One of the most prestigious honors we award products is inclusion on our Buy It for Life gear roundup. This list represents products that WIRED writers have personally used for years, and as the name implies, they should last you for the rest of your life with proper care and warranty support. There’s only one coffee maker on that list, the Moccamaster KBGV Select, and you can currently pick it up from Amazon for up to $40 off its list price, depending on the color.

These drip coffee makers are seriously built to last, handmade in the Netherlands with solid steel and copper components. They’re fully repairable, which means they’ll keep churning out hot mugs of perfect coffee even after the five-year warranty ends. There are a variety of models, but we like the KBGV Select because it can also brew a half carafe instead of a full carafe, a useful trick for smaller households or an afternoon energy burst.

Extremely precise temperature control means you get excellent coffee every time, managing to consistently heat within a range of 4 degree Celsius. Technivorm is one of less than a dozen companies producing SCA-certified coffee makers for home use, and the Moccamaster models take up a noticeable chunk of that list.

It has all the features you’d expect from a drip coffee maker, like a hot plate for the carafe that has an automatic shut off, which automatically adjusts temperature based on whether you brewed a full or half carafe. The reservoir is 1.25 liters, so you can brew up to 10 cups of coffee at once, and it takes just four to six minutes from start to finish.

This model is available in a huge variety of colors, and your discount will vary based on which you think will match your kitchen best. I found the best price of $317 on the Turquoise, with the Apricot and Matte Black right behind at $320, as well as lesser discounts on the Off-White, Polished silver, and Juniper varieties. While we think it’s worth spending the extra cash for something that will last you years to come, you can always peruse our other favorite coffee makers if you’re looking for something more wallet-friendly.

To train artificial intelligence (AI) models, researchers need good data and lots of it. However, most real-world data has already been used, leading scientists to generate synthetic data. While the generated data helps solve the issue of quantity, it may not always have good quality, and assessing its quality has been overlooked.

Wei Gao, associate professor of electrical and computer engineering at the University of Pittsburgh Swanson School of Engineering, has collaborated with researchers from Peking University to develop analytical metrics to qualitatively evaluate the quality of synthetic wireless data. The researchers have created a novel framework that significantly improves the task-driven training of AI models using synthetic wireless data.

Their work is detailed on the arXiv preprint server in a study titled “Data Can Speak for Itself: Quality-Guided Utilization of Wireless Synthetic Data,” which received the Best Paper Award in June at the MobiSys 2025 International Conference on Mobile Systems, Applications, and Services.

Assessing affinity and diversity

“Synthetic data is vital for training AI models, but for modalities such as images, video, or sound, and especially wireless signals, generating good data can be difficult,” said Gao, who also directs the Pitt Intelligent Systems Laboratory.

Gao has developed metrics to quantify affinity and diversity, essential qualities for synthetic data to be used for effectively training AI models.

“Generated data shouldn’t be random,” said Gao. “Take human faces. If you’re training an AI model to identify human faces, you need to ensure that the images of faces represent actual faces. They can’t have three eyes or two noses. They must have affinity.”

The images also need diversity. Training an AI model on a million images of an identical face won’t achieve much. While the faces must have affinity, they must also be different, as human faces are. As Gao noted, “AI models learn from variation.”

Different tasks have different requirements for judging affinity and diversity. Recognizing a specific human face is different than distinguishing it from that of a dog or a cat, with each task having unique data requirements. Therefore, in systemically assessing the quality of synthetic data, the team applied a task-specific approach.

“We applied our method to downstream tasks and evaluated the existing work of synthesizing data,” said Gao. “We found that most synthetic data achieved good diversity, but some had problems satisfying affinity, especially wireless signals.”

The challenge of synthetic wireless data

Today, wireless signals are used in technologies such as home and sleep monitoring, interactive gaming, and virtual reality. Cell phone and Wi-Fi signals, as radio waves, hit objects and bounce back toward their source. These signals can be interpreted to indicate everything from sleep patterns to the shape of a person sitting on a couch.

To advance this technology, researchers need more wireless data to train models to recognize human behaviors in the signal patterns. However, as a waveform, the signals are difficult for humans to evaluate.

It’s not like human faces, which can be clearly defined. “Our research found that current synthetic wireless data is limited in its affinity,” said Gao. “This leads to mislabeled data and degraded task performance.”

To improve affinity in wireless signals, the researchers took a semi-supervised learning approach. “We used a small amount of labeled synthetic data, which was verified as legitimate,” Gao said. “We used this data to teach the model what is and isn’t legitimate.”

Gao and his collaborators developed SynCheck, a framework that filters out synthetic wireless samples with low affinity and labels the remaining samples during iterative training of a model.

“We found that our system improves performance by 4.3% whereas a nonselective use of synthetic wireless data degrades performance by 13.4%,” Gao noted.

This research takes an important first step toward ensuring not just an endless stream of data, but of quality data that scientists can use to train more sophisticated AI models.

The rise of e-commerce has brought unprecedented convenience to consumers, but it has also created fertile ground for deceptive practices in online marketplaces. A growing body of research is now focusing on the detection of fake or misleading product reviews, often referred to as spam reviews. These are deliberately written to either unfairly promote a product or damage a competitor’s reputation.

These reviews frequently use fabricated profiles or carefully crafted language, making them difficult to distinguish from genuine customer feedback. Moreover, the use of large language models, colloquially known as generative AI, are now being used to generate authentic-seeming spam reviews.

The impact of spam reviews is significant. Consumers may be persuaded to purchase low-quality goods, while legitimate businesses suffer reputational harm. Ultimately, this might erode trust in digital marketplaces. However, distinguishing between authentic opinions and deceptive ones is difficult.

For their article published in the International Journal of Services, Economics and Management researchers have turned to computational opinion mining, which involves analyzing text to extract sentiment and meaning, to detect patterns indicative of fraudulent activity.

Traditional techniques include filtering for suspicious keywords, monitoring abnormal posting patterns, assessing reviewer credibility, and employing verification tools such as anti-spam CAPTCHAs.

More recently, advances in machine learning (ML) and natural language processing (NLP), which allows a computer to interpret human language, have enabled automated systems to detect the subtle linguistic and contextual cues that often reveal fabricated content.

The researchers explain that central to their approach is the creation of ground truth datasets. These are curated examples of real and fake reviews. These datasets provide a reference for training machine learning models to recognize subtle indicators of deception, including unusual writing styles, sentiment inconsistencies, or anomalies in sentence structure.

The new approach then combines multiple algorithms into a hybrid classifier. A deep learning framework, such as a convolutional neural network (CNN), which is adept at identifying complex patterns, is paired with a traditional statistical classifier. The accuracy rate of this hybrid is between 96% and 99% when tested on standard datasets.

As global e-commerce continues to expand, accurate spam detection systems will become increasingly important in maintaining the reliability of digital marketplaces, reinforcing transparency and trustworthiness.