Politicians are often accused of giving robotic answers when facing questions – but politicians in the UK may just have been shown how it's really done. The android Ai-Da, which is claimed to be the world's first ultra-realistic AI robot artist, was questioned by a committee in the British parliament on Tuesday. The politicians from the Communications and Digital Committee in the House of Lords asked the robot – named after the 19th century computer pioneer Ada Lovelace – about the relationship between artificial intelligence, robots, and the arts. "I do not have subjective experiences despite being able to talk about where I am and depend on computer programmes and algorithms who are very not alive. I can still create art," said the robot.

Two extremely rare "unreleased, one-of-a-kind, never-digitized" Nintendo NES games have appeared on eBay, according to a tweet from the Video Game History Foundation's Frank Cifaldi, seen by Kotaku. One of those, called Battlefields of Napoleon, was only ever released in Japan. The other is a cartridge from Rare, and appears to be the demo of one of the few games ever developed for the Nintendo Power Glove. According to the eBay listing for Battlefields of Napoleon, the game was "rescued from a dumpster after The Learning Company acquired Brøderbund in 1998 and subsequently discarded most of the historical assets." The items in the lot include a WATA certified prototype on a development board and two additional CHR ROMs.

Torch.AI today announced the company has joined Insurtech Vanguards, an initiative led by Guidewire, a leading Property & Casualty (P&C) cloud platform provider. With this, Torch.AI will bring data and artificial intelligence (AI) capabilities to insurance organizations across the insurance value chain. Guidewire's Insurtech Vanguards program is an initiative to help insurers learn about solutions from leading insurtechs and how to best work with them. As part of the program, Torch.AI joins one of the most reputable software providers in the insurance industry today to bring unique enterprise-level data infrastructure technology to a large and growing network of Guidewire customers and partners. "The large majority of insurance organizations today have not been able to optimize data usage across the business, still relying on error-prone, manual processes and complex point solutions," said Jason Eidam, Torch.AI's VP of Commercial Markets.

Laguna Hills, Calif. – DATE, 2022 – BrainChip Holdings Ltd (ASX: BRN, OTCQX: BRCHF, ADR: BCHPY), the world's first commercial producer of ultra-low power neuromorphic AI IP, has extended the breadth and depth of its neuromorphic IP with two new patents granted by the US Patents and Trademarks Office (USPTO), and the acquisition of previously licensed technology from Toulouse Tech Transfer (TTT). These latest additions of technical assets reinforce BrainChip's event-based processor differentiation for high performance, ultra-low power AI inference and on-chip learning. BrainChip also acquired full ownership of the IP rights related to JAST learning rule and algorithms from French technology transfer-based company TTT, including issued patent EP3324344 and pending patents US2019/0286944 and EP3324343. The invention related to the acquired IP rights include pattern detection algorithms that provide BrainChip with significant competitive advantages. The company held an exclusive license for the IP prior to their acquisition.

This paper presents a fast algorithm for estimating hidden states of Bayesian state space models. The algorithm is a variation of amortized simulation-based inference algorithms, where a large number of artificial datasets are generated at the first stage, and then a flexible model is trained to predict the variables of interest. In contrast to those proposed earlier, the procedure described in this paper makes it possible to train estimators for hidden states by concentrating only on certain characteristics of the marginal posterior distributions and introducing inductive bias. Illustrations using the examples of the stochastic volatility model, nonlinear dynamic stochastic general equilibrium model, and seasonal adjustment procedure with breaks in seasonality show that the algorithm has sufficient accuracy for practical use. Moreover, after pretraining, which takes several hours, finding the posterior distribution for any dataset takes from hundredths to tenths of a second.

Multi-hop QA requires reasoning over multiple supporting facts to answer the question. However, the existing QA models always rely on shortcuts, e.g., providing the true answer by only one fact, rather than multi-hop reasoning, which is referred as $\textit{disconnected reasoning}$ problem. To alleviate this issue, we propose a novel counterfactual multihop QA, a causal-effect approach that enables to reduce the disconnected reasoning. It builds upon explicitly modeling of causality: 1) the direct causal effects of disconnected reasoning and 2) the causal effect of true multi-hop reasoning from the total causal effect. With the causal graph, a counterfactual inference is proposed to disentangle the disconnected reasoning from the total causal effect, which provides us a new perspective and technology to learn a QA model that exploits the true multi-hop reasoning instead of shortcuts. Extensive experiments have conducted on the benchmark HotpotQA dataset, which demonstrate that the proposed method can achieve notable improvement on reducing disconnected reasoning. For example, our method achieves 5.8% higher points of its Supp$_s$ score on HotpotQA through true multihop reasoning. The code is available at supplementary material.

Current implementations of Gradient Boosting Machines are mostly designed for single-target regression tasks and commonly assume independence between responses when used in multivariate settings. As such, these models are not well suited if non-negligible dependencies exist between targets. To overcome this limitation, we present an extension of XGBoostLSS that models multiple targets and their dependencies in a probabilistic regression setting. Empirical results show that our approach outperforms existing GBMs with respect to runtime and compares well in terms of accuracy.

Recent studies show that Natural Language Processing (NLP) technologies propagate societal biases about demographic groups associated with attributes such as gender, race, and nationality. To create interventions and mitigate these biases and associated harms, it is vital to be able to detect and measure such biases. While existing works propose bias evaluation and mitigation methods for various tasks, there remains a need to cohesively understand the biases and the specific harms they measure, and how different measures compare with each other. To address this gap, this work presents a practical framework of harms and a series of questions that practitioners can answer to guide the development of bias measures. As a validation of our framework and documentation questions, we also present several case studies of how existing bias measures in NLP -- both intrinsic measures of bias in representations and extrinsic measures of bias of downstream applications -- can be aligned with different harms and how our proposed documentation questions facilitates more holistic understanding of what bias measures are measuring.

Multihop Question Answering is a complex Natural Language Processing task that requires multiple steps of reasoning to find the correct answer to a given question. Previous research has explored the use of models based on Graph Neural Networks for tackling this task. Various architectures have been proposed, including Relational Graph Convolutional Networks (RGCN). For these many node types and relations between them have been introduced, such as simple entity co-occurrences, modelling coreferences, or "reasoning paths" from questions to answers via intermediary entities. Nevertheless, a thoughtful analysis on which relations, node types, embeddings and architecture are the most beneficial for this task is still missing. In this paper we explore a number of RGCN-based Multihop QA models, graph relations, and node embeddings, and empirically explore the influence of each on Multihop QA performance on the WikiHop dataset.

We present an empirical study on methods for span finding, the selection of consecutive tokens in text for some downstream tasks. We focus on approaches that can be employed in training end-to-end information extraction systems, and find there is no definitive solution without considering task properties, and provide our observations to help with future design choices: 1) a tagging approach often yields higher precision while span enumeration and boundary prediction provide higher recall; 2) span type information can benefit a boundary prediction approach; 3) additional contextualization does not help span finding in most cases.