When people think of artificial intelligence, the images that often come to mind are of the sinister robots that populate the worlds of "The Terminator," "i, Robot," "Westworld," and "Blade Runner." For many years, fiction has told us that AI is often used for evil rather than for good. But what we may not usually associate with AI is art and poetry -- yet that's exactly what Ai-Da, a highly realistic robot invented by Aidan Meller in Oxford, central England, spends her time creating. Ai-Da is the world's first ultra-realistic humanoid robot artist, and on Friday she gave a public performance of poetry that she wrote using her algorithms in celebration of the great Italian poet Dante. The recital took place at the University of Oxford's renowned Ashmolean Museum as part of an exhibition marking the 700th anniversary of Dante's death.

As with many of us, after three years of staying home, I realized a few months ago that I'd had it with the pandemic. Having traveled up to 70% for years before Covid-19, my initial reaction to being in the same time zone and same building and bed was pure gratitude, even bliss. I wanted to go out and experience the world again. During our last Omicron-initiated staycation over the Christmas holidays, my 14-year-old son stated in his very polished, diplomatic and convincing style that he was bored. As soon as we learned Omicron was manageable, and there would be a break from lockdowns and fewer travel restrictions, we decided to get on with it and book some memorable holidays. So came the trips to Costa Rica and the Dominican Republic.

Previous security research efforts orbiting around graphs have been exclusively focusing on either (de-)anonymizing the graphs or understanding the security and privacy issues of graph neural networks. Little attention has been paid to understand the privacy risks of integrating the output from graph embedding models (e.g., node embeddings) with complex downstream machine learning pipelines. In this paper, we fill this gap and propose a novel model-agnostic graph recovery attack that exploits the implicit graph structural information preserved in the embeddings of graph nodes. We show that an adversary can recover edges with decent accuracy by only gaining access to the node embedding matrix of the original graph without interactions with the node embedding models. We demonstrate the effectiveness and applicability of our graph recovery attack through extensive experiments.

In this paper, we propose FFCI, a framework for fine-grained summarization evaluation that comprises four elements: faithfulness (degree of factual consistency with the source), focus (precision of summary content relative to the reference), coverage (recall of summary content relative to the reference), and inter-sentential coherence (document fluency between adjacent sentences). We construct a novel dataset for focus, coverage, and inter-sentential coherence, and develop automatic methods for evaluating each of the four dimensions of FFCI based on cross-comparison of evaluation metrics and model-based evaluation methods, including question answering (QA) approaches, semantic textual similarity (STS), next-sentence prediction (NSP), and scores derived from 19 pre-trained language models. We then apply the developed metrics in evaluating a broad range of summarization models across two datasets, with some surprising findings.

Instead of analyzing days, weeks, or months of recorded footage for animal behavior studies, researchers can now turn the task over to an open-source algorithm capable of picking up even subtle actions. Not only could this save researchers time, it could also improve the lives of animals in zoos, the wild, and the lab. Animal behavior: Animals can't talk to humans (yet), but they can communicate through their behavior -- an animal that's eating less than normal may be experiencing depression or illness, for example. Much of what we can learn from animal behavior requires that we study them for an extended period of time. Being able to analyze animal behavior is hugely important for both people and animals.

On the internet and in the real world women are drastically under-represented than men in all types of things whether it be profession or the entertainment industry. If a need of information arises we automatically go to Wikipedia but Wikipedia is not as diverse as you might think because women are underrepresented over there as well. The website does not give the credit and recognition that is deserved by women because only 20% of all biographies on Wikipedia belong to women. That percentage deteriorates further when we talk about industries that have male domination like science and underrepresented historical communities. Addressing the fact that gender bias exists especially on the internet, Angela Fan who is an AI researcher for Meta, worked on this topic for her PhD.

Welcome to our April 2022 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. This month, we hear from our latest new voice in AI, talk about AI images, investigate data justice, and watch an AI system play bridge. In our latest episode of New voices in AI, we caught up with Maria De-Arteaga who told us about her work and journey into algorithmic fairness and human algorithm collaboration. You can find all episodes in the series here. In this article, Thom Badings and Nils Jansen write about their work on controllers for autonomous systems that won them, and co-authors Alessandro Abate, David Parker, Hasan Poonawala, and Marielle Stoelinga, a distinguished paper award at AAAI 2022.

However, current autoregressive approaches suffer from high latency. In this paper, we focus on non-autoregressive translation (NAT) for this problem for its efficiency advantage. We identify that current constrained NAT models, which are based on iterative editing, do not handle low-frequency constraints well. To this end, we propose a plug-in algorithm for this line of work, i.e., Aligned Constrained Training (ACT), which alleviates this problem by familiarizing the model with the source-side context of the constraints. Experiments on the general and domain datasets show that our model improves over the backbone constrained NAT model in constraint preservation and translation quality, especially for rare constraints.

Aidan Connolly is the President of AgriTech Capital, a food/farm futurologist, and author of "2-1-4-3, Plan your Explosive Business Growth," Described as the world's least digitized industry by McKinsey analysts (joint last position with hunting), the food producers of the world could only agree that agriculture has struggled to avail of the breakthroughs in technology that have transformed other industries. Uber has disrupted transportation, Netflix the movies, Airbnb the hotel business, online money movers who hold no cash now dominate banking and we purchase apps from companies who don't make them. Yet, farming seems to have changed little in the 10,000 years since the first animals were domesticated, and many believe that it will change little in the coming decades. However, I contend that this view is myopic and fails to recognize the degree of disruption already happening in farming. Sean Moffitt, managing director of Futureproofing, listed the 30 new technologies that both are currently seeing the greatest dollar investments and that industries will require to futureproof themselves for the next decade.

By Joyjit Chatterjee and Nina Dethlefs, University of Hull Cottingham Artificial Intelligence (AI) is much more than just a buzzword nowadays. It powers facial recognition in smartphones and computers, translation between foreign languages, systems which filter spam emails and identify toxic content on social media, and can even detect cancerous tumours. These examples, along with countless other existing and emerging applications of AI, help make people's daily lives easier, especially in the developed world. As of October 2021, 44 countries were reported to have their own national AI strategic plans, showing their willingness to forge ahead in the global AI race. These include emerging economies like China and India, which are leading the way in building national AI plans within the developing world.