Word embeddings have demonstrated strong performance on NLP tasks. However, lack of interpretability and the unsupervised nature of word embeddings have limited their use within computational social science and digital humanities. We propose the use of informative priors to create interpretable and domain-informed dimensions for probabilistic word embeddings. Experimental results show that sensible priors can capture latent semantic concepts better than or on-par with the current state of the art, while retaining the simplicity and generalizability of using priors.

Compressing word embeddings is important for deploying NLP models in memory-constrained settings. However, understanding what makes compressed embeddings perform well on downstream tasks is challenging---existing measures of compression quality often fail to distinguish between embeddings that perform well and those that do not. We thus propose the eigenspace overlap score as a new measure. We relate the eigenspace overlap score to downstream performance by developing generalization bounds for the compressed embeddings in terms of this score, in the context of linear and logistic regression. We then show that we can lower bound the eigenspace overlap score for a simple uniform quantization compression method, helping to explain the strong empirical performance of this method. Finally, we show that by using the eigenspace overlap score as a selection criterion between embeddings drawn from a representative set we compressed, we can efficiently identify the better performing embedding with up to $2\times$ lower selection error rates than the next best measure of compression quality, and avoid the cost of training a model for each task of interest.

W e present a systematic study of biases in natural language generation (NLG) by analyzing text generated from prompts that contain mentions of different demographic groups. In this work, we introduce the notion of the regard towards a demographic, use the varying levels of regard towards different demographics as a defining metric for bias in NLG, and analyze the extent to which sentiment scores are a relevant proxy metric for regard. To this end, we collect strategically-generated text from language models and manually annotate the text with both sentiment and regard scores. Additionally, we build an automatic regard classifier through transfer learning, so that we can analyze biases in unseen text. Together, these methods reveal the extent of the biased nature of language model generations. Our analysis provides a study of biases in NLG, bias metrics and correlated human judgments, and empirical evidence on the usefulness of our annotated dataset.

The government's peak advisory body on tech and science has turned its attention to the development of an artificial intelligence ethics framework and lifelong learning of STEM skills. The National Science and Technology Council met for the third time in Brisbane last week, after it was launched to replace the Commonwealth Science Council in February this year. The meeting was chaired by Industry Minister Karen Andrews, with education minister Dan Tehan also in attendance. Council members include Professor Genevieve Bell, Professor Barbara Howlett, Professor Debra Henly and Professor Brian Schmidt. They were briefed on the government's progress in developing a national artificial intelligence ethics framework, and the "strong engagement" from stakeholders during consultation.

A blue-collar worker is a working class person who performs manual labor. Blue-collar work may involve skilled or unskilled manufacturing, mining, sanitation, custodial work, textile manufacturing, power plant operations, farming, commercial fishing, landscaping, pest control, food processing, oil field work, waste disposal, recycling, electrical, plumbing, construction, mechanic, maintenance, warehousing, shipping, technical installation, and many other types of physical work. Blue-collar work often involves something being physically built or maintained. In contrast, the white-collar worker typically performs work in an office environment and may involve sitting at a computer or desk. A third type of work is a service worker (pink collar) whose labor is related to customer interaction, entertainment, sales or other service-oriented work.

PERTH, Australia – There is pervasive use of artificial intelligence and machine learning (AI/ML) across the health care industry in Australia, and excitement is building on the opportunities it offers to technologies and ultimately to patients, Ausbiotech CEO Lorraine Chiroiu told BioWorld. "AI/ML is transforming clinical practice in terms of clinical trials, diagnosis, treatment, decision-making, early detection and preventative health," she said. AI is being used for everything from smart medical records to the systems that help set appointments, to hospital records and diagnostic and pathology tests. It's being used in diagnostics for cancer patients to redirect the best treatment regimens based on a number of patient variables, and patient records can be aggregated so that algorithms can narrow down diagnoses. AI is changing the precision around surgeries like knee replacements by using robotic surgery to diagnose the exact angles, Brandon Capital Managing Director Chris Nave told BioWorld.

Dr Ian Pearson, an ex-cybernetics engineer, says our species risks a future "robot uprising". The futurologist said manufacturers who do not follow guidelines risk leaving robots to turn against us. He told Daily Star Online: "Military robots obviously would be able to kill people, but only a few. "To be an existential threat, there would need to be many millions of them that have become a threat without anyone noticing, and that seems unlikely. "Although again, it assumes a modicum of intelligence in regulation. "Robots plus online AI is a different threat.

GENEVA – Top players in global tech companies kicked off work Monday to draw up global ethical standards related to data and artificial intelligence, with Microsoft's president voicing a "sense of urgency. Some two dozen high-ranking representatives of the global and Swiss economies, as well as scientists and academics, met in Geneva for the first Swiss Global Digital Summit aimed at seeking agreement on ethical guidelines to steer technological development. The participants, including the heads of Credit Suisse, UBS and Adecco, and high-level representatives from Facebook, Google, Huawei and IBM, are due to meet again at the World Economic Forum in Davos next January. There, they will launch the Swiss Digital Initiative (SDI) and present a list of concrete projects, which could include things like the development of a "transparency label" or a "label. After the signing ceremony in Davos, "we are really going to go into practice, into implementation of concrete projects, and that is the proof of the pudding," former Swiss president Doris Leuthard, who will head SDI, told reporters.

According to a release issued earlier this month by the Los Alamos National Laboratory (LANL), researchers have developed a machine learning approach called transfer learning that lets them model novel materials by learning from data collected about millions of other compounds. The new approach can be applied to new molecules in milliseconds, enabling research into a far greater number of compounds over much longer timescales. The new technique, called ANI-1ccx potential, promises to advance the capabilities of researchers in many fields and improve the accuracy of machine learning-based potentials in future studies of metal alloys and detonation physics. "Our quantum mechanical calculations to create ANI-1ccx potential were conducted over two years with time split on the Comet supercomputer at the San Diego Supercomputer Center and the Badger supercomputer at LANL," said Olexandr Isayev, paper author and a pharmacy professor at the University of North Carolina at Chapel Hill. "We chose these two supercomputers to train our neural networks as there are few machines that can run these – due to the high memory and core requirements."

"As you ramble through Life, Brother, Whatever be your goal. Keep your eye upon the doughnut, And not upon the hole." Advice for those who drink coffee and eat "sinkers." Artificial Intelligence is not new. Another thing that is not new is people over-promising and under-delivering results around AI.