Benchmarks have been essential for driving progress in machine learning. A better understanding of LLM capabilities on real world tasks is vital for safe development. Designing adequate LLM benchmarks is challenging: Data from real-world tasks is hard to collect, public availability of static evaluation data results in test data contamination and benchmark overfitting, and periodically generating new evaluation data is tedious and may result in temporally inconsistent results. We introduce HelloFresh, based on continuous streams of real-world data generated by intrinsically motivated human labelers. It covers recent events from X (formerly Twitter) community notes and edits of Wikipedia pages, mitigating the risk of test data contamination and benchmark overfitting. Any X user can propose an X note to add additional context to a misleading post (formerly tweet); if the community classifies it as helpful, it is shown with the post. Similarly, Wikipedia relies on community-based consensus, allowing users to edit articles or revert edits made by other users. Verifying whether an X note is helpful or whether a Wikipedia edit should be accepted are hard tasks that require grounding by q
Get up to 55% off and free meal boxes using a HelloFresh coupon code in July 2026。 Discover our best codes and discounts to let you save time and money
The aim of this project is to review the products and recipes contained within popular subscription meal kits to determine if they are suitable for wider use among people who are food insecure. Across the 6-week period, weekly meal kits from both HelloFresh and Markey Spoon were purchased, resulting in thirty-six individual meals that were prepared and assessed. Meals were assessed based on the content included in the meal kit compared with the recipe card and the nutrition panel, the costs of the individual foods if purchased at one of two major supermarkets and the ease of preparation. Australia. Households were comprised of two, 2-person households who were provided with 2 meals each week, and two, single-person households who were provided with one meal each week. The findings of this research suggest that while the meal kits are convenient and, in general, the recipes are easy to follow, and the meals would be made again, the high levels of salt and fat may preclude these kits from regular inclusion in a healthy diet. The meal kits were also found to be more costly than the same ingredients if purchased from a major supermarket. However, the convenience of having most of the foods needed to prepare a full meal with little to no wastage may counterbalance this cost. Meal kits may be a useful component of a healthy diet, that can increase meals prepared and consumed in the home, and thanks to the clear instructions and pre-portioned ingredients, may reduce stress related to food preparation.
The rise of digital speech-based assistants (e.g., Amazon's Alexa, Apple's Siri, or Hellofresh's chatbot) gained increased popularity and has also found its way into healthcare. A current example is WHO's WhatsApp chatbot that informs users about COVID-19. Such assistants can educate patients, independent of location and time, which is an incredible benefit for patients. However, in addition to existing technical challenges (including the development and optimization of speech recognition algorithms), there are also challenges in human-chatbot-interactions. This study investigates the role of a human-like design (including human name, greeting, and human avatar) of a travel vaccination advice chatbot on the perception of its users. Specifically, we aim to understand whether and how anthropomorphism (the perception of humanness and social presence in objects, animals, and machines) affects perceived trustworthiness and ultimately service satisfaction. In an online experiment with 78 participants, two chatbot designs (with human-like design elements vs. without these elements) were compared. The results show that perceived social presence significantly increases perceived trustworthiness and service satisfaction. Thus, we recommend that practitioners implement a human-like design travel vaccination counseling and similar counseling processes via chatbots. Der Einsatz von Sprachassistenten (Alltagsbeispiele sind Alexa von Amazon, Siri von Apple oder der Chatbot von Hellofresh) hat auch im Gesundheitswesen Einzug gehalten. Ein aktuelles Beispiel ist der WhatsApp Chatbot der WHO, welcher Nutzer:innen über COVID-19 aufklärt. Sprachassistenten haben die Fähigkeit, Patienten und Patientinnen orts- und zeitunabhängig aufzuklären, wodurch Mitarbeiter:innen entlastet werden. Jedoch gibt es neben den technischen (u. a. Entwicklung und Optimierung der Algorithmen für die Spracherkennung) auch Herausforderungen in der Mensch-Chatbot-Interaktion. In dieser Studie wird untersucht, welche Rolle die menschenähnliche Gestaltung (u. a. menschlicher Name, Begrüßung, menschlicher Avatar) eines Reiseimpfberatungschatbots auf dessen Wahrnehmung durch Nutzer:innen hat. Spezifisch geht es darum zu verstehen, ob und wie sich Anthropomorphismus (die Wahrnehmung von Menschlichkeit und sozialer Präsenz in Objekten, Tieren und Maschinen) auf die wahrgenommene Vertrauenswürdigkeit und letztendliche Zufriedenheit mit dem Service auswirkt. In einem Online-Experiment mit 78 Teilnehmer:innen, wurden zwei Chatbot-Gestaltungen (mit menschenähnlichen Gestaltungselementen vs. ohne diese Elemente) verglichen. Die Ergebnisse zeigen, dass die wahrgenommene soziale Präsenz signifikant die wahrgenommene Vertrauenswürdigkeit und die Zufriedenheit mit dem Service des Chatbots erhöhen. Somit ist die Implikation, dass bei der Reiseimpfberatung und ähnlichen Beratungsprozessen eine menschliche Gestaltung der Sprachassistenten zu empfehlen ist.
Leftovers are particularly at risk of being discarded, and therefore a main component of household food waste. This study provides insights into sources of heterogeneity in leftover management behaviours, with a particular focus on the use of meal kits providing matched portion and ingredient sizes, and identifies consumer segments via a latent class analysis. We investigate whether belonging to a segment with positive attitudes toward leftovers, and engagement in conscious leftover management behaviours decreases the amounts of dinner leftovers and food waste. Besides, we demonstrate that several food waste antecedents, emotions, personal norms, intention and dinner procurement routines elicit leftover management segment membership. In addition to examining such individual differences, we also investigate the role of meal-level determinants, in particular, whether meal kits heterogeneously affect dinner leftovers depending on the consumer's leftover management segment. Data was collected from 868 households from six countries, using an online survey and diaries. Results of the latent class analysis point towards five consumer segments. We found differences in dinner leftovers amount across classes and detected heterogeneous effects of meal kits. That is, meal kits were able to diminish leftovers in two segments, but not in the other segments. These results provide novel insights into consumer heterogeneity regarding the occurrence, antecedents, and potential solutions of leftovers and resulting household food waste. Implications for both theory and policy are discussed.
Families face many barriers in providing nutritious home-cooked family meals. Meal kit subscription services are increasingly popular among families and may address obstacles to cooking at home and facilitate shared family meals. This study aimed to understand why families use meal kits and what they perceived to be the main impacts on family dynamics, nutrition, social and mental health. Sixteen primary meal providers with at least one child 18 years and under living at home, were recruited via social media (e.g., Facebook, Twitter) to participate in a semi-structured interview using Zoom videoconferencing. Interviews were conducted with participants who currently purchased and used commercially available meal kits in Australia (e.g., HelloFresh, Marley Spoon). Thematic analysis of interview transcripts revealed that women, as the primary carers responsible for family meals, primarily reported the role that meal kits played in reducing their mental load through reduced food-related decision making, enhanced family participation in meal preparation, and opportunities for food literacy. Additionally, meal kits were reported to reduce food eaten away-from-home with the majority of participants perceiving meal kits to provide nutritionally dense meals and appropriate portion sizes aligned with National dietary guidelines. This study provides important insights into the potential physical, mental and social health benefits of meal kits in supporting families to cook and eat meals together at home. While meal kits have the capacity to positively influence population health and wellbeing, it is necessary that meal kit subscription services address the nutritional quality of their meals and provide evidence-based nutrition messaging to facilitate improvements in food literacy and nutritional intake.
People are cooking at home less often and relying more on food prepared outside of the home, which is often of less nutritional value than home-cooked meals. The foodservice industry has endeavored to address barriers with the introduction of commercial meal kit subscription services (MKSSs). We aimed to assess and compare the nutritional qualities of MKSSs available in Australia. Average nutritional qualities per serve of 12 recipes (from four weekly boxes of three meals serving two people) were analyzed from five MKKSs (Dinnerly, HelloFresh™, MarleySpoon™, Pepper Leaf, Thomas Farms Kitchen). On average, MKSSs provided adequate serves of core foods, particularly of vegetables (2.3 ± 1.6-3.1 ± 1.8 serves per serve). Energy content ranged between 2891 ± 539 and 3904 kJ ± 890 per serve. All MKKSs were high in fat (39.5 ± 9.5-59.6 ± 11.2% of energy) and sodium (723 ± 404-1426 ± 688 mg per serve). All MKSSs met suggested dietary target level of dietary fiber for women, but none for men. If MKSS providers can modify recipes to reduce added salt and fat and increase dietary fiber, they have the potential to provide both men and women with nutritious meals that more closely align with the dietary guidelines for the prevention of chronic disease, especially if meals are used as an alternative to energy-dense nutrient-poor takeaway and convenience foods.
Meal kits are popular for consumers seeking greater convenience in preparing meals at home. The market share for meal kit subscription services (MKSSs) is growing in developed nations including Australia, however, literature about their health promoting qualities, e.g. nutritional composition, is scarce. This study aimed to assess the characteristics and nutritional composition of meals offered from an MKSS over 12 months. Nutritional data were extracted from recipes available to order from HelloFresh in Australia from 1 July 2017 to 30 June 2018. In total, 346 (251 unique) recipes were retrieved. Per serve (median size 580 g), meals contained a median of 2840 kJ (678 kcal) of energy, 58 g carbohydrate (14 g sugar), 44 g protein, 28 g total fat (8 g saturated fat) and 839 mg sodium. Median energy from macronutrients was total fat (38%), carbohydrates (34%), protein (25%) and saturated fat (11%). This paper is the first to describe characteristics of recipes available from an MKSS over a 12-month period of time. With their growing popularity, meal kit delivery services have the capacity to influence consumer food behaviours, diets and subsequently population health. MKSSs may function to promote health though education, training, and enabling home cooking behaviours, and may be a powerful commitment device for home cooking behaviour change. However, it is important for health professionals, including dietitians and nutritionists, to understand the nutritional risks, benefits and suitability of this contemporary mealtime option before recommending them to clients and members of the public as part of health promotion. Meal kit delivery services are growing in popularity in developed countries, complementing busy lifestyles with pre-measured ingredients and recipe instructions delivered to the home. These meal kits have the ability to influence consumer diets and population health, and may support health promoting diet behaviours, e.g. eating vegetables, and enable home cooking. In this study, we reviewed a years' worth of recipes from a popular meal kit service. We report that a typical recipe contained approximately nine different ingredients, comprising three vegetables and required three ingredients from the home pantry. Meals took ∼35 min to prepare and were found to be relatively high in energy from fat and protein, and relatively low in energy from carbohydrates. The level of sodium varied widely and some meals exceeded the Australian Suggested Dietary Target for sodium (<2000 mg). Meal kit recipes were found to have health promoting qualities, frequently including vegetable ingredients, however, improvements to recipes would make these meal kits more health promoting. Current diet intakes and the nutritional composition of meal kits recipes should be reviewed before being recommended by health professionals.
Scientists have identified new clues that could help astronomers spot one of the most famous hypothetical alien megastructures: a Dyson sphere。 The study finds that red dwarfs and white dwarfs are the most promising stars to examine, since advanced civilizations could potentially build energy-harvesting swarms around them more easily。 These objects
With residential proxies all the rage, CISA urges router users to be vigilant
Researchers solved the mystery of how soft lithium dendrites crack the hard ceramic inside solid-state batteries, triggering short circuits。 The breakthrough could help engineers build safer, longer-lasting batteries for smartphones, electric vehicles, and other electronics
NASA is marking the United States' 250th birthday with four striking red, white, and blue images of deep space from the Chandra X-ray Observatory。 The collection features an exploded star, a stellar nursery, a galaxy where stars are rapidly forming, and a galaxy cluster that provides evidence for dark matter
Gold may have a secret self-defense system that helps it resist tarnishing。 Researchers discovered that atoms on gold surfaces reorganize themselves into patterns that block oxygen from reacting with the metal, suppressing oxidation by up to a trillion-fold。 Beyond explaining why gold jewelry stays bright for generations, the finding could help sci
Conductive ink is painted directly onto the skin in colorful custom designs, drying into working electrodes
Four nearby white dwarf stars have been discovered hiding in plain sight beside brighter red dwarf companions。 Hubble's ultraviolet observations finally revealed the long-hidden stellar remnants, including one just 25 light-years away that took nearly three decades to confirm。 The findings match long-standing predictions and suggest our corner of t
The biggest smartphone makers keep on trucking in the face of component shortages and economic uncertainty
NASA's Hubble Space Telescope has captured a spectacular red, white, and blue view of one of the Milky Way's oldest star clusters to celebrate the nation's 250th anniversary。 Hidden within the ancient cluster are clues to how exploding stars helped transform the young universe into one capable of forming planets and, eventually, life
A detailed independent study found that SpaceX's Starship is every bit as revolutionary as expected, while revealing both its impressive capabilities and its biggest remaining hurdles。 It also introduces an ambitious European rocket concept that could offer a very different route to affordable super heavy launches
Scientists have developed a new framework that could finally apply the laws of thermodynamics to real, ever-changing black holes instead of only perfectly stable ones。 The advance may improve our understanding of black hole mergers, evaporation, and the powerful gravitational wave events detected by observatories like LIGO
K2-18b is one of the most promising worlds for the search for extraterrestrial life, so astronomers conducted an unusually powerful radio survey using both the VLA and MeerKAT telescopes。 Advanced software analyzed millions of signals, filtering out Earth-based interference and other false positives。 No convincing artificial radio transmissions wer