1 Why people judge humans differently from machines The role of perceived agency and

2025-04-30 1 0 1MB 8 页 10玖币

侵权投诉

Why people judge humans differently from

machines: The role of perceived agency and

experience

Jingling Zhang∗, Jane Conway†, C´

esar A. Hidalgo∗‡

∗Center for Collective Learning, ANITI, IAST, TSE, IRIT, University of Toulouse

†Centre for Creative Technologies and School of Psychology, University of Galway

‡Center for Collective Learning, CIAS, Corvinus University

Abstract—People are known to judge artiﬁcial intelligence

using a utilitarian moral philosophy and humans using a moral

philosophy emphasizing perceived intentions. But why do people

judge humans and machines differently? Psychology suggests that

people may have different mind perception models of humans and

machines, and thus, will treat human-like robots more similarly

to the way they treat humans. Here we present a randomized ex-

periment where we manipulated people’s perception of machine

agency (e.g., ability to plan, act) and experience (e.g., ability to

feel) to explore whether people judge machines that are perceived

to be more similar to humans along these two dimensions more

similarly to the way they judge humans. We ﬁnd that people’s

judgments of machines become more similar to that of humans

when they perceive machines as having more agency but not

more experience. Our ﬁndings indicate that people’s use of

different moral philosophies to judge humans and machines can

be explained by a progression of mind perception models where

the perception of agency plays a prominent role. These ﬁndings

add to the body of evidence suggesting that people’s judgment

of machines becomes more similar to that of humans motivating

further work on dimensions modulating people’s judgment of

human and machine actions.

I. INTRODUCTION

Do people judge human and machine actions equally?

Recent empirical studies suggest this is not the case. In fact,

several studies have shown that people make strong differences

when judging humans and machines.

Consider the recent experiments from Malle et al. (2015)

asking people to judge a trolley problem [10], [15]. In a trolley

problem, people can pull a lever to deviate an out-of-control

trolley sacriﬁcing a few people to save many. Malle et al.

(2015) found that people expected robots to pull the lever and

act utilitarianly (sacriﬁce one person to save four) compared

to humans (which were not judged as severely for not pulling

the lever) [21]. This idea was expanded by [14]. Using a

set of over 80 randomized experiments comparing people’s

reactions to the actions of humans and machines, the authors

concluded that people judge humans and machines using

different moral philosophies: a consequentialist philosophy

(focused on outcomes) for machines and a moral philosophy

focused more on intention when it comes to humans.

But why do people use different moral philosophies to

judge humans and machines? Psychology suggests that people

may perceive the minds of machines and humans differently

[8], [11], and therefore, may treat more human-like robots

more similarly to the way they treat humans [9]. This idea

is related to various experiments where robots were endowed

with human-like features [19], [27], [16], [24], [22], [28], [29],

[23]. For instance, Powers and Kiesler (2006) used a robot with

tunable chin length and tone of voice to explore the connection

between the robot’s appearance and its perceived personality

[24]. Waytz et al. (2014) compared anthropomorphized and

non-anthropomorphized self-driving cars to show that people

trust the anthropomorphized self-driving cars more [27]. Malle

et al. (2016) explored the impact of a robot’s appearance in

people’s judgment of moral actions (trolley problem), ﬁnding

that people judge more human-like robots more similarly to

the way they judge humans [22]. Yet, these experiments did

not provide an explicit quantitative mind perception model

explaining people’s judgment of more and less human-like

machines.

Here we explore how perceived agency and experience,

two key dimensions of mind perception [11], affect people’s

judgments of machines.

Agency is related to an agents ability to plan (e.g., to create

a strategy for action that considers potential consequences)

and to act (e.g., the capacity to affect or control the immediate

environment). Thus, agency is related to moral responsibility

for performed actions (higher agency, higher expected respon-

sibility) [17].

Experience, in the context of this paper, is used to describe

the ability to feel (e.g., the ability to experience sensations

such as pain, sadness, guilt, or anger). It is, thus, related to

the concept of moral status (not to be confused with the idea

of expertise) and to the right of an agent to be treated with

dignity.

These two dimensions represent a basic mind perception

model that has been used previously to explain the cognition

and behavior of alters using representations of their perceived

mental abilities [3], [4], [8], [11]. Usually, mind perception

models involve low dimensional representations of an alter’s

characteristics, such as the warmth and competence model

used to explain stereotypes [5]. That model, for instance, says

that people tend to protect those high in warmth and low in

competence (e.g., babies) but fear those high in competence

and low in warmth (e.g., killer robots).

arXiv:2210.10081v2 [cs.CY] 19 Sep 2023

Fig. 1. A. Schematic Illustration of mind perception models of humans and machines. Research by Gray et al.(2007) showed that humans are perceived

as having high levels of agency and experience while machines are perceived as having intermediate levels of agency and low levels of experience [11].

B. Schematic illustration of our experimental design. The experimental design involved manipulating people’s perceptions of machines by changing their

descriptions. Our experiment involved four machine conditions (from high agency and experience (A+, E+) to low agency and experience (A-, E-).C. The

experimental manipulation involved changing a few words in the description of machines. A+ and E+ words are associated with high agency and experience.

A- and E- words are associated with low agency and experience. D. Result of manipulation check. Mean ratings of agency and experience with 95 percent

conﬁdence intervals. Agency ratings are based on the perceived ability to self-restraint, tell right from wrong, and remember things. Experience ratings are

based on three factors: the perceived ability to feel afraid or fearful, being aware of things, and having a personality. These factors come from Gray et al.

(2007) [11].

Here, we present evidence for the mind perception model

used by Gray et al. (2007), which decomposes mental abilities

into agency and experience. This is not completely unrelated

to other models, such as the warmth and competence model

[5], since competence is related to agency. We use this model

to explore how people’s mind perception of machines affects

their moral reasoning [11]. In fact, Gray and Wegner have

shown that moral reasoning about agents (e.g., deserving pun-

ishment for wrongdoing) correlates with these two dimensions.

This is aligned with ideas in philosophy, law, and cognitive

science, which relate moral responsibility to agency [2], [26]

and rights and privileges to experience.

Agency and experience provide an interesting framework to

explain people’s judgment of machines because humans and

machines occupy different positions in this mind perception

space (Figure 1 A). This tells us that we should expect people

to judge humans and machines differently and that we could

expect people to judge humans and machines more similarly

when they come together in a person’s mind perception space.

In our experiment, we manipulated people’s perception of

machines along the mind perception space deﬁned by agency

and experience using an experimental design involving four

machine conditions (See Figure 1. B). The four machine

conditions range from high-agency and high-experience ma-

chine A+, E+to a low-agency and low-experience machine

A−, E−(passing through a high-agency low-experience ma-

chine A+, E−and a low-agency high-experience machine

A−, E+). Overall, we ﬁnd that manipulating the perceived

agency of machines, but not its perceived experience, changes

people’s reaction to machine actions so that their judgment be-

comes more similar to how people judge humans. Technically,

people’s judgment of more human-like machines depends less

on the perceived outcome of a scenario (the perceived level of

harm) and more on its perceived intention. We interpret this

as evidence that people’s switch between these different moral

philosophies when judging humans and machines is partly

explained by people’s mental models and that these different

modes of judgment are connected through a progression

of intermediate steps that maps onto the agency-experience

mind perception space. Our ﬁndings provide evidence of this

progression by showing that moving people’s perception of

machines along this mind perception space correlates with

changes in judgment. These ﬁndings contribute to the grow-

ing literature exploring people’s judgment of machines and

artiﬁcial intelligence [14], [22], [1], [25], [18].

II. METHODS

We recruited 719 participants (49.24% female, 49.93%

male, 0.83% other; mean age 37.79, SD = 14) of UK nation-

ality from Proliﬁc (https://www.proliﬁc.co). Participants were

compensated £1.00 for the 8-minute study. After removing the

answers of participants who failed the attention check or who

provided the same numerical answer for all questions, we got

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

1Whypeoplejudgehumansdifferentlyfrommachines:TheroleofperceivedagencyandexperienceJinglingZhang∗,JaneConway†,C´esarA.Hidalgo∗‡∗CenterforCollectiveLearning,ANITI,IAST,TSE,IRIT,UniversityofToulouse†CentreforCreativeTechnologiesandSchoolofPsychology,UniversityofGalway‡CenterforCollectiveLearning,CIAS,C...

展开>> 收起<<

1 Why people judge humans differently from machines The role of perceived agency and.pdf

共8页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

1 Why people judge humans differently from machines The role of perceived agency and

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: