03/17 — LNDN
︎ 02
Anthropomorphic
Loopholes.
Anthrop exp.00 — reddit 0217
understanding / time
understanding / time
︎ Anthropomorphic
Loopholes
2017
︎ Research
︎ Perception <>
AI, HRI
Publication:
︎
Loopholes
2017
︎ Research
︎ Perception <>
AI, HRI
Publication:
︎
Large
scale experiment carried out on Reddit, in which 940 participants took part in examining
the psychological rules that govern our anthropomorphic expectations of AI.
The experiment tested the relation between the use of humanlike adjectives when describing AI assistants (the density of the diagram being proportional to the frequency of use) both over time and in relationship to users’ understanding of the backend technology.
The resulting maps show that not only do humanlike perceptions tend to persist over time, but they are surprisingly more common amongst those claiming to understand the backend technology.
The experiment tested the relation between the use of humanlike adjectives when describing AI assistants (the density of the diagram being proportional to the frequency of use) both over time and in relationship to users’ understanding of the backend technology.
The resulting maps show that not only do humanlike perceptions tend to persist over time, but they are surprisingly more common amongst those claiming to understand the backend technology.
1-36 — TIME VARIATION (MONTHS)
1-10 UNDERSTANDING VARIATION
︎︎︎