Text-conditioned image generation models have recently achieved astonishing results in image quality and text alignment and are consequently employed in a fast-growing number of applications. Since they are highly data-driven, relying on billion-sized datasets randomly scraped from the internet, they also suffer, as we demonstrate, from degenerated and biased human behavior. In turn, they may even reinforce such biases. To help combat these undesired side effects, we present safe latent diffusion (SLD). Specifically, to measure the inappropriate degeneration due to unfiltered and imbalanced training sets, we establish a novel image generation test bed-inappropriate image prompts (I2P)-containing dedicated, real-world image-to-text prompts covering concepts such as nudity and violence. As our exhaustive empirical evaluation demonstrates, the introduced SLD removes and suppresses inappropriate image parts during the diffusion process, with no additional training required and no adverse effect on overall image quality or text alignment.

Artificial intelligence, machine learning, and algorithmic techniques in general, provide two crucial abilities with the potential to improve decision-making in the context of allocation of scarce societal resources. They have the ability to flexibly and accurately model treatment response at the individual level, potentially allowing us to better match available resources to individuals. In addition, they have the ability to reason simultaneously about the effects of matching sets of scarce resources to populations of individuals. In this work, we leverage these abilities to study algorithmic allocation of scarce societal resources in the context of homelessness. In communities throughout the United States, there is constant demand for an array of homeless services intended to address different levels of need. Allocations of housing services must match households to appropriate services that continuously fluctuate in availability, while inefficiencies in allocation could “waste” scarce resources as households will remain in-need and re-enter the homeless system, increasing the overall demand for homeless services. This complex allocation problem introduces novel technical and ethical challenges. Using administrative data from a regional homeless system, we formulate the problem of “optimal” allocation of resources given data on households with need for homeless services. The optimization problem aims to allocate available resources such that predicted probabilities of household re-entry are minimized. The key element of this work is its use of a counterfactual prediction approach that predicts household probabilities of re-entry into homeless services if assigned to each service. Through these counterfactual predictions, we find that this approach has the potential to improve the efficiency of the homeless system by reducing re-entry, and, therefore, system-wide demand. However, efficiency comes with trade-offs - a significant fraction of households are assigned to services that increase probability of re-entry. To address this issue as well as the inherent fairness considerations present in any context where there are insufficient resources to meet demand, we discuss the efficiency, equity, and fairness issues that arise in our work and consider potential implications for homeless policies.

Organizations are increasingly adopting machine learning (ML) for personnel assessment. However, concerns exist about fairness in designing and implementing ML assessments. Supervised ML models are trained to model patterns in data, meaning ML models tend to yield predictions that reflect subgroup differences in applicant attributes in the training data, regardless of the underlying cause of subgroup differences. In this study, we systematically under- and oversampled minority (Black and Hispanic) applicants to manipulate adverse impact ratios in training data and investigated how training data adverse impact ratios affect ML model adverse impact and accuracy. We used self-reports and interview transcripts from job applicants (N = 2,501) to train 9,702 ML models to predict screening decisions. We found that training data adverse impact related linearly to ML model adverse impact. However, removing adverse impact from training data only slightly reduced ML model adverse impact and tended to negatively affect ML model accuracy. We observed consistent effects across self-reports and interview transcripts, whether oversampling real (i.e., bootstrapping) or synthetic observations. As our study relied on limited predictor sets from one organization, the observed effects on adverse impact may be attenuated among more accurate ML models.

On April 24th, 2023, reports circulated that Blizzard Entertainment was being sued by former Chinese publishing partner NetEase after servers shutdown in January when the two failed to reach a continuation agreement. However, a day later, it turns out that NetEase was in fact not suing the company -- instead, as reported by PC Gamer, the suit is being brought by a single individual who is known to be a serial litigant with no history with NetEase. It appears the court documents listened NetEase erroneously; the company does not have anything to do with the lawsuit. Originally, MMO-focused gaming website Wowhead noticed the suit. Since this story was originally published, those court documents have been re-published to reflect that the suits are coming from a Yang Jun; all mentions of NetEase have been removed.

"The View" co-host Whoopi Goldberg criticized re-editing books in an effort to avoid offending modern audiences and argued "that's how kids learn." Among the most tragic events in human cultural history was the destruction of works from the great library of Alexandria. Blamed on Julius Caesar as well as later Christian and Muslim zealots, the net loss of knowledge from this font of ancient wisdom roughly coincided with what we call the Dark Ages, and we may be repeating history. From its beginnings one of the great promises of computer technology was the possibility of maintaining a library of all human writing that could not burn, that would neither fade nor wither. The irony, that has not been considered closely enough, is how easily this same technology can revise or fabricate literary and historical classics, which is tantamount to destroying them.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. When the U.S. Supreme Court decides in the coming months whether to weaken a powerful shield protecting internet companies, the ruling also could have implications for rapidly developing technologies like artificial intelligence chatbot ChatGPT. The justices are due to rule by the end of June whether Alphabet Inc's YouTube can be sued over its video recommendations to users. That case tests whether a U.S. law that protects technology platforms from legal responsibility for content posted online by their users also applies when companies use algorithms to target users with recommendations.

Andy Warhol probably never said that thing about everyone in the future getting their 15 minutes of fame. It might have been Swedish art collector Pontus Hultén. Warhol is the household name, though, so he gets the credit. But he did say this: "Being good in business is the most fascinating kind of art." Warhol won his first advertising award in 1952.

The pop singer invites creators to generate new music using software trained on her voice.

The UK's Competition and Markets Authority (CMA) has introduced a new bill that would give it the power slap the biggest tech companies with a fine worth billions if they don't comply with its rules. It's a multi-faceted bill that's aimed at protecting consumers and encouraging competition, and it will allow the CMA to directly enforce the law instead of having to go through the court. If the bill passes, the agency's Digital Markets Unit (DMU) will be able to enforce a set of rules on how companies it deems to have "strategic market status" in key digital services have to operate. The CMA didn't name any specific company in its announcement, but the DMU will most likely identify Google, Apple and Amazon as organizations with strategic market status. The DMU could require them to be more transparent on how their app store review systems work or to open up their data to rivals -- in Google's case, it could be a rival search engine. If these companies fail to abide by the new rules, the DMU could fine them up to 10 percent of their global turnover.

Fox News correspondent David Spunt has the latest as the Supreme Court weighs whether tech companies should be legally liable for harmful content on their platforms on'Special Report.' An impending Supreme Court ruling focusing on whether legal protections given to Big Tech extend to their algorithms and recommendation features could have significant implications for future cases surrounding artificial intelligence, according to experts. In late February, the Supreme Court heard oral arguments examining the extent of legal immunity given to tech companies that allow third-party users to publish content on their platforms. One of two cases, Gonzalez v. Google, focuses on recommendations and algorithms used by sites like YouTube, allowing accounts to arrange and promote content to users. Section 230, which allows online platforms significant leeway regarding responsibility for users' speech, has been challenged multiple times in the Supreme Court.