Scholarlink

<p>What makes AI controversial? This article reports on an international expert consultation that we conducted between 2021 and 2023 to analyse recent controversies about AI in four countries: the United Kingdom, France, Germany, and Canada. Adopting a standpoint approach, this online consultation was designed to activate the collective intelligence of AI &amp; Society expert communities to identify the most important but possibly overlooked controversies about AI during a 10&#xa0;year period, 2012–2022. Based on our analysis of the consultation responses, we discern two contrasting discursive dynamics of problematisation in relation to AI during the relevant period: one revolving around the demonstration of “AI frictions”, the public staging of situated problems arising from the deployment of AI-based applications such as recommendation algorithms and image recognition in specific professional contexts and everyday environments. The other is associated with generalised and abstract forms of controversiality regarding generative AI technologies (GAIT), such as ChatGPT. Through a relational analysis of ‘topic-friction’ couplings latent in the consultation responses, we go on to distinguish two different types of ‘associational profiles’ (Marres <CitationRef CitationID="CR47">2015</CitationRef>) for AI controversies: while situated AI frictions are defined by sustained connections with specific topics, GAIT controversies were marked by broad and thin connections across a broad range of topics. In capturing the decline and ascent of these AI controversy types, our consultation enables us to characterise a watershed moment in public debates about AI as a shift from deep to wide controversies. We conclude with the reminder that deep AI frictions have distinctive potential for democratic governance, because they tend to implicate organised communities and stakeholders with specific, demonstrable stakes in technology deployment. Wide GAIT controversies, by contrast, distribute concern across many domains in a diffuse manner, implicating general publics rather than affected communities, which creates special challenges for democratic accountability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A tale of two controversies: situated AI frictions and the generative AI tsunami

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月4日

查看全文
  • Chiara Poletti,
  • Noortje Marres,
  • Donato Ricci,
  • Christian Katzenbach,
  • Lena Marie Henkes,
  • Guillaume Dandurand

摘要

What makes AI controversial? This article reports on an international expert consultation that we conducted between 2021 and 2023 to analyse recent controversies about AI in four countries: the United Kingdom, France, Germany, and Canada. Adopting a standpoint approach, this online consultation was designed to activate the collective intelligence of AI & Society expert communities to identify the most important but possibly overlooked controversies about AI during a 10 year period, 2012–2022. Based on our analysis of the consultation responses, we discern two contrasting discursive dynamics of problematisation in relation to AI during the relevant period: one revolving around the demonstration of “AI frictions”, the public staging of situated problems arising from the deployment of AI-based applications such as recommendation algorithms and image recognition in specific professional contexts and everyday environments. The other is associated with generalised and abstract forms of controversiality regarding generative AI technologies (GAIT), such as ChatGPT. Through a relational analysis of ‘topic-friction’ couplings latent in the consultation responses, we go on to distinguish two different types of ‘associational profiles’ (Marres 2015) for AI controversies: while situated AI frictions are defined by sustained connections with specific topics, GAIT controversies were marked by broad and thin connections across a broad range of topics. In capturing the decline and ascent of these AI controversy types, our consultation enables us to characterise a watershed moment in public debates about AI as a shift from deep to wide controversies. We conclude with the reminder that deep AI frictions have distinctive potential for democratic governance, because they tend to implicate organised communities and stakeholders with specific, demonstrable stakes in technology deployment. Wide GAIT controversies, by contrast, distribute concern across many domains in a diffuse manner, implicating general publics rather than affected communities, which creates special challenges for democratic accountability.