In this study, we examine the reliability of AI-based Voting Advice Applications (VAAs) and large language models (LLMs) in providing objective political information. Our analysis is based upon a comparison with party responses to 38 statements of the Wahl-O-Mat, a well-established German online tool that helps inform voters by comparing their views with political party positions. For the LLMs, we identify significant biases. They exhibit a strong alignment (over 75% on average) with left-wing parties and a substantially lower alignment with center-right (smaller 50%) and right-wing parties (around 30%). Furthermore, for the VAAs, intended to objectively inform voters, we found substantial deviations from the parties' stated positions in Wahl-O-Mat: While one VAA deviated in 25% of cases, another VAA showed deviations in more than 50% of cases. For the latter, we even observed that simple prompt injections led to severe hallucinations, including false claims such as non-existent connections between political parties and right-wing extremist ties.

Within the realm of privacy-preserving machine learning, empirical privacy defenses have been proposed as a solution to achieve satisfactory levels of training data privacy without a significant drop in model utility. Most existing defenses against membership inference attacks assume access to reference data, defined as an additional dataset coming from the same (or a similar) underlying distribution as training data. Despite the common use of reference data, previous works are notably reticent about defining and evaluating reference data privacy. As gains in model utility and/or training data privacy may come at the expense of reference data privacy, it is essential that all three aspects are duly considered. In this paper, we first examine the availability of reference data and its privacy treatment in previous works and demonstrate its necessity for fairly comparing defenses. Second, we propose a baseline defense that enables the utility-privacy tradeoff with respect to both training and reference data to be easily understood. Our method is formulated as an empirical risk minimization with a constraint on the generalization error, which, in practice, can be evaluated as a weighted empirical risk minimization (WERM) over the training and reference datasets. Although we conceived of WERM as a simple baseline, our experiments show that, surprisingly, it outperforms the most well-studied and current state-of-the-art empirical privacy defenses using reference data for nearly all relative privacy levels of reference and training data. Our investigation also reveals that these existing methods are unable to effectively trade off reference data privacy for model utility and/or training data privacy. Overall, our work highlights the need for a proper evaluation of the triad model utility / training data privacy / reference data privacy when comparing privacy defenses.

The 1970s were one of the most overtly political decades for science fiction filmmaking. Humor writer Tom Gerencer grew up watching movies such as Logan's Run, Silent Running, and Beneath the Planet of the Apes, all of which contain clear political messages. "We were watching industrialization do what it's continued to do now, getting worse and worse and worse, and we had a lot of voices back then saying, 'No, we have to stop this,' and rightly so," Gerencer says in Episode 543 of the Geek's Guide to the Galaxy podcast. Geek's Guide to the Galaxy host David Barr Kirtley was impressed by the continued relevance of many '70s science fiction movies, whether it's the idea of a deadly new virus in The Andromeda Strain or the threat of artificial intelligence in Colossus: The Forbin Project. "If you look at some of the issues they're dealing with--pandemics, AI, ecological collapse, youth culture, nuclear war--you would have to say that they did a pretty good job of honing in on some of the issues that were going to be important over the coming decades," he says.

Fox News contributor Jonathan Turley describes how ChatGPT falsely accused him and other professors of sexual harassment, made up a fake Washington Post story and concocted a fake quote as some news sites invest into AI written news stories. George Washington University law professor Jonathan Turley doubled down on warnings surrounding the dangers of artificial intelligence (AI) on Monday after he was falsely accused of sexual harassment by the online bot ChatGPT, which cited a fabricated article supporting the allegation. Turley, a Fox News contributor, has been outspoken about the pitfalls of artificial intelligence and has publicly expressed concerns with the disinformation dangers of the ChatGPT bot, the latest iteration of the AI chatbot. Last week, a UCLA professor and friend of Turley's notified him that his name appeared in a search while he was conducting research on ChatGPT. The bot was asked to cite "five examples" of "sexual harassment" by U.S. law professors with "quotes from relevant newspaper articles" to support it.

In response to the growing public attention given to ChatGPT, the Europol Innovation Lab organised a number of workshops with subject matter experts from across Europol to explore how criminals can abuse large language models (LLMs) such as ChatGPT, as well as how it may assist investigators in their daily work.

In June of 2020, given the latest bolus of articles re: "technology" applications in healthcare, I ruminated about the deployment (and risk) of artificial intelligence (AI) and machine learning (ML) technologies in the space. The utilization of technology to assist in care delivery, whether off-the-shelf solutions or custom designed AI products to empower decision making/care management, is necessary but should be approached with caution. As I'd noted, and continue to believe, AI and ML are constructs that require a bit of near-term expectation management in healthcare but do have application when deployed with solution-driven clarity. As suggested, while the efficacy and value of AI and ML will improve with time, they are not "the" answer that will remedy the myriad care and cost delivery questions surrounding healthcare in the United States. Owing to space constraints and the fact I am not an AI guru, this column is an overly simplistic noodling of recent AI foibles outside of healthcare that tell a larger story. As in 2020, to level set, I am not an AI programmer, don't code in Python, and have never built a ML algorithm.

Despite the buzz around artificial intelligence (AI), most industry insiders know that the use of machine learning (ML) in drug discovery is nothing new. For more than a decade, researchers have used computational techniques for many purposes, such as finding hits, modelling drug-protein interactions, and predicting reaction rates. As AI has taken off in other industries, countless start-ups have emerged promising to transform drug discovery and design with AI-based technologies. While a few "AI-native" candidates are in clinical trials, around 90% remain in discovery or preclinical development, so it will take years to see if the bets pay off. This begs the question: Is AI for drug discovery more hype than hope?

Robot with violin is followed by cloned businessmen. For decades, scientists and tech visionaries have envisioned a day when computers become so powerful that they become smarter than the human race. There is no shortage of science fiction stories and movies about robot uprisings. We are very far from that scary scenario, but at the same time artificial intelligence (AI) is no longer sci-fi. Many applications of AI abound today in business, and it is even being used in creative professions.

Robot with violin is followed by cloned businessmen. For decades, scientists and tech visionaries have envisioned a day when computers become so powerful that they become smarter than the human race. There is no shortage of science fiction stories and movies about robot uprisings. We are very far from that scary scenario, but at the same time artificial intelligence (AI) is no longer sci-fi. Many applications of AI abound today in business, and it is even being used in creative professions.

For decades, sci-fi movies have predicted a future in which humans lose control of intelligent machines and chaos ensues. Those apocalyptic portrayals of artificial intelligence may seem like a distant or unrealistic future. But the seeds of a reality in which we lose control of the machines we build are being sown today. "What is the biggest software-related issue to impact the economy in Puget Sound in 2019?" "Software in the cockpit of an airplane, software that the pilots couldn't turn off," Smith said. Smith was referring to the multi-billion dollar fallout from Boeing's faulty 737 Max software that resulted in two crashes killing 346 people. Boeing's manufacturing center is based in Renton, Wash.