Dataset Information

Direct Human-AI Comparison in the Animal-AI Environment.

ABSTRACT: Artificial Intelligence is making rapid and remarkable progress in the development of more sophisticated and powerful systems. However, the acknowledgement of several problems with modern machine learning approaches has prompted a shift in AI benchmarking away from task-oriented testing (such as Chess and Go) towards ability-oriented testing, in which AI systems are tested on their capacity to solve certain kinds of novel problems. The Animal-AI Environment is one such benchmark which aims to apply the ability-oriented testing used in comparative psychology to AI systems. Here, we present the first direct human-AI comparison in the Animal-AI Environment, using children aged 6-10 (n = 52). We found that children of all ages were significantly better than a sample of 30 AIs across most of the tests we examined, as well as performing significantly better than the two top-scoring AIs, "ironbar" and "Trrrrr," from the Animal-AI Olympics Competition 2019. While children and AIs performed similarly on basic navigational tasks, AIs performed significantly worse in more complex cognitive tests, including detour tasks, spatial elimination tasks, and object permanence tasks, indicating that AIs lack several cognitive abilities that children aged 6-10 possess. Both children and AIs performed poorly on tool-use tasks, suggesting that these tests are challenging for both biological and non-biological machines.

SUBMITTER: Voudouris K

PROVIDER: S-EPMC9172850 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Direct Human-AI Comparison in the Animal-AI Environment.

Voudouris Konstantinos K Crosby Matthew M Beyret Benjamin B Hernández-Orallo José J Shanahan Murray M Halina Marta M Cheke Lucy G LG

Frontiers in psychology 20220524

Artificial Intelligence is making rapid and remarkable progress in the development of more sophisticated and powerful systems. However, the acknowledgement of several problems with modern machine learning approaches has prompted a shift in AI benchmarking away from task-oriented testing (such as Chess and Go) towards <i>ability</i>-oriented testing, in which AI systems are tested on their capacity to solve certain <i>kinds</i> of novel problems. The Animal-AI Environment is one such benchmark wh ...[more]

PMID: 35686061

Similar Datasets

Project description:The medical authority in China, especially in Wuhan city, reported on December 2019 a large number of highly fatal, rapidly spreading viral pneumonia caused by an unknown coronavirus. The common history of all the patients was their visiting a Wuhan's whole food store, where live animals and seafood are sold. Irrespective of the efforts of the Chinese authorities, the virus spread rapidly all over the world by travelers, provoking widespread attention by the media and panic. Many previous coronavirus epidemics had been recorded, such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS), and the recently newly discovered epidemic is named coronavirus disease of 2019 (COVID-19). This disease is caused by SARS Coronavirus-2 (SARS-CoV-2), and this virus is antigenically related to the SARS virus (SARS-CoV), which had been detected in 2002, depending on clinical, serological, and molecular findings. There is rapid competition among the researchers to discover the source of the virus, understand the mechanism of the disease development, establish treatment strategies, and determine the factors affecting the incidence of infection and severity of the disease, and focus on the production of a vaccine. Coronaviruses are a group of single-stranded, positive-sense RNA genome viruses; its genome length varies from 26 to 32 kb. Coronavirus causes mild to severe respiratory disorders. In December 2019, several cases of pneumonia of unknown causes were found in Wuhan city, which is located in the Hubei province in China. Chinese health authorities investigated the problem and found that a new virus caused such infection and, using next-generation sequencing, found the 2019 novel coronavirus (2019-nCoV). It has been transferred from humans to humans and animals to humans (zoonotic). Coronaviruses cause multiple respiratory problems, varying from common cold to severe infections such as SARS. General symptoms of infection include fatigue, cough, and breathing problems such as shortness of breath, as described by World Health Organization. Serious cases may result in pneumonia, renal failure, and even death. We address current information about the new SARS Coronavirus-2 as well as the COVID-19 disease caused by it in this review.

Project description:BackgroundData-driven research is a very important component of One Health. As the core part of the global One Health index (GOHI), the global One Health Intrinsic Drivers index (IDI) is a framework for evaluating the baseline conditions of human-animal-environment health. This study aims to assess the global performance in terms of GOH-IDI, compare it across different World Bank regions, and analyze the relationships between GOH-IDI and national economic levels.MethodsThe raw data among 146 countries were collected from authoritative databases and official reports in November 2021. Descriptive statistical analysis, data visualization and manipulation, Shapiro normality test and ridge maps were used to evaluate and identify the spatial and classificatory distribution of GOH-IDI. This paper uses the World Bank regional classification and the World Bank income groups to analyse the relationship between GOH-IDI and regional economic levels, and completes the case studies of representative countries.ResultsThe performance of One Health Intrinsic Driver in 146 countries was evaluated. The mean (standard deviation, SD) score of GOH-IDI is 54.05 (4.95). The values (mean SD) of different regions are North America (60.44, 2.36), Europe and Central Asia (57.73, 3.29), Middle East and North Africa (57.02, 2.56), East Asia and Pacific (53.87, 5.22), Latin America and the Caribbean (53.75, 2.20), South Asia (52.45, 2.61) and sub-Saharan Africa (48.27, 2.48). Gross national income per capita was moderately correlated with GOH-IDI (R2 = 0.651, Deviance explained = 66.6%, P < 0.005). Low income countries have the best performance in some secondary indicators, including Non-communicable Diseases and Mental Health and Health risks. Five indicators are not statistically different at each economic level, including Animal Epidemic Disease, Animal Biodiversity, Air Quality and Climate Change, Land Resources and Environmental Biodiversity.ConclusionsThe GOH-IDI is a crucial tool to evaluate the situation of One Health. There are inter-regional differences in GOH-IDI significantly at the worldwide level. The best performing region for GOH-IDI was North America and the worst was sub-Saharan Africa. There is a positive correlation between the GOH-IDI and country economic status, with high-income countries performing well in most indicators. GOH-IDI facilitates researchers' understanding of the multidimensional situation in each country and invests more attention in scientific questions that need to be addressed urgently.

Dataset Information

Direct Human-AI Comparison in the Animal-AI Environment.

Publications

Direct Human-AI Comparison in the Animal-AI Environment.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets