Mapping the role and benefits of metacognition in Human-AI collaboration. A systematic review

作者
Jorge Mario Alvarino Diaz,Paul M. Salmon,Assoc. Prof. Gemma Read,Nicki Newton
标识
DOI:10.17605/osf.io/kr54a
摘要

Metacognition offers a promising response to mitigating the risks associated with human inherent errors in judgement and general overreliance on mental heuristics when making judgements under uncertainty (decision-making under pressure) (Carpenter et al., 2019; Lee et al., 2010). Recent research shows that human biases may be significantly amplified when directly interacting with biased AI systems or by merely observing flawed AI generated insights (Alabed et al., 2022; Glickman & Sharot, 2024). Given AI’s lack of physiological cues typically employed in human interaction and humans’ positive perception of AI judgements, AI users are frequently unaware of the extent of the AI’s influence, making them more vulnerable to it, causing a snowball effect where small errors in judgement escalate into much larger ones (Glickman & Sharot, 2024). AI systems lack metacognitive capabilities, and its expanding computational resources may compound this issue even further. Consequently, enhanced metacognitive awareness may empower AI users to judiciously monitor and assess their thought processes as they increasingly query, collaborate with and receive inputs from AI systems. This can support robust analyses, enhanced decision-making and improved learning. Metacognitive aware AI users may apply informed intuition to expertly draw upon their experience and knowledge while recognising their own and the AI’s cognitive limitations; and acknowledge the need for divergent perspectives and information. By cultivating heightened metacognitive practices, experts and decision-makers can critically assess the efficacy of their own and the AI's knowledge, judgment, confidence levels, and intuition within specific contexts, ultimately leading to more balanced and accurate outcomes in collaborative settings (Sidra & Mason, 2024). Research on this topic may help map out the role and benefits of metacognition in Human-AI Collaboration. It could also contribute to other areas such as human factors in the design of AI tools, epistemology, and cognitive science.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
2秒前
wct发布了新的文献求助10
2秒前
无极微光的应助被Sky千变采纳,获得20
2秒前
白南完成签到,获得积分10
3秒前
weng发布了新的文献求助10
4秒前
wxh完成签到,获得积分10
4秒前
4秒前
慕念发布了新的文献求助10
4秒前
鱼湘完成签到,获得积分10
4秒前
LiangKeCheng发布了新的文献求助10
5秒前
tangerine完成签到,获得积分20
5秒前
keliya完成签到 ,获得积分10
7秒前
今后的应助被辛勤泥猴桃采纳,获得10
7秒前
8秒前
bff完成签到,获得积分10
9秒前
迷途书童发布了新的文献求助10
9秒前
Akim的应助被wct采纳,获得10
9秒前
XuHT完成签到,获得积分10
9秒前
xxx完成签到 ,获得积分10
10秒前
八十发布了新的文献求助10
10秒前
烨无殇完成签到,获得积分10
10秒前
华仔的应助被典雅翅膀采纳,获得10
10秒前
10秒前
hh完成签到 ,获得积分10
11秒前
11秒前
苹果芷雪发布了新的文献求助10
11秒前
彩色蚂蚁发布了新的文献求助10
11秒前
Atheist的应助被BGW采纳,获得10
12秒前
爆米花的应助被webel采纳,获得10
12秒前
光亮凉面完成签到,获得积分10
13秒前
13秒前
科研通AI2S的应助被科研通管家采纳,获得10
13秒前
天天快乐的应助被科研通管家采纳,获得10
13秒前
13秒前
科目三的应助被科研通管家采纳,获得10
13秒前
CodeCraft的应助被zhangerya采纳,获得10
13秒前
大猫发布了新的文献求助20
13秒前
风格的应助被科研通管家采纳,获得20
13秒前
充电宝的应助被科研通管家采纳,获得10
14秒前
Hello的应助被科研通管家采纳,获得10
14秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
拉伸高分子熔体中缠结行为的模拟和小角中子散射研究 1500
The World Atlas of Wine 9th edition 1000
奈特基础组织学:配套组织病理学(第4版)(Netter's Essential Histology With Correlated Histopathology) 800
John McDowells Bildungstheorie 800
八旗制度の研究 600
Analyses de reconnaissances de dettes passées devant les échevins d'Ypres (1249-1291) 600
热门求助领域 (近24小时)
化学 材料科学 医学 生物 纳米技术 计算机科学 工程类 有机化学 化学工程 生物化学 物理 复合材料 内科学 催化作用 光电子学 人工智能 细胞生物学 无机化学 物理化学 数学
热门帖子
关注 科研通微信公众号,转发送积分 8160632
求助须知:如何正确求助?哪些是违规求助? 9555512
关于积分的说明 21728801
捐赠科研通 7750444
什么是DOI,文献DOI怎么找? 3427429
关于科研通互助平台的介绍 2552456
邀请新用户注册赠送积分活动 2471817