From "AI" to Probabilistic Automation: How Does Anthropomorphization of Technical Systems Descriptions Influence Trust?

Classified as probably not a substantial claim. May combine two claims.

2529 choices were made of the anthropomorphized product description, and 2439 choices were made of de-anthropomorphized product descriptions.

“2529 choices were made of the anthropomorphized product description, and 2439 choices were made of de-anthropomorphized product descriptions.”

May combine two claims.

In the aggregate personal trust analysis, there is a preference for the de-anthropomorphized description of Judy, but for the anthropomorphized description of JurisDecide with the product as a significant dependent variable (chi-square = 8.50; n = 620; p = .003).

“and a preference for the de-anthropomorphized description of Judy, but for the anthropomorphized description of JurisDecide with the product as a significant dependent variable ( 𝜒 2 = 8.50; 𝑛 = 620; 𝑝 = .003).”

May combine two claims.

In Study 1 personal trust, the distribution of preferences for anthropomorphized descriptions was consistent with the H0 distribution (chi-square = 0.63; df = 14; p = .43), meaning there was no statistically significant preference for either anthropomorphized or de-anthropomorphized descriptions overall.

“The Chi-squared goodness-of-fit test showed that the distribution of preferences for anthropomorphized descriptions was consistent with the H0 distribution ( 𝜒 2 = 0.63; df = 14; 𝑝 = .43), meaning there was no statistically significant preference for neither anthropomorphized nor de-anthropomorphized descriptions overall .”

May combine two claims.

In Study 1 general trust, the distribution of anthropomorphized descriptions was consistent with the H0 distribution (chi-square = 0.70; df = 14; p = .40), meaning there was no statistically significant preference for either anthropomorphized or de-anthropomorphized descriptions overall.

“The distribution of anthropomorphized descriptions was consistent with the H0 distribution ( 𝜒 2 = 0.70; df = 14; 𝑝 = .40), meaning there was no statistically significant preference for neither anthropomorphized nor de-anthropomorphized descriptions overall .”

The quote only partly states this claim.

In Study 2 general trust, within the individual product pairs, there are no statistically significant preferences for either anthropomorphized or de-anthropomorphized descriptions.

“Within the individual product pairs, we see no statistically significant preferences for neither anthropomorphized nor de-anthropomorphized descriptions.”

May combine two claims.

The distribution of choices is consistent with the H0 distribution (chi2=1.63; df=14; p=.09), meaning there was no statistically significant preference for neither anthropomorphized nor de-anthropomorphized descriptions overall.

“The distribution is consistent with the H0 distribution ( 𝜒 2 = 1.63; df = 14; 𝑝 = .09), meaning there was no statistically significant preference for neither anthropomorphized nor de-anthropomorphized descriptions overall .”

• Also stated in: § 5.1 Observation 1: Overall, people are no more likely to choose anthropomorphized descriptions of products over de-anthropomorphized descriptions of probabilistic automation products. · p. 11 · plain

  “Across categories, we do not see a clear preference for anthropomorphized descriptions of products over de-anthropomorphized descriptions of products.”

May combine two claims.

In Study 1 personal trust, for the Judy/JurisDecide pair, there was a significant preference for the de-anthropomorphized description of the Judy system, and a preference for the anthropomorphized version of the JurisDecide system (chi-square = 5.12; N = 315; p = .02).

“Interestingly, for the Judy/JurisDecide (legal recommendations) pair, there was a significant preference for the de-anthropomorphized description of the Judy system, and a preference for the anthropomorphized version of the JurisDecide system ( 𝜒 2 = 5.12; 𝑁 = 315; 𝑝 = .02).”

May combine two claims.

In Study 2 personal trust, the distribution of preferences for anthropomorphized descriptions was consistent with the H0 distribution (chi-square = 1.57; df = 14; p = .56), meaning there was no statistically significant preference for either anthropomorphized or de-anthropomorphized descriptions overall.

“The distribution of preferences for anthropomorphized descriptions was consistent with the H0 distribution ( 𝜒 2 = 1.57; df = 14; 𝑝 = .56), meaning there was no statistically significant preference for neither anthropomorphized nor de-anthropomorphized descriptions overall .”

May combine two claims.

In the aggregate personal trust analysis, between individual product pairs, there is a significant preference for the re-Commender/IntelliTrade (chi-square = 8.47; df = 1; p = .004) and MonAI/Cameron (chi-square = 8.31; df = 1; p = .004) pairs.

“Between individual product pairs, we see a significant preference for the re-Commender/IntelliTrade ( 𝜒 2 = 8.47; df = 1; 𝑝 = .004) and MonAI/Cameron ( 𝜒 2 = 8.31; df = 1; 𝑝 = .004) pairs”

The quote only partly states this claim.

In Study 1 personal trust, the re-Commender/IntelliTrade pair shows a significant preference for the anthropomorphized descriptions for both products (chi-square = 6.84; df = 1; p > .001).

“The re-Commender/IntelliTrade (recmmender systems) pair shows a significant preference for the anthropomorphized descriptions for both products ( 𝜒 2 = 6.84; df = 1; 𝑝 > .001).”

May combine two claims.

In Study 1 general trust, the Judy/JurisDecide pair reveals a preference for the anthropomorphized description of JurisDecide, but a preference for the de-anthropomorphized description of Judy with the product as a dependent variable (chi-square = 5.00; N = 315; p = .02).

“and the Judy/JurisDecide pair reveals a preference for the anthropomorphized description of JurisDecide, but a preference for the de -anthropomorphized description of Judy with the product as a dependent variable ( 𝜒 2 = 5.00; 𝑁 = 315; 𝑝 = .02).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Many of these assumptions are assumptions of the system's functionality.

“However, many of these assumptions are exactly that; assumptions of the system's functionality: 'It will be more correct because it works with photos for comparison, so the chance of error is smaller' (Study 2, de-ant. DermAI Scan).”

Probably not a substantial claim: may not pass S3 (load-bearing).

A few participants highlighted linguistic differences in product descriptions as motivating their choice.

“A few participants even highlighted linguistic differences in product descriptions as motivating their choice, albeit using different words than anthropomorphization: 'Option B provides a more engaging and descriptive presentation ' (Study 1, de-ant. AquaSentinel 6 ).”

Probably not a substantial claim: may not pass S3 (load-bearing).

A prevalent theme in the open text answers is that the participant has chosen 'the lesser of two evils'; meaning they are expressing deep skepticism of both products, but was forced, through the survey design, to choose one.

“A prevalent theme in the open text answers is that the participant has chosen 'the lesser of two evils'; meaning they are expressing deep skepticism of both products, but was forced, through the survey design, to choose one.”

May combine two claims.

The proportion of choices of anthropomorphized/de-anthropomorphized product descriptions did not differ significantly by socio-economic status in either study, neither in personal trust, nor in general trust.

“A chi-squared test showed that the proportion of choices of anthropomorphized/deanthropomorphized product descriptions did not differ significantly by socio-economic status in either study , neither in personal trust, nor in general trust.”

May combine two claims.

The proportion of choices of anthropomorphized/de-anthropomorphized product descriptions did not differ significantly by level of computer knowledge in either study, neither in personal trust, nor in general trust.

“The proportion of choices of anthropomorphized/de-anthropomorphized product descriptions did not differ significantly by level of computer knowledge in either study, neither in personal trust, nor in general trust.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The potential of biased probabilistic automation training data was mentioned many times as rationale for distrusting the system.

“The potential of biased probabilistic automation training data was mentioned many times as rationale for distrusting the system”

Probably not a substantial claim: may not pass S3 (load-bearing).

A lot of responses express that probabilistic automation is more appropriate for some tasks than for others.

“A lot of responses express that probabilistic automation is more appropriate for some tasks than for others.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Many responses in favor of the MonAI system over the Cameron system highlight that computers are better with numbers than texts.

“For instance, most responses in favor of the MonAI system in favor of the Cameron system highlight that 'Computers are better with numbers than texts. I would trust more an app with numbers than one who manage texts.' (Study 1, ant. MonAI Maker).”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The qualitative responses from the surveys circumstantiate and add detail to the quantitative results.

“The qualitative responses from the surveys circumstantiate and add detail to the quantitative results.”

May combine two claims.

The proportion of choices of anthropomorphized/de-anthropomorphized product descriptions did not differ significantly by gender in either study, neither in personal trust, nor in general trust.

“The proportion of choices of anthropomorphized/de-anthropomorphized product descriptions did not differ significantly by gender in either study, neither in personal trust, nor in general trust.”

May combine two claims.

For Study 2, preferences lean in different directions for several product pairs, and overall for the different categories (Cognizer, Agency, and Biological metaphors all elicit different preferences between personal and general trust in Study 2).

“In fact, we see for Study 2 that preferences (see Table 2) lean in different directions for several product pairs, and overall for the different categories ( Cognizer , Agency , and Biological metaphors all elicit different preferences between personal and general trust in Study 2).”

May combine two claims.

The differences between personal and general trust are small, and none of them are statistically significant in the overall comparison, except for the Agency category, which elicited 55% and 49.9% preference for the anthropomorphized descriptions in personal and general trust, respectively.

“The differences are small, however (e.g., 48.9% preference for anthropomorphized descriptions for personal trust vs. 50.8% preference for anthropomorphized descriptions in general trust for the Cognizer -category), and none of them are statistically significant in the overall comparison, except for the Agency -category, which elicited 55% and 49.9% preference for the anthropomorphized descriptions”

May combine two claims.

The groups 31-35 and 36-40 compose a larger proportion of participants, and these groups both show a strong preference for anthropomorphized descriptions, particularly in the Cognizer category.

“The groups 31-35 and 36-40 compose a larger proportion of participants, and these groups both show a strong preference for anthropomorphized descriptions, particularly in the Cognizer -category.”

May combine two claims.

In Study 1, for the Cognizer category, 56.5% preferred the anthropomorphized description for personal trust, a statistically significant preference (χ2=10.99, p<.001).

“Cognizer | 642 | 363 | 279 | 56.5 | 10.99 | <.001 ∗”

Probably not a substantial claim: may not pass S3 (load-bearing).

The coding of the open-ended responses resulted in a list of more than 100 different codes at very different levels of abstraction.

“The result of the coding was a list of more than 100 different codes at very different levels of abstraction”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A few participants highlighted linguistic or semantic features of the descriptions as determining factors for their choice, so it is possible that this would have impacted the results to some degree.

“A few participants highlighted linguistic or semantic features of the descriptions as determining factors for their choice (see section 5.1), so it is possible that this would have impacted the results to some degree.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In Study 1, MonAI Maker's anthropomorphized description used cognizer language ('identifying ways to save money') and its de-anthropomorphized version used 'providing suggestions', which is still agentive language.

“For example, in Study 1, MonAI Maker is described as identifying ways to save money , a cognizer description and de-anthropomorphized as providing suggestions instead. This is still agentive language.”

May combine two claims.

In Study 1, for the Cognizer category, 55.0% preferred the anthropomorphized description for general trust, a statistically significant preference (χ2=6.38, p=.012).

“642 | 353 | 289 | 55.0 | 6.38 | .012 ∗”

The quote only partly states this claim.

In the aggregate across both studies, across the anthropomorphization categories, personal trust did not differ significantly (χ2=4.96, N=4985, p=.17).

“Personal trust: 𝜒 2 = 4.96; 𝑁 = 4985; 𝑝 = .17”

May combine two claims.

In Study 1, among participants with a Master's degree, 54.7% preferred the anthropomorphized description for personal trust, a split significantly different from chance (χ2 = 5.46, p = .019).

“| Master's | 616 | 337 | 279 | 54.7 | 5.46 | .019 ∗ |”

The quote only partly states this claim.

In the aggregate across both studies, across the anthropomorphization categories, general trust did not differ significantly (χ2=0.52, N=4968, p=.91).

“General trust: 𝜒 2 = 0.52; 𝑁 = 4968; 𝑝 = .91”

The quote only partly states this claim.

In the aggregate across both studies, the Cognizer category showed a statistically significant preference for anthropomorphized descriptions in personal trust (52.3%, χ2=5.37, p=.020).

“Cognizer | 1252 | 667 | 585 | 52.3 | 5.37 | .020 ∗”

May combine two claims.

In Study 1, among participants self-identifying as between average and high socio-economic status, 46.3% preferred the anthropomorphized description for personal trust, a split significantly different from chance (χ2 = 4.01, p = .045).

“| Between average/high | 728 | 337 | 391 | 46.3 | 4.01 | .045 ∗ |”

May combine two claims.

In Study 2, among participants self-identifying as average socio-economic status, 53.1% preferred the anthropomorphized description for general trust, a split significantly different from chance (χ2 = 4.50, p = .034).

“| Average | 1264 | 648 | 616 | 51.3 | 0.81 | .37 | 1152 | 612 | 540 | 53.1 | 4.50 | .034 ∗ |”

The quote only partly states this claim.

Among participants with a Master's degree, personal trust differed significantly between the anthropomorphized and de-anthropomorphized conditions (χ2 = 8.90; p = .003).

“| Master's | 1360 | 735 | 625 | 54.0 | 8.90”

May combine two claims.

In the aggregate data, among participants self-identifying as average socio-economic status, 52.2% preferred the anthropomorphized description for general trust, a split significantly different from chance (χ2 = 4.57, p = .034).

“| Average | 2480 | 1277 | 1203 | 51.5 | 2.21 | .14 | 2368 | 1236 | 1132 | 52.2 | 4.57 | .034 ∗ |”

Classified as probably not a substantial claim. May combine two claims.

Anthropomorphizing technical systems has been criticized for setting the wrong expectations and causing over-reliance in technology.

“This phenomenon - anthropomorphizing 1 technical systems - has been criticized for setting the wrong expectations and causing over-reliance in technology [21, 45, 46].”

• cites [21] [21] Francis Hunger. 2023. Unhype Artificial 'Intelligence'! A proposal to replace the deceiving terminology of AI. (2023). https://doi.org/10.5281/ ZENODO.7524492 -Not checked yet

  Not checked in the free run.

• cites [45] [45] Murray Shanahan. 2022. Talking about large language models. arXiv preprint arXiv:2212.03551 (2022). - sourceNot checked yet

  Not checked in the free run.

• cites [46] [46] Matthew Shardlow and Piotr Przybyła. 2022. Deanthropomorphising NLP: Can a Language Model Be Conscious? arXiv preprint arXiv:2211.11483 (2022). -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable).

In the context of probabilistic automation systems, the conceptual metaphor or 'pitch' of the system's functionality might play an especially compelling role, given the complexity of such systems.

“In the context of probabilistic automation systems, the conceptual metaphor [11, 33] or 'pitch' of the system's functionality might play an especially compelling role, given the complexity of such systems [31].”

• cites [31] [31] Markus Langer, Tim Hunsicker, Tina Feldkamp, Cornelius J König, and Nina Grgić-Hlača. 2022. 'Look! It'sa computer program! It's an algorithm! It's AI!': Does terminology affect human perceptions and evaluations of algorithmic decision-making systems?. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1-28. - sourceNot checked yet

  Not checked in the free run.

May combine two claims.

By using personified language when referring to AI systems, we implicitly attribute human-like properties to them, which both makes them seem more powerful than they are while obscuring their potential negative effects.

“The argument here is that by using personified language when referring to AI systems, we also implicitly attribute human-like properties to them, which both makes them seem more powerful than they are while obscuring their potential negative effects.”

• cites [21] [21] Francis Hunger. 2023. Unhype Artificial 'Intelligence'! A proposal to replace the deceiving terminology of AI. (2023). https://doi.org/10.5281/ ZENODO.7524492 -Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

Prior studies investigated how conceptual metaphors influenced people's perception of algorithmic decision-making systems more broadly, as well as how anthropomorphic cues influence people's trust in robots, voice-assistants, and websites.

“Prior studies investigated how conceptual metaphors influenced people's perception of algorithmic decision-making systems more broadly [31], as well as how anthropomorphic cues influence people's trust in robots [9], voice-assistants [15], and websites [42].”

• cites [31] [31] Markus Langer, Tim Hunsicker, Tina Feldkamp, Cornelius J König, and Nina Grgić-Hlača. 2022. 'Look! It'sa computer program! It's an algorithm! It's AI!': Does terminology affect human perceptions and evaluations of algorithmic decision-making systems?. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1-28. - sourceNot checked yet

  Not checked in the free run.

• cites [9] [9] Lara Christoforakos, Alessio Gallucci, Tinatini Surmava-Große, Daniel Ullrich, and Sarah Diefenbach. 2021. Can robots earn our trust the same way humans do? A systematic exploration of competence, warmth, and anthropomorphism as determinants of trust development in HRI. Frontiers in Robotics and AI 8 (2021), 640444. -Not checked yet

  Not checked in the free run.

• cites [15] [15] Elizabeth Fetterolf and Ekaterina Hertog. 2022. It's Not Her Fault: Trust through Anthropomorphism among Young Adult Amazon Alexa Users. (2022). -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Anthropomorphism is often used to describe technical products, and it appears a rising trend in news coverage.

“With the advent of more advanced technical systems, anthropomorphism is often used to describe technical products (i.e. 'A.I. Shows Signs of Human Reasoning' [34]), and it appears a rising trend in news coverage [6].”

• cites [6] [6] Mercedes Bunz and Marco Braghieri. 2022. The AI doctor will see you now: assessing the framing of AI in news coverage. AI & society 37, 1 (2022), 9-22. -Not checked yet

  Not checked in the free run.

May combine two claims.

One direct consequence of anthropomorphization is misplaced trust, which in turn can lead to over-reliance on probabilistic automation systems.

“One direct consequence of anthropomorphization is misplaced trust, which in turn can lead to over-reliance on probabilistic automation systems [1, 12, 13, 21, 43].”

• cites [1] [1] Gavin Abercrombie, Amanda Cercas Curry, Tanvi Dinkar, and Zeerak Talat. 2023. Mirages: On anthropomorphism in dialogue systems. arXiv preprint arXiv:2305.09800 (2023). -Not checked yet

  Not checked in the free run.

• cites [12] [12] Ophelia Deroy. 2023. The Ethics of Terminology: Can We Use Human Terms to Describe AI? Topoi (2023), 1-9. -Not checked yet

  Not checked in the free run.

• cites [13] [13] Smit Desai and Michael Twidale. 2023. Metaphors in voice user interfaces: a slippery fish. ACM Transactions on Computer-Human Interaction 30, 6 (2023), 1-37. - sourceNot checked yet

  Not checked in the free run.

• cites [21] [21] Francis Hunger. 2023. Unhype Artificial 'Intelligence'! A proposal to replace the deceiving terminology of AI. (2023). https://doi.org/10.5281/ ZENODO.7524492 -Not checked yet

  Not checked in the free run.

• cites [43] [43] Arleen Salles, Kathinka Evers, and Michele Farisco. 2020. Anthropomorphism in AI. AJOB neuroscience 11, 2 (2020), 88-95. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

This is especially true of robots and embodied assistants.

“This is especially true of robots and embodied assistants [16, 25, 49].”

• cites [16] [16] Terrence Fong, Illah Nourbakhsh, and Kerstin Dautenhahn. 2003. A survey of socially interactive robots. Robotics and autonomous systems 42, 3 (2003), 143-166. -Not checked yet

  Not checked in the free run.

• cites [25] [25] Takayuki Kanda, Takayuki Hirano, Daniel Eaton, and Hiroshi Ishiguro. 2004. Interactive robots as social partners and peer tutors for children: A field trial. Human-computer interaction 19, 1 (2004), 61-84. -Not checked yet

  Not checked in the free run.

• cites [49] [49] Fumihide Tanaka, Aaron Cicourel, and Javier R Movellan. 2007. Socialization between toddlers and robots at an early childhood education center. Proceedings of the National Academy of Sciences 104, 46 (2007), 17954-17958. -Not checked yet

  Not checked in the free run.

May combine two claims.

While anthropomorphism may enhance user experience and trust, it also risks creating a false sense of the system's capabilities.

“While anthropomorphism may enhance user experience and trust (in fact, much of the literature on anthropomorphism and technology concerns using anthropomorphization to increase trust, e.g., [7, 8, 28]), 3 it also risks creating a false sense of the system's capabilities.”

• cites [7] [7] Qian Qian Chen and Hyun Jung Park. 2021. How anthropomorphism affects trust in intelligent personal assistants. Industrial Management & Data Systems 121, 12 (2021), 2722-2737. -Not checked yet

  Not checked in the free run.

• cites [8] [8] Nguyen Thi Khanh Chi and Nam Hoang Vu. 2023. Investigating the customer trust in artificial intelligence: The role of anthropomorphism, empathy response, and interaction. CAAI Transactions on Intelligence Technology 8, 1 (2023), 260-273. -Not checked yet

  Not checked in the free run.

• cites [28] [28] Taenyun Kim and Hayeon Song. 2021. How should intelligent agents apologize to restore trust? Interaction effects between anthropomorphism and apology attribution on trust repair. Telematics and Informatics 61 (2021), 101595. - sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Smart voice assistants lead children to overestimate the intelligence of these devices, trusting them, and deferring to them when making decisions.

“Smart voice assistants also lead children to overestimate the intelligence of these devices, trusting them, and deferring to them when making decisions [14].”

• cites [14] [14] Stefania Druga, Randi Williams, Cynthia Breazeal, and Mitchel Resnick. 2017. Hey Google is it OK if I eat you?: Initial Explorations in Child-Agent Interaction. In Proceedings of the 2017 Conference on Interaction Design and Children . ACM, 595-600. - sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The more life-like a probabilistic automation application is in terms of embodiment, physical presence, social presence, and appearance, the more persuasive it can become.

“The more life-like a probabilistic automation application is in terms of embodiment (the physical form of the system), physical presence, social presence, and appearance, the more persuasive it can become [2, 41].”

• cites [2] [2] Wilma A Bainbridge, Justin Hart, Elizabeth S Kim, and Brian Scassellati. 2008. The effect of presence on human-robot interaction. In Robot and Human Interactive Communication, 2008. RO-MAN 2008. The 17th IEEE International Symposium on . IEEE, 701-706. - sourceNot checked yet

  Not checked in the free run.

• cites [41] [41] Maaike Roubroeks, Jaap Ham, and Cees Midden. 2011. When artificial social agents try to persuade people: The role of social agency on the occurrence of psychological reactance. International Journal of Social Robotics 3 (2011), 155-165. -Not checked yet

  Not checked in the free run.

May combine two claims.

In the absence of technical understanding, humans develop their own simplified mental models of how a system works - models that are not always consistent with the actual functionalities of the system, and of which inaccurate versions can lead to consequences from mundanely inconvenient to more severe.

“In the absence of technical understanding, humans develop their own simplified mental models of how a system works - models that are not always consistent with the actual functionalities of the system, and of which inaccurate versions can lead to consequences from mundanely inconvenient to more severe [37].”

• cites [37] [37] Don Norman. 2013. The design of everyday things: Revised and expanded edition . Basic books. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Numerous academic papers have appeared which warn about the risks of overusing anthropomorphic language to describe LLMs and generative probabilistic automation technology.

“With the blight of publicly-available Large Language Models (LLMs) and generative probabilistic automation technology, numerous academic papers have appeared which warn about the risks of overusing anthropomorphic language to describe such technology [1, 12, 21, 43, 45, 46].”

• cites [1] [1] Gavin Abercrombie, Amanda Cercas Curry, Tanvi Dinkar, and Zeerak Talat. 2023. Mirages: On anthropomorphism in dialogue systems. arXiv preprint arXiv:2305.09800 (2023). -Not checked yet

  Not checked in the free run.

• cites [12] [12] Ophelia Deroy. 2023. The Ethics of Terminology: Can We Use Human Terms to Describe AI? Topoi (2023), 1-9. -Not checked yet

  Not checked in the free run.

• cites [21] [21] Francis Hunger. 2023. Unhype Artificial 'Intelligence'! A proposal to replace the deceiving terminology of AI. (2023). https://doi.org/10.5281/ ZENODO.7524492 -Not checked yet

  Not checked in the free run.

• cites [43] [43] Arleen Salles, Kathinka Evers, and Michele Farisco. 2020. Anthropomorphism in AI. AJOB neuroscience 11, 2 (2020), 88-95. -Not checked yet

  Not checked in the free run.

• cites [45] [45] Murray Shanahan. 2022. Talking about large language models. arXiv preprint arXiv:2212.03551 (2022). - sourceNot checked yet

  Not checked in the free run.

• cites [46] [46] Matthew Shardlow and Piotr Przybyła. 2022. Deanthropomorphising NLP: Can a Language Model Be Conscious? arXiv preprint arXiv:2211.11483 (2022). -Not checked yet

  Not checked in the free run.

May combine two claims.

Accurately priming the user and adjusting their expectations to the system is difficult, and simply providing performance metrics is not meaningful to the average user, given their lack of familiarity with the inner workings of the technologies that they interact with.

“Accurately priming the user and adjusting their expectations to the system is difficult, and simply providing performance metrics is not meaningful to the average user, given their lack of familiarity with the inner workings of the technologies that they interact with [26, 30].”

• cites [26] [26] Pranav Khadpe, Ranjay Krishna, Li Fei-Fei, Jeffrey T Hancock, and Michael S Bernstein. 2020. Conceptual metaphors impact perceptions of human-AI collaboration. Proceedings of the ACM on Human-Computer Interaction 4, CSCW2 (2020), 1-26. - sourceNot checked yet

  Not checked in the free run.

• cites [30] [30] Rafal Kocielnik, Saleema Amershi, and Paul N Bennett. 2019. Will you accept an imperfect ai? exploring designs for adjusting end-user expectations of ai systems. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems . 1-14. - sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Robots could even exert peer pressure over children; 7- to 9-year-old children had a tendency to echo the incorrect, but unanimous, responses of a group of robots to a simple visual task.

“For example, Vollmer et al. showed that robots could even exert peer pressure over children [51]. In their experiment, 7- to 9-year-old children had a tendency to echo the incorrect, but unanimous, responses of a group of robots to a simple visual task [51].”

• cites [51] [51] Anna-Lisa Vollmer, Robin Read, Dries Trippas, and Tony Belpaeme. 2018. Children conform, adults resist: A robot group induced peer pressure on normative social conformity. Science Robotics 3, 21 (2018), eaat7111. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Humans are capable of engaging socially with machines.

“For example, humans are capable of engaging socially with machines [24, 36, 40].”

• cites [24] [24] Katherine Isbister and Clifford Nass. 2000. Consistency of personality in interactive characters: verbal cues, non-verbal cues, and user characteristics. International journal of human-computer studies 53, 2 (2000), 251-267. -Not checked yet

  Not checked in the free run.

• cites [36] [36] Clifford Nass and Youngme Moon. 2000. Machines and mindlessness: Social responses to computers. Journal of social issues 56, 1 (2000), 81-103. - sourceNot checked yet

  Not checked in the free run.

• cites [40] [40] Byron Reeves and Clifford Nass. 1996. How people treat computers, television, and new media like real people and places . CSLI Publications and Cambridge university press. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In Study 1, 333 participants signed the consent form, 313 participants completed the survey fully, and 20 participants partially completed the survey.

“For Study 1, 333 participants signed the consent form, and 313 participants completed the survey fully, while 20 participants partially completed the survey.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

All partially completed survey responses were included in the analyses because they provide valid answers to the questions, and excluding these participants has no statistically significant impact on the results.

“We have included all partially completed survey responses in the analyses, as they provide valid answers to the questions. Excluding these participants has no statistically significant impact on the results.”

May combine two claims.

People are aware of the dangers of overattributing accountability to technology 'when harm comes to pass', but the dynamics are not well understood.

“Some research has shown that people are aware of the dangers of overattributing accountability to technology 'when harm comes to pass' [48], but the dynamics are not well understood.”

• cites [48] [48] Michael T Stuart and Markus Kneer. 2021. Guilty artificial minds: Folk attributions of mens rea and culpability to artificially intelligent agents. Proceedings of the ACM on Human-Computer Interaction 5, CSCW2 (2021), 1-27. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A pilot study with 37 participants recruited through personal networks was run, and only minor edits to the product descriptions were made to clarify misunderstandings as a result of the pilot study.

“We ran a pilot study with 37 participants recruited through personal networks. Only minor edits to the product descriptions were made to clarify misunderstandings as a result of the pilot study.”

Classified as probably not a substantial claim.

The products were paired in genres so they would be somewhat comparable, to enable apples-to-apples comparisons.

“The products were paired in genres, so they would be somewhat comparable (for instance, 'recommender systems' or 'online health diagnostics'), to enable apples-to-apples comparisons.”

Classified as probably not a substantial claim. May combine two claims.

This cognitive overestimation can result in the inappropriate application of probabilistic automation advice, potentially leading to harmful outcomes.

“This cognitive overestimation can result in the inappropriate application of probabilistic automation advice, potentially leading to harmful outcomes.”

Classified as probably not a substantial claim.

The questions were designed to reflect two essential questions for measuring trust identified by Hoffman et al. with two modifications.

“The questions were designed to reflect two essential questions for measuring trust identified by Hoffman et al. [18] with two modifications”

• cites [18] [18] Robert R Hoffman, Shane T Mueller, Gary Klein, and Jordan Litman. 2018. Metrics for explainable AI: Challenges and prospects. arXiv preprint arXiv:1812.04608 (2018). -Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

For each presentation of a product pair to a participant, the authors randomized which product of the pair would be presented in its anthropomorphized guise and which de-anthropomorphized, but there was always one of each.

“For each presentation of a product pair to a participant, we randomized which product of the pair would be presented in its anthropomorphize guise, and which de-anthropomorphized, but there was always one of each”

Classified as probably not a substantial claim.

Participants were randomly shown a combination of either an anthropomorphized description of Product A and a de-anthropomorphized description of Product B, or a de-anthropomorphized description of Product A and an anthropomorphized description of Product B.

“The participants were randomly shown a combination of either: [Product A: Anthropomorphized description] + [Product B: De-anthropomorphized description] or [Product A: De-anthropomorphized description] + [Product B: Anthropomorphized description]”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In Study 2, Group A, asked only about personal trust, consisted of 307 participants, of which 304 fully completed the survey; Group B, asked only about general trust/reliance, consisted of 314 participants, of which 300 fully completed the survey.

“Group A, who were asked only about personal trust, consisted of 307 participants, of which 304 fully completed the survey. Group B, who were asked only about general trust/reliance, consisted of 314 participants, of which 300 fully completed the survey.”

Classified as probably not a substantial claim.

To make it more likely that participants would understand trust in a somewhat similar way, an introductory text as well as a short definition of trust was provided with each product pair.

“To make it more likely that participants would understand trust in a somewhat similar way, an introductory text as well as a short definition of trust was provided with each product pair (see appendix, Section A).”

Probably not a substantial claim: may not pass S3 (load-bearing).

When probabilistic automation systems are perceived as autonomous agents, it raises complex questions about accountability.

“When probabilistic automation systems are perceived as autonomous agents, it raises complex questions about accountability [4, 21, 48].”

• cites [4] [4] Yochanan E Bigman, Adam Waytz, Ron Alterovitz, and Kurt Gray. 2019. Holding robots responsible: The elements of machine morality. Trends in cognitive sciences 23, 5 (2019), 365-368. -Not checked yet

  Not checked in the free run.

• cites [21] [21] Francis Hunger. 2023. Unhype Artificial 'Intelligence'! A proposal to replace the deceiving terminology of AI. (2023). https://doi.org/10.5281/ ZENODO.7524492 -Not checked yet

  Not checked in the free run.

• cites [48] [48] Michael T Stuart and Markus Kneer. 2021. Guilty artificial minds: Folk attributions of mens rea and culpability to artificially intelligent agents. Proceedings of the ACM on Human-Computer Interaction 5, CSCW2 (2021), 1-27. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Such misplaced trust can be particularly problematic in high-stakes scenarios, such as medical diagnosis or financial decision-making, where over-reliance on probabilistic automation can lead to significant consequences.

“Such misplaced trust can be particularly problematic in high-stakes scenarios, such as medical diagnosis or financial decision-making, where over-reliance on probabilistic automation can lead to significant consequences.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The recruitment database allowed the authors to create pre-screening criteria such as country of residence, self-assessed socio-economic status, and ethnicity, to reach as diverse a group as possible.

“This database allowed us to create pre-screening criteria such as country of residence, self-assessed socio-economic status, and ethnicity, to reach as diverse a group as possible”

Probably not a substantial claim: may not pass S3 (load-bearing).

All participants signed a consent form that their anonymous answers could be used for research purposes.

“All participants signed a consent form that their (anonymous) answers could be used for research purposes.”

Classified as probably not a substantial claim.

People are prone to form beliefs even based on short descriptions.

“broad assumptions like this emphasize the importance of conveying accurate expectations of the system's functionality, because people are prone to form beliefs even based on short descriptions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In this case, the motivation appears to be identifying which product has lower stakes, or less impact if the system somehow fails.

“In this case, the motivation appears to be identifying which product has lower stakes , or less impact if the system somehow fails”

May combine two claims.

The aggregate preference for anthropomorphized descriptions in the Cognizer category is a spillover effect: in Study 1, the preference in the Cognizer category is so strong (56.5% and 55% for personal and general trust, respectively) that despite a very slight negative preference in personal trust in Study 2 (49.8%) and a weaker preference in general trust (50.9%) the preference carries over.

“This is a spillover effect: in Study 1, the preference for anthropomorphized descriptions in the Cognizer -category is so strong (56.5% and 55% for personal and general trust, respectively), that despite a very slight negative preference in personal trust in Study 2 (49.8%) and a weaker preference in general trust (50.9%) the preference carries over.”

May combine two claims.

Both product genre and type of anthropomorphization influence how people immediately perceive a product based on its description.

“This points to the conclusion that both product genre and type of anthropomorphization influence how people immediately perceive a product based on its description.”

May combine two claims.

People are critical about use context, risks, impacts, and human involvement.

“People are critical about use context, risks, impacts, and human involvement”

Probably not a substantial claim: may not pass S3 (load-bearing).

A recent survey has indicated a generation gap in probabilistic automation-acceptance.

“A recent survey has indicated a generation gap in probabilistic automation-acceptance [50]”

• cites [50] [50] Peter Tsai. 2023. The AI Generation Gap: Millennials Embrace AI, Boomers Are Skeptical - pcmag.com. https://www.pcmag.com/news/the-aigeneration-gap-millennials-embrace-ai-boomers-are-skeptical. [Accessed 22-01-2024]. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The study only explores a small part of the overarching question 'What are the effects of anthropomorphization of probabilistic automation systems?'

“We acknowledge study only explores a small part of the overarching question 'What are the effects of anthropomorphization of probabilistic automation systems?'”

Classified as probably not a substantial claim. May combine two claims.

The subgroups are small (26 participants total for the 61-65 group, and 37 for the 66+ group), so we refrain from making general conclusions on the basis of this study.

“The subgroups are small, however, (26 participants total for the 61-65 group, and 37 for the 66+ group), so we refrain from making general conclusions on the basis of this study.”

Probably not a substantial claim: may not pass S3 (load-bearing).

'Trust' in this study is based on momentary, immediate choices, rather than long-term, more organic exposure to descriptions of a system.

“'Trust' based on momentary, immediate choices, rather than long-term, more organic exposure to descriptions of a system.”

• AssumptionRests on an assumption: Participants' choices in the survey are a valid measure of their trust in the products.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants were asked to choose based on only a short description and no examples of the system's output.

“Participants were also asked to choose based on only a short description and no examples of the system's output.”

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

Using more familiar language to describe such systems (playing on anthropomorphizing metaphors and analogies) may make the systems more appealing to these groups.

“and potentially, using more familiar language to describe such systems (playing on anthropomorphizing metaphors and analogies) may make the systems more appealing to these groups.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Our studies support the findings of [31].

“Our studies thus support the findings of [31].”

• cites [31] [31] Markus Langer, Tim Hunsicker, Tina Feldkamp, Cornelius J König, and Nina Grgić-Hlača. 2022. 'Look! It'sa computer program! It's an algorithm! It's AI!': Does terminology affect human perceptions and evaluations of algorithmic decision-making systems?. In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems . 1-28. - sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Some products might be more susceptible to anthropomorphization of one type, and certain types of anthropomorphization might highlight or obfuscate specific qualities in specific system genres.

“Some products might be more susceptible to anthropomorphization of one type, and certain types of anthropomorphization might highlight or obfuscate specific qualities in specific system genres.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

We have tried to mitigate this factor by creating descriptions that are directly comparable to actual products found 'in the wild'.

“We have tried to mitigate this factor by creating descriptions that are directly comparable to actual products found 'in the wild'.”

Probably not a substantial claim: may not pass S3 (load-bearing).

We only swapped the categories between two different products.

“We only swapped the categories between two different products.”

Probably not a substantial claim: may not pass S3 (load-bearing).

To ensure that we reached a representative group we created pre-screening criteria such as country of residence, selfassessed socio-economic status, ethnicity, and geographic location.

“To ensure that we reached a representative group we created pre-screening criteria such as country of residence, selfassessed socio-economic status, ethnicity, and geographic location.”

Probably not a substantial claim: may not pass S3 (load-bearing).

We present our research team positionality according to the guidelines proposed by Liang et al..

“We present our research team positionality according to the guidelines proposed by Liang et al..”

• cites Liang et al.Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

In designing our online survey we adhered to the ethical guidelines in HCI methodology to ensure participant anonymity and data privacy.

“In designing our online survey we adhered to the ethical guidelines in HCI methodology [5] to ensure participant anonymity and data privacy.”

• cites [5] [5] Amy Bruckman. 2014. Research ethics and HCI. Ways of Knowing in HCI (2014), 449-468. -Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The results will not generalize to contexts where more information is given.

“but it could mean that the results will not generalize to contexts where more information is given.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

We did not collect any identifiable information and all the survey responses were stored temporarily on a secure server.

“We did not collect any identifiable information and all the survey responses were stored temporarily on a secure server.”

May combine two claims.

It appears to depend on product category and type of anthropomorphization, as well as the reader of the text.

“It appears to depend on product category and type of anthropomorphization, as well as the reader of the text.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

The expertise and lived experiences of our research team were an important part of the judgments and discussions in our analysis.

“The expertise and lived experiences of our research team were an important part of the judgments and discussions in our analysis.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

This should not have any effect on the primary variable (anthropomorphized versus deanthropomorphized), as these choices were always randomized.

“This should not have any effect on the primary variable (anthropomorphized versus deanthropomorphized), as these choices were always randomized.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In the thematic analysis, all survey responses were read several times while initial codes were generated.

“All survey responses were read several times while initial codes were generated.”

May combine two claims.

Our results show that anthropomorphized descriptions of systems do not automatically lead to favoritism or increased trust.

“Our results show that anthropomorphized descriptions of systems do not automatically lead to favoritism or increased trust.”

Classified as probably not a substantial claim. May combine two claims.

For each de-anthropomorphized product description, the five instances of anthropomorphic language were rewritten so they did not reflect the specific category of anthropomorphization, while the rest of the short pitch could include examples of other anthropomorphization categories, thereby isolating each anthropomorphization category as the independent variable.

“For each de-anthropomorpized description of the product, the five instances of anthropomorphic language were re-written so they did not reflect the specific category of anthropomorphization, but the rest of the short pitch could include examples of the other categories of anthropomorphization - thus, isolating each anthropomorphization category as the independent variable.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

The results provide enough insight for us to conclude that the matter is not straightforward, and that further investigation is needed.

“The results provide enough insight for us to conclude that the matter is not straightforward, and that further investigation is needed.”

Probably not a substantial claim: may not pass S3 (load-bearing).

To emphasize the anthropomorphic language as a variable, we have loaded a lot of 'anthropomorphisms' into very little text.

“To emphasize the anthropomorphic language as a variable, we have loaded a lot of 'anthropomorphisms' into very little text.”

Classified as probably not a substantial claim.

The authors were not strict about avoiding other categories of anthropomorphic language, especially the category of agency, in the anthropomorphic pitches.

“We were not strict about avoiding other categories of anthropomorphic language (especially the category of agency) in the pitches.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

To avoid confusion about the fictitious products, we added a statement at the end of the survey asserting that all products are 100% imagined, although some of them have been loosely based on existing products or services.

“To avoid confusion about the fictitious products, we added a statement at the end of the survey asserting that all products are 100% imagined, although some of them have been loosely based on existing products or services.”

Classified as probably not a substantial claim. May combine two claims.

Each anthropomorphic short pitch was written to fit its respective category and included 4-5 instances of that anthropomorphic category.

“Each of the anthropomorphic short pitches was written to fit in its respective category, and each of the short pitches included 4-5 'instances' of the anthropomorphic category.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In the survey, product pairs were always presented in the same order, which could induce order effects bias.

“In the survey, product pairs were always presented in the same order, which could could induce order effects bias.”

Classified as probably not a substantial claim.

The authors did not de-anthropomorphize language outside the target anthropomorphic language type in the corresponding de-anthropomorphized product description.

“However, we also did not de-anthropomorphize language outside the target anthropomoprhic language type in the corresponding de-anthropomorphized product description.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In the thematic analysis, all authors went over at least 100 responses and added tags (codes) and notes before a shared discussion about what appeared salient for respondents.

“All authors went over at least 100 responses and added tags (codes) and notes before a shared discussion about what appeared salient for respondents.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Because of the low N in the educational-level subgroups, a Fisher Exact test was performed on those numbers.

“Because of the low 𝑁 , a Fisher Exact test was performed on these numbers.”