Neural conditional language generation models achieve the state-of-the-art in Neural Machine Translation (NMT) but are highly dependent on the quality of parallel training dataset. When trained on low-quality datasets, these models are prone to various error types, including hallucinations, i.e. outputs that are fluent, but unrelated to the source sentences. These errors are particularly dangerous, because on the surface the translation can be perceived as a correct output, especially if the reader does not understand the source language. We present a case study focusing on model understanding and regularisation to reduce hallucinations in NMT. We first use feature attribution methods to study the behaviour of an NMT model that produces hallucinations. We then leverage these methods to propose a novel loss function that substantially helps reduce hallucinations and does not require retraining the model from scratch.

Twitter's former CEO Jack Dorsey has given an interview to Breaking Points in his first media appearance since stepping down from the social media giant in 2021. After touching on Elon Musk's Twitter takeover and surrounding controversies, Dorsey also addressed censorship battles with governments, his view of Twitter's role in the free-speech debate, as well as the future of artificial intelligence technology. Dorsey said he was happy when Musk made moves to become more involved with Twitter. "At the very start, I was hoping for years that [Musk] would, and I asked him many times to join our board at least. But when he decided to make a bid for the company, or join the board and then make a bid for the company … it felt great," said Dorsey.

Despite making concessions to any state-level agency that will listen, Microsoft's would-be purchase of mega-game publisher Activision Blizzard continues to run into regulatory hurdles. The latest obstacle for the deal is the United States Federal Trade Commission, which has filed an injunction and restraining order to block the nearly $70 billion purchase from going through. The following proceedings will determine whether the US government allows the largest deal in video game history to proceed. The various steps in the regulatory process are complicated, but competently explained by the Associated Press. Essentially, the FTC had already signaled its opposition to the Microsoft-Activision deal on anti-trust grounds late last year but had not completed the process of formally blocking the merger.

On May 30, the Center for AI Safety released a public warning of the risk artificial intelligence poses to humanity. The one-sentence statement signed by more than 350 scientists, business executives and public figures asserts: "Mitigating the risk of extinction from A.I. should be a global priority alongside other societal scale risks such as pandemics and nuclear war." It is hard not to sense the brutal double irony in this declaration. First, some of the signatories – including the CEOs of Google DeepMind and OpenAI – warning about the end of civilisation represent companies that are responsible for creating this technology in the first place. Second, it is exactly these same companies that have the power to ensure that AI actually benefits humanity, or at the very least does not do harm.

The news: An algorithm funded by the World Bank to determine which families should get financial assistance in Jordan likely excludes people who should qualify, an investigation from Humans Rights Watch has found. Why it matters: The organization identified several fundamental problems with the algorithmic system that resulted in bias and inaccuracies. It ranks families applying for aid from least poor to poorest using a secret calculus that assigns weights to 57 socioeconomic indicators. Applicants say that the calculus is not reflective of reality, and oversimplifies people's economic situation. The bigger picture: AI ethics researchers are calling for more scrutiny around the increasing use of algorithms in welfare systems.

Last month, Sam Altman, the CEO of OpenAI and face of the artificial intelligence boom, sat in front of members of Congress urging them to regulate artificial intelligence (AI). As lawmakers on the Senate judiciary subcommittee asked the 38-year-old tech mogul about the nature of his business, Altman argued that the AI industry could be dangerous and that the government needs to step in. "I think if this technology goes wrong, it can go quite wrong," Altman said. "We want to be vocal about that." How governments should regulate artificial intelligence is a topic of increasing urgency in countries around the world, as advancements reach the general public and threaten to upend entire industries.

"The questions asked don't reflect the reality we exist in," says Abdelhamad, a father of two who makes 250 dinars ($353) a month and struggles to make ends meet, as quoted in the report. Takaful also reinforces existing gender-based discrimination by relying on sexist legal codes. The cash assistance is provided to Jordanian citizens only, and one indicator the algorithm takes into account is the size of a household. Although Jordanian men who marry a noncitizen can pass on citizenship to their spouse, Jordanian women who do so cannot. For such women, this results in a lower reportable household size, making them less likely to receive assistance.

The detection of political fake statements is crucial for maintaining information integrity and preventing the spread of misinformation in society. Historically, state-of-the-art machine learning models employed various methods for detecting deceptive statements. These methods include the use of metadata (W. Wang et al., 2018), n-grams analysis (Singh et al., 2021), and linguistic (Wu et al., 2022) and stylometric (Islam et al., 2020) features. Recent advancements in large language models, such as GPT-3 (Brown et al., 2020) have achieved state-of-the-art performance on a wide range of tasks. In this study, we conducted experiments with GPT-3 on the LIAR dataset (W. Wang et al., 2018) and achieved higher accuracy than state-of-the-art models without using any additional meta or linguistic features. Additionally, we experimented with zero-shot learning using a carefully designed prompt and achieved near state-of-the-art performance. An advantage of this approach is that the model provided evidence for its decision, which adds transparency to the model's decision-making and offers a chance for users to verify the validity of the evidence provided.

To prevent the costly and inefficient use of resources on low-quality annotations, we want a method for creating a pool of dependable annotators who can effectively complete difficult tasks, such as evaluating automatic summarization. Thus, we investigate the recruitment of high-quality Amazon Mechanical Turk workers via a two-step pipeline. We show that we can successfully filter out subpar workers before they carry out the evaluations and obtain high-agreement annotations with similar constraints on resources. Although our workers demonstrate a strong consensus among themselves and CloudResearch workers, their alignment with expert judgments on a subset of the data is not as expected and needs further training in correctness. This paper still serves as a best practice for the recruitment of qualified annotators in other challenging annotation tasks.

This paper presents our participation in the FinNLP-2023 shared task on multi-lingual environmental, social, and corporate governance issue identification (ML-ESG). The task's objective is to classify news articles based on the 35 ESG key issues defined by the MSCI ESG rating guidelines. Our approach focuses on the English and French subtasks, employing the CerebrasGPT, OPT, and Pythia models, along with the zero-shot and GPT3Mix Augmentation techniques. We utilize various encoder models, such as RoBERTa, DeBERTa, and FinBERT, subjecting them to knowledge distillation and additional training. Our approach yielded exceptional results, securing the first position in the English text subtask with F1-score 0.69 and the second position in the French text subtask with F1-score 0.78. These outcomes underscore the effectiveness of our methodology in identifying ESG issues in news articles across different languages. Our findings contribute to the exploration of ESG topics and highlight the potential of leveraging advanced language models for ESG issue identification.