Postdoc: Can AI Learn to Solve Analogies Like Children Can?

26 Jan 2024
Job Information

Organisation/Company

University of Amsterdam (UvA)

Research Field

Cultural studies

Researcher Profile

Recognised Researcher (R2)

Country

Netherlands

Application Deadline

1 Mar 2024 - 22:59 (UTC)

Type of Contract

Temporary

Job Status

Not Applicable

Hours Per Week

38.0

Is the job funded through the EU Research Framework Programme?

Not funded by an EU programme

Is the Job related to staff position within a Research Infrastructure?

No

Offer Description

Are you interested in doing research at the intersection of Developmental Science, Computational modeling and Cognitive AI? We are looking for a postdoc to join our research project entitled “Learning to solve analogies: Why do children excel where AI models fail?”, funded by NWO. This project is led by dr. Claire Stevenson (PI) and dr. Han van der Maas.

The public summary of the project is as follows. Body is to feet as tree is to …? Children become capable of solving such analogies around four-to-five years of age (answer: roots). Modern artificial intelligence (AI) systems, like young children, often provide less optimal answers such as “leaves”. Analogical reasoning lies at the heart of human intelligence. Why can't AI systems reason by analogy as well as children do? This project examines how children learn to solve analogies that challenge AI, generating insights to create more robust and trustworthy AI-systems, while providing key insights into human intelligence and how it develops.

What are you going to do?
The focus of this postdoc position is to examine how, when and why children shift from associative reasoning to successful analogy solving across various domains (e.g., verbal and visual). The aim is to identify the phases children progress through when learning to solve analogies and to identify mechanisms that enable children’s transition to successful analogy solving. Then, in collaboration with the project team, we can perhaps translate these mechanisms to AI models to help enable analogical reasoning in these systems.

The main research tasks will include organizing data collection with children (in person and online with Prowise Learn), creating and/or adapting computational models of analogy solving development, and complex statistical analyses including latent Markov modeling (latent transition analysis) and longitudinal analyses (e.g., latent change score models).

Requirements
Specific Requirements

We are looking for a candidate holding a PhD in cognitive science, developmental psychology or a related discipline. Given that the project lies at the intersection of psychology, cognitive science, and AI, a successful candidate has an interest and the capacity to work and interact in a multidisciplinary team. The ideal candidate speaks Dutch (for data collection with children), has excellent written and spoken command of English, a creative mind, strong statistical and computational modeling skills and embraces open science practices.

The successful candidate is familiar with:

• longitudinal data analysis using big data
• latent Markov modeling (a.k.a. latent transition analysis)
• computational modeling of cognition (or analogical reasoning in particular)
• coding in R and Python

Additional Information
Benefits

The position concerns temporary employment of 27 hours per week for two years. This can be expanded to 38 hours per week by taking on additional teaching tasks.

Your salary, depending on your relevant experience on commencement of the employment contract, ranges from €4322 to €5929 (scale 11) gross per month based on full-time employment and in keeping with the Collective Labour Agreement of Dutch Universities . We additionally offer an extensive package of secondary benefits, including 8% holiday allowance and a year-end bonus of 8.3%.

The UvA offers excellent possibilities for further professional development and education.

What else do we offer:

• a position in which initiative and input are highly valued
• an enthusiastic and warm team that is open to new colleagues
• an inspiring academic and international working environment in the heart of Amsterdam.

Selection process

Do you recognize yourself in the job profile? Then we look forward to receiving your application by March 1, 2024. You may apply online by using the link below.

Applications in one .pdf should include:

• a curriculum vitae and a list of 2-3 references that can be contacted in the process;
• a letter of motivation describing your background and interests relative to the position;
• one or two exemplary writing samples.

Additional comments

Do you have any questions or do you require additional information? Please contact:

• Claire Stevenson , Assistant Professor of Psychological Methods: [email protected]

Website for additional job details

https://www.academictransfer.com/337039/

Work Location(s)

Number of offers available

1

Company/Institute

Faculty of Social and Behavioural Sciences

Country

Netherlands

City

Amsterdam

Postal Code

1018WV

Street

Nieuwe Achtergracht 166

Geofield

Where to apply

Website

https://www.academictransfer.com/en/337039/postdoc-can-ai-learn-to-solve-analog…

Contact

City

Amsterdam

Website

http://www.uva.nl/

Street

Spui 21

Postal Code

1012 WX

STATUS: EXPIRED