Welcome to our November 2021 monthly digest where you can catch up with any AIhub stories you may have missed, get the low-down on recent events, and much more. We have a bumper edition this month covering science communication, avoiding AI hype, events, awards, AI thanksgiving, and much, much more.
The AIhub coffee corner captures the musings of AI experts over a 30-minute conversation. We have not one, but two discussions to bring you this month.
The first, AIhub coffee corner: are deep learning’s returns diminishing?, was stimulated by an article that appeared recently in IEEE Spectrum. The article reported that deep-learning models are becoming more and more accurate, but that the computing power needed to achieve this accuracy is increasing at such a rate that, to further reduce the error, the cost and environmental impact is going to be unsustainably high. The article certainly generated an interesting discussion.
To mark Thanksgiving, our trustees talked about the things in the AI community for which they are grateful. Find out what they like about AI and the community in AIhub coffee corner: AI thanksgiving.
This month, we were delighted to announce the launch of scicomm.io, a joint science communication project from AIhub and Robohub. This new science communication platform aims to empower people to share stories about their robotics and AI work. As well as offering science communication training sessions, we have a series of byte-sized videos to enable you to get quickly up-to-speed with science communication for AI and robotics. Find out why science communication is important, how to talk to the media, and about some of the different ways in which you can communicate your work. We have also produced guides with tips for turning your research into a blog post and for avoiding hype when promoting your research.
Speaking of hype, we recently took to Twitter with a couple of informative threads on how to avoid generating AI hype, and how to spot AI hype. You can find the related articles, and our handy 10-point PDF guides on these topics below:
How to avoid hype when promoting your AI research | PDF
How to spot AI hype | PDF
This month saw the running of a number of conferences and events, many of which had associated best paper awards.
The Conference on Robot Learning (CoRL) took place earlier in the month in a hybrid format. You can catch-up with any of the talks you may have missed in this video archive that the organisers have put together.
We caught up with the CoRL best paper award winners, Tao Chen, Jie Xu and Pulkit Agrawal, who told us all about their winning work on a system for general in-hand object re-orientation, in this interview.
You can find out more about the other shortlisted papers and the winner of the best system paper award here.
The Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) took place both in Punta Cana and online. AIhub ambassador Nedjma Ousidhoum has put together this collection of livetweeted keynotes and papers, sorted by language.
You can find out who received the outstanding paper awards here.
Another big event to take place during November was the Asian Conference on Machine Learning (ACML 2021). A completely virtual gathering comprised talks, workshops and tutorials. You can find recording of the presentations here. The winners of the best paper awards were announced here.
The Stanford HAI fall conference comprised a discussion of four policy proposals that respond to the issues and opportunities created by artificial intelligence. The premise was that each policy proposal posed a challenge to the status quo. These proposals were presented to panels of experts who debated the merits and issues surrounding each policy. Find out more and watch the recordings here.
RoboCup president Peter Stone was interviewed on the Flash Forward podcast about robot sports, and whether robots will ever defeat human athletes. You can listen to the episode here – Peter’s segment starts at minute 19 and lasts about 5 minutes.
You can also read what Peter, and colleague Elliott Hauser, had to say about bringing robots into the real world in this interview.
Janelle Shane has been playing with Pixray Swirl. Pixray is an image generation system, and the Swirl version is a development that makes an animation from the generated image. You can see some of the cool results in Janelle’s blogpost.
Pablo Samuel Castro uses machine learning models for creative purposes. In this performance he explores using the losses and gradients obtained during the training of a simple function approximator as a mechanism for creating musical dissonance and visual distortion in a solo piano performance setting. You can find out more in his arXiv paper.
Our latest focus series sustainable cities and communities (as part of our wider series on the UN sustainable development goals) will launch in December. Keep your eyes peeled for exciting research stories from the community. If you are interested in contributing, there is still time to do so. Contact us here to find out how to get involved.
As always, don’t forget to check out content from all of our other focus collections:
Good health and well-being
Life below water
Affordable and clean energy
Life on land