An ecosystemic view on information, data, and knowledge: insights on agential AI and relational ethics

AI and Ethics, Год журнала: 2025, Номер unknown

Опубликована: Фев. 17, 2025

Язык: Английский

No Consciousness? No Meaning (and no AGI)!

Опубликована: Апрель 2, 2025

The recent developments in artificial intelligence (AI), particularly light of the impressive capabilities transformer-based Large Language Models (LLMs), have reignited discussion cognitive science regarding whether computational devices could possess semantic understanding or they are merely mimicking human intelligence. Recent research has highlighted limitations LLMs’ reasoning, suggesting that gap between mere symbol manipulation (syntax) and deeper (semantics) remains wide open. While LLMs overcome certain aspects grounding problem through feedback, still lack true understanding, struggling with common-sense reasoning abstract thinking. This paper argues while adding sensory inputs embodying AI sensorimotor integration environment might enhance its ability to connect symbols real-world meaning, this alone would not close syntax semantics. True meaning-making also requires a connection subjective experience, which current lacks. path AGI must address fundamental relationship manipulation, data processing, pattern matching, probabilistic best guesses knowledge conscious experience. A transition from can occur only if it possesses is closely tied understanding. Recognition furnish new philosophical insights into longstanding practical questions for theories biology provide more meaningful tests than Turing test.

Язык: Английский

Suitability of Chinese GenAI Platforms for Early Childhood Education: A Multifaceted Evaluation

Опубликована: Апрель 21, 2025

Generative artificial intelligence (GenAI) offers transformative potential for early childhood education (ECE), yet concerns remain regarding its suitability and ethical implications young children. This study evaluated 10 freely available Chinese GenAI platforms in ECE, considering technical performance, pedagogical adaptability, ethical/safety considerations. Using mixed methods (quantitative scoring qualitative content analysis), were assessed on multimodal support, response speed, teaching generation, activity organization, personalization, compliance, appropriateness, algorithmic fairness. Results showed significant variability with some excelling aspects while others demonstrated stronger adaptability. Doubao iFlytek Spark strong DeepSeek excelled However, all presented areas improvement, particularly support the transparency of guidelines. research a novel framework evaluating which allows more nuanced context-specific assessment compared to existing frameworks. highlights need specialized ECE datasets, transparency, robust guidelines protect learners. Findings provide practical guidance educators, developers, policymakers navigating GenAI's development.

Язык: Английский