We start 2024 with a packed monthly digest, where you can catch up with any AIhub stories you may have missed, peruse the latest news, recap recent events, and more. This month, we continue our coverage of NeurIPS, meet the first interviewee in our AAAI Doctoral Consortium series, and find out how to build AI openly.
The AAAI/SIGAI Doctoral Consortium provides an opportunity for a group of PhD students to discuss and explore their research interests and career objectives in an interdisciplinary workshop together with a panel of established researchers. Over the course of the next few months, we’ll be meeting the participants and finding out more about their work, PhD life, and their future research plans. In the first interview of the series, Changhoon Kim told us about his research on enhancing the reliability of image generative AI.
Bo Li and colleagues won an outstanding datasets and benchmark track award at NeurIPS 2023 for their work DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models. In this interview, Bo tells us about the research, the team’s methodology, and key findings.
As part of the programme at the Conference on Neural Information Processing Systems (NeurIPS 2023), a series of invited talks covered a range of fascinating topics. In her presentation, Linda Smith spoke about work monitoring young babies and how the findings could inform ML research. Lora Aroyo tackled the subject of responsible AI, specifically looking at the data annotation process and what this means for models that use those data.
To model climate on a local-scale, researchers commonly use statistical downscaling (SD) to map the coarse resolution of climate models to the required local-scale. The use of deep-learning to facilitate SD often leads to violation of physical properties. In this blog post, Jose González-Abad writes about work that investigates the scope of this problem and lays the foundation for a framework that guarantees physical relationships between groups of downscaled climate variables.
In this blogpost, Yi Chen, Ramya Korlakai Vinayak and Babak Hassibi write about work presented at the Eleventh AAAI Conference on Human Computation and Crowdsourcing (HCOMP 2023) in which they introduce Active Crowdclustering, an algorithm that finds clusters in a dataset with unlabeled items by querying pairs of items for similarity.
Diogo Carvalho, Francisco Melo and Pedro Santos won an ECAI 2023 outstanding paper award for their paper Theoretical Remarks on Feudal Hierarchies and Reinforcement Learning. In this blogpost, Diogo explains hierarchical reinforcement learning, and summarises how the team showed that Q-learning solves the hierarchical decision making process.
In their paper Model Checking for Closed-Loop Robot Reactive Planning, Christopher Chandler, Bernd Porr, Alice Miller and Giulia Lafratta show how model checking can be used to create multi-step plans for a differential drive wheeled robot so that it can avoid immediate danger. In this interview, Christopher tells us about model checking and how it is used in the context of autonomous robotic systems.
The Feminist AI lecture series (organised by the University of Arts Linz), which ran from September 2023 to January 2024, presented inspiring lectures on gender and AI. The recordings from the five events are available here.
The Auschwitz Pledge Foundation has recently launched the Erase Indifference Challenge 2024, a competition that aims to support innovative projects leveraging technology to combat indifference to discrimination. They are offering grants of up to €30,000 for the three winning projects. The deadline to enter is 11 February, and you can find out more here.
We and AI and The Scottish AI Alliance have joined forces on an introductory AI course. The five-week course is perfect for anyone looking to understand how AI is being used in our world and the rapid changes it is making. It is designed to be accessible to all, regardless of prior knowledge or experience with AI. There is still time to sign up if you are interested.
In a recent TED talk, Percy Liang spoke about the necessity to build AI openly. He presented his vision for a transparent, participatory future, one that credits contributors and gives everyone a voice.