Even after TikTok removed swastika jewelry from its online shop, I was algorithmically nudged toward a web of Nazi-related products during searches, like "double lightning bolt" and "ss" necklaces. My journey on TikTok Shop started out with a search for "hip hop jewelry." It's an innocuous search query multiple users have likely typed in, hoping to find something to wear. While browsing the cheap jewelry, I was struck by what TikTok's algorithm repeatedly suggested that I might also be interested in: jewelry with blatant Nazi symbolism. TikTok continues to struggle with moderation as its in-app ecommerce store gains traction with younger users.

The chatbot-enabled Friend necklace eavesdrops on your life and provides a running commentary that's snarky and unhelpful. Worse, it can also make the people around you uneasy. The AI-powered Friend pendant is now out in the world. If you live in the US or Canada, you can buy one for $129. The smooth plastic disc is just under 2 inches in diameter; it looks and feels a little like a beefy Apple AirTag. Inside are some LEDs and a Bluetooth radio that connects you (through your iPhone) to a chatbot in the cloud that's powered by Google's Gemini 2.5 model. You can tap on the disc to ask your Friend questions as it dangles around your neck, and it responds to your voice prompts by sending you text messages through the companion app.

Recent advances in text-to-image (T2I) generation have led to impressive visual results. However, these models still face significant challenges when handling complex prompt, particularly those involving multiple subjects with distinct attributes. Inspired by the human drawing process, which first outlines the composition and then incrementally adds details, we propose Detail++, a training-free framework that introduces a novel Progressive Detail Injection (PDI) strategy to address this limitation. Specifically, we decompose a complex prompt into a sequence of simplified sub-prompts, guiding the generation process in stages. This staged generation leverages the inherent layout-controlling capacity of self-attention to first ensure global composition, followed by precise refinement. To achieve accurate binding between attributes and corresponding subjects, we exploit cross-attention mechanisms and further introduce a Centroid Alignment Loss at test time to reduce binding noise and enhance attribute consistency. Extensive experiments on T2I-CompBench and a newly constructed style composition benchmark demonstrate that Detail++ significantly outperforms existing methods, particularly in scenarios involving multiple objects and complex stylistic conditions.

Clustered Moore automata (CMA) are subsequen-tial Moore transducers whose states can contain smaller CMA that operate on a faster timescale, subject to an Artinian limitation.

The new year is a time for reflection, renewal, and rampant speculation about what wonders (or fresh hell) the future might hold. No place does this mix of anxiety and forward-looking techno-evangelism spring forth more profusely than at CES. The giant consumer tech showcase is barreling down on Las Vegas starting January 7, bringing with it a whirlwind of fuss about the newest gadgets and devices. And yes, you bet all these things are going to be packed full of AI features. You're probably going to be asked to wear many of them.

Long-term memory is important for chatbots and dialogue systems (DS) to create consistent and human-like conversations, evidenced by numerous developed memory-augmented DS (MADS). To evaluate the effectiveness of such MADS, existing commonly used evaluation metrics, like retrieval accuracy and perplexity (PPL), mainly focus on query-oriented factualness and language quality assessment. However, these metrics often lack practical value. Moreover, the evaluation dimensions are insufficient for human-like assessment in DS. Regarding memory-recalling paradigms, current evaluation schemes only consider passive memory retrieval while ignoring diverse memory recall with rich triggering factors, e.g., emotions and surroundings, which can be essential in emotional support scenarios. To bridge the gap, we construct a novel Memory-Augmented Dialogue Benchmark (MADail-Bench) covering various memory-recalling paradigms based on cognitive science and psychology theories. The benchmark assesses two tasks separately: memory retrieval and memory recognition with the incorporation of both passive and proactive memory recall data. We introduce new scoring criteria to the evaluation, including memory injection, emotion support (ES) proficiency, and intimacy, to comprehensively assess generated responses. Results from cutting-edge embedding models and large language models on this benchmark indicate the potential for further advancement. Extensive testing further reveals correlations between memory injection, ES proficiency, and intimacy.

If you want to coast through meetings, keep track of everyone you meet, or just remember the name of that obscure dog food your veterinarian told you to feed your pooch, there's a necklace for that. Plaud is an AI company that makes the creatively named Plaud Note--a slim ChatGPT-enabled audio recorder that can be stuck on the back of your phone or slipped into a shirt pocket to record, transcribe, and summarize your conversations. The company's newest offering is called the Plaud NotePin (the naming scheme doesn't get any better here), and it takes basically all the same features of the Note and packs them into a wearable device about the size of a lipstick tube. The NotePin can be worn as a necklace, a wristwatch, or a pin, or clipped onto something like a lapel. It costs 169 and lets you record up to 300 minutes of audio per month.

Circuits are supposed to accurately describe how a neural network performs a specific task, but do they really? We evaluate three circuits found in the literature (IOI, greater-than, and docstring) in an adversarial manner, considering inputs where the circuit's behavior maximally diverges from the full model. Concretely, we measure the KL divergence between the full model's output and the circuit's output, calculated through resample ablation, and we analyze the worst-performing inputs. Our results show that the circuits for the IOI and docstring tasks fail to behave similarly to the full model even on completely benign inputs from the original task, indicating that more robust circuits are needed for safety-critical applications.

Large Language Models (LLMs) have shown superior capability to solve reasoning problems with programs. While being a promising direction, most of such frameworks are trained and evaluated in settings with a prior knowledge of task requirements. However, as LLMs become more capable, it is necessary to assess their reasoning abilities in more realistic scenarios where many real-world problems are open-ended with ambiguous scope, and often require multiple formalisms to solve. To investigate this, we introduce the task of reasoning in the wild, where an LLM is tasked to solve a reasoning problem of unknown type by identifying the subproblems and their corresponding formalisms, and writing a program to solve each subproblem, guided by a tactic. We create a large tactic-guided trajectory dataset containing detailed solutions to a diverse set of reasoning problems, ranging from well-defined single-form reasoning (e.g., math, logic), to ambiguous and hybrid ones (e.g., commonsense, combined math and logic). This allows us to test various aspects of LLMs reasoning at the fine-grained level such as the selection and execution of tactics, and the tendency to take undesired shortcuts. In experiments, we highlight that existing LLMs fail significantly on problems with ambiguous and mixed scope, revealing critical limitations and overfitting issues (e.g. accuracy on GSM8K drops by at least 50\%). We further show the potential of finetuning a local LLM on the tactic-guided trajectories in achieving better performance. Project repo is available at github.com/gblackout/Reason-in-the-Wild

If you're looking for a birthday gift for the Pokémon fan who has everything (and we mean, every toy, card, item of apparel, game, Happy Meal collectable, etc.), the jewelry brand King Ice may have the solution. King Ice posted a photo on X and its website Wednesday teasing a new line of Pokémon jewelry pieces. The photo features the familiar face of Pikachu with a Poké Ball on his head decked out entirely in jewels. The X caption reads: "Collection dropping 6/14/24." This is not the first time the jewelry and clothing brand have collaborated with a big video game franchise. King Ice also sells a line of bejeweled Xbox themed necklaces, rings and earrings.