The paper argues that (a) Machine Learning (ML) constitutes a cultural tool capable of learning through perceiving patterns in data, (b) the kind of learning ML is capable of nevertheless constitutes a more circumscribed kind of learning compared with how that concept has been interpreted in sociocultural (S-c) theory; and, (c) the development of ML is therefore further extending and distributing the complex relationship between human and machine cognition and learning. The paper explores these contentions by firstly, providing a broad-based account of the conception of cultural tools in S-c Theory. Secondly, offering a genealogy of ML, including the model of learning that underpins ML and highlights the challenge that a cultural too capable of some kind of learning presents for the extant S-c conception of a cultural tool. Thirdly, identifying the new human-machine working-learning problem the ML model of learning is generating. Finally, argues the concept of recontextualization offers a way to address that problem by providing a holistic perspective on the relationship between ML and IPL models of learning. In making this argument the paper distinguishes between the ML predictive and the Chat GPT answer to question(s) model of learning.

该论文认为，(a) 机器学习 (ML) 构成了一种能够通过感知数据模式进行学习的文化工具，(b) 与该概念的学习方式相比，ML 能够进行的学习类型仍然构成了一种更受限制的学习类型。已被社会文化（Sc）理论解释；(c) 因此，机器学习的发展正在进一步扩展和分布人类与机器认知和学习之间的复杂关系。本文首先对 Sc 理论中文化工具的概念进行了广泛的阐述，从而探讨了这些争论。其次，提供机器学习的谱系，包括支撑机器学习的学习模型，并强调具有某种学习能力的文化对现有的文化工具 Sc 概念提出的挑战。第三，识别 ML 学习模型正在生成的新人机工作学习问题。最后提出概念重新情境化提供了一种解决该问题的方法，它提供了关于 ML 和 IPL 学习模型之间关系的整体视角。在提出这一论点时，本文区分了 ML 预测学习模型和 Chat GPT 问题回答学习模型。