). On the other hand, a unique feature of HAI consists in its potential to suggest original analyses, alternative to the mainstream ones (e.g. connectionist vs. symbolic linguistics). Humanistic AI is not limited to giving new answers to traditional research questions, but allows to completely rephrase the outdated ones. Never losing sight of the human factor and the social context, new (ecological, evolutionary) approaches are suited to the investigation of the embodied and culturally-aware human mind alongside the analysis of its interaction with cognitive artifacts in a given environment (e.g. “cultural affordances”, evolutionary perspectives on human societies).

Research on foundations and application of AI techniques in Humanistic AI


Semantic and language technologies, cognitive computing and knowledge extraction

  • Semantic Web methods, tools, and resources
  • Web science
  • Embodied cognition
  • Ontology and linked data design
  • Knowledge graph integration
  • Automated extraction of ontologies and knowledge graphs from text and multimodal sources
  • Human-level natural language understanding
  • Cognitive computing
  • Frame semantics
  • Social robotics
  • Corpus and text analysis/mining
  • Multilingual and language independent text representations
  • Neural machine translation engines
  • Quality estimation for (machine) translation
  • Conversational interfaces

AI in the management and preservation of cultural heritage, libraries, archives, works

  • Digitization and semantic representation of museum heritage data and scientific collections
  • Computational analysis of linguistic, literary, narratological, philological and historical text
  • Cultural heritage knowledge graphs
  • Machine learning in audio-visual and artwork documentation
  • Deep learning for deciphering ancient scripts
  • Image analysis and 3D modelling for archaeological data interpretation
  • Historical network analysis
  • Sentiment analysis in spatio-temporal narratives and geographical perception
  • Data visualization in Digital Humanities
  • Digital cultural object collections and digital archives
  • Data management, long-term preservation, and integration of (heterogeneous) catalogs, data, metadata, and software architectures
  • Text encoding and markup languages
  • Transcription of audio and video recordings
  • Automated or assisted classification of content: artworks, audiovisual, demo-ethno-anthropological objects, etc.
  • Digitization of images
  • Prosopographic databases
  • Semantic publishing
  • Digital critical editions of sources
  • Semantic metadata design for digital editions
  • Digital acquisition of primary sources for improved transcription and readability

Creativity and cultural change

  • Human creative thinking and artificial creativity
  • Automated metaphor generation
  • Automated analysis of macroscopic trends in cultural change
  • Transdisciplinarity
  • Hybridisation of computational and humanistic methods
  • Historical development of artificial intelligence and cybernetics