Volvo is looking to boost its reputation for safety with the release of a new “multi-adaptive safety belt” that uses real-time data from the vehicle’s sensors to better protect the person wearing it.Seatbelt technology hasn’t changed much since Volvo patented one of the first modern three-point safety belts in the early 1960s. But cars have changed significantly, adding sensors, cameras, and high-powered computers to power advanced driver assist features and anti-crash technology.Now, Volvo wants to put those gadgets to work for seatbelts. Modern safety belts use load limiters to control how much force the safety belt applies on the human body during a crash. Volvo says its new safety belt expands the load-limiting profiles from three to 11 and increases the possible number of settings, enabling it to tailor its performance to specific situations and individuals.As such, Volvo can use sensor data to customize seatbelts based on a person’s height, weight, body shape, and seating position. A larger occupant, for example, would receive a higher belt load setting to help reduce the risk of a head injury in a crash, while a smaller person in a milder crash would receive a lower belt load setting to reduce the risk of rib fractures.During a crash, Volvo says its vehicles’ safety systems will share sensor data — such as direction, speed, and passenger posture — with multi-adaptive seatbelts to determine how much force to apply to the occupant’s body. And using over-the-air software updates, Volvo promises that the seatbelts can improve over time.Volvo says its new safety belt expands the load-limiting profiles and increases the possible number of settings. Image: VolvoVolvo has previously deviated from traditional practices to introduce new technologies meant to underscore its commitment to safety. The company limits the top speed on all of its vehicles to 112 mph — notably below the 155 mph established by a “gentleman’s agreement” between Mercedes-Benz, Audi, and BMW to reduce the number of fatalities on the Autobahn.The new seatbelts will debut in the Volvo EX60, the automaker’s mid-sized electric SUV which is scheduled to come out next year.See More:

Volvo has introduced a new seatbelt technology that can customize the protection it provides in real time. The "multi-adaptive safety belt" system, as the automaker is calling it, uses data input from both interior and exterior sensors to change protection settings based on various factors. It can take a person's height, weight, body shape and seating position into account, as well as the direction and speed of the vehicle. The system can communicate all those information to the seatbelt "in the blink of an eye" so that it can optimize protection for the passenger.  If the passenger is on the larger side, for instance, they will receive a higher belt load setting to reduce the risk of a head injury in the event of a serious crash. For milder crashes, someone with a smaller frame will receive a lower belt load setting to prevent rib injuries. Volvo didn't specifically say if the system also takes the position of a seatbelt on women into account, since it doesn't always fit right over a woman's chest. However, the automaker explained that the system expands the number of load-limiting profiles to 11. Load limiters control how much force a seatbelt applies on the body during a crash. Typically, seatbelts only have three load-limiting profiles, but Volvo expanding them to 11 means the system can better optimize the protection a passenger gets.  Volvo used information from five decades of safety research and from a database of over 80,000 people involved in real-life accidents to design the new safety belt. The system was also created to incorporate improvements rolled out via over the-air software updates, which the company expects to release as it gets more data and insights.  "The world first multi-adaptive safety belt is another milestone for automotive safety and a great example of how we leverage real-time data with the ambition to help save millions of more lives," said Åsa Haglund, head of Volvo Cars Safety Centre. "This marks a major upgrade to the modern three-point safety belt, a Volvo invention introduced in 1959, estimated to have saved over a million lives." Volvo engineer Nils Bohlin designed the modern three-point seatbelt and made its patent available for use by other automakers. The company didn't say whether it'll be as generous with the multi-adaptive safety belt, but the new system will debut in the all-electric Volvo EX60 midsize SUV sometime next year.This article originally appeared on Engadget at https://www.engadget.com/transportation/volvo-is-introducing-the-first-multi-adaptive-seatbelt-technology-on-the-ex60-ev-070017403.html?src=rss

Latest   Machine Learning Beyond the Prompt: What Google’s LLM Advice Doesn’t Quite Tell You 0 like May 18, 2025 Share this post Author(s): Mayank Bohra Originally published on Towards AI. Image by the author Alright, let’s talk about prompt engineering. Every other week, it seems there is a new set of secrets or magical techniques guaranteed to unlock AI perfection. Recently, a whitepaper from Google made the rounds, outlining their take on getting better results from Large Language Models. Look, effective prompting is absolutely necessary. It’s the interface layer, how we communicate our intent to these incredibly powerful, yet often frustrating opaque, models. Think of it like giving instructions to a brilliant but slightly eccentric junior engineer who only understands natural language. You need to be clear, specific, and provide context. But let’s be pragmatic. The idea that a few prompt tweaks will magically “10x” your results for every task is marketing hype, not engineering reality. These models, for all their capabilities, are fundamentally pattern-matching machines operating within a probabilistic space. They don’t understand in the way a human does. Prompting is about nudging that pattern matching closer to the desired outcome. So, what did Google’s advice cover, and what’s the experience builder’s take on it? The techniques generally boil down to principles we’ve known for a while: clarity, structure, providing examples and iteration. The Fundamentals: Clarity, Structure, Context Much of the advice centers on making your intent unambiguous. This is ground zero for dealing with LLMs. They excel at finding patterns in vast amounts of data, but they stumble on vagueness. Being Specific and Detailed: This isn’t a secret; it’s just good communication. If you ask for “information about AI”, you’ll get something generic. If you ask for “a summary of recent advancements in Generative AI model architecture published in research papers since April 2025, focusing on MoE models”, you give the model a much better target. Defining Output Format: Models are flexible text generators. If you don’t specify structure (JSON, bullet points, a specific paragraph format), you’ll get whatever feels statistically probable based on the training data, which is often inconsistent. Telling the model “Respond in JSON format with keys ‘summary’ and ‘key_findings’” isn’t magic; it’s setting clear requirements. Providing Context: Models have limited context windows. Showing your entire codebase or all user documentation in won’t work. You need to curate teh relevant information. This principle is the entire foundation of Retrieval Augmented Generation (RAG), where you retrieve relevant chunks of data and then provide them as context to the prompt. Prompting alone without relevant external knowledge only leverage the model’s internal training data, which might be outdated or insufficient for domain-specific tasks. These points are foundational. They’re less about discovering hidden model behaviors and more about mitigating the inherent ambiguity of natural language and the model’s lack of true world understanding. Structuring the Conversation: Roles and Delimiters Assigning a role (“Act as an expert historian…”) or using delimiters (like “` or — -) are simple yet effective ways to guide the model’s behavior and separate instructions from input. Assigning a Role: This is a trick to prime the model to generate text consistent with a certain persona or knowledge domain it learned during training. It leverage the fact that the model has seen countless examples of different writing styles and knowledge expressions. It works, but it’s a heuristic, not a guarantee of factual accuracy or perfect adherence to the role. Using Delimiters: Essential for programmatic prompting. When you’re building an application that feeds user input into a prompt, you must use delimiters (e.g., triple backticks, XML tags) to clearly separated the user’s potentially malicious input from your system instructions. This is a critical security measure against prompt injection attacks, not just a formatting tip. Nudging the Model’s Reasoning: Few-shot and Step-by-Step Some techniques go beyond just structuring the input; they attempt to influence the model’s internal processing. Few-shot Prompts: Providing a few examples of input/output pairs (‘Input X → Output Y’, Input A → Output B, Input C→ ?’) if often far more effective than just describing the task. Why? Because the model learns the desired mapping from the examples. It’s pattern recognition again. This is powerful for teaching specific formats or interpreting nuanced instructions that hard to describe purely verbally. It’s basically in-context learning. Breaking Down Complex Tasks: Asking the model to think step-by-step (or implementing techniques like Chain-of-Thought or Tree-of-Thought prompting outside the model) encourages it to show intermediate steps. This often leads to more accurate final results, especially for reasoning-heavy tasks. Why? It mimics hwo humans solve problems and forces the model to allocate computational steps sequentially. It’s less about a secret instruction and more about guiding the model through a multi-step process rather than expecting it to leap to the answer in one go. The Engineering Angle: Testing and Iteration The advice also includes testing and iteration. Again, this isn’t unique to prompt engineering. It’s fundamental to all software development. Test and Iterate: You write a prompt, you test it with various inputs, you see where it fails or is suboptimal, you tweak the prompt, and you test again. This loop is the reality of building anything reliable with LLMs. It highlights that prompting is often empirical; you figure out what works by trying it. This is the opposite of a predictable, documented API. The Hard Truth: Where Prompt Engineering Hits a Wall Here’s where the pragmatic view really kicks in. Prompt engineering, while crucial, has significant limitations, especially for building robust, production-grade applications: Context Window Limits: There’s only so much information you can cram into a prompt. Long documents, complex histories, or large datasets are out. This is why RAG systems are essential — they manage and retrieve relevant context dynamically. Prompting alone doesn’t solve the knowledge bottleneck. Factual Accuracy and Hallucinations: No amount of prompting can guarantee a model won’t invent facts or confidently present misinformation. Prompting can sometimes mitigate this by, for examples, telling the model to stick only to the provided context (RAG), but it doesn’t fix the underlying issue that the model is a text predictor, not a truth engine. Model Bias and Undesired Behavior: Prompts can influence output, but they can’t easily override biases embedded in the training data or prevent the model from generating harmful or inappropriate content in unexpected ways. Guardrails need to be implemented *outside* the prompt layer. Complexity Ceiling: For truly complex, multi-step processes requiring external tool use, decision making, and dynamic state, pure prompting breaks down. This is the domain of AI agents, which use LLMs as the controller but rely on external memory, planning modules, and tool interaction to achieve goals. Prompting is just one part of the agent’s loop. Maintainability: Try managing dozens or hundreds of complex, multi-line prompts across different features in a large application. Versioning them? Testing changes? This quickly becomes an engineering nightmare. Prompts are code, but often undocumented, untestable code living in strings. Prompt Injection: As mentioned with delimiters, allowing external input (from users, databases, APIs) into a prompt opens the door to prompt injection attacks, where malicious input hijacks the model’s instructions. Robust applications need sanitization and architectural safeguards beyond just a delimiter trick. What no one tells you in the prompt “secrets” articles is that the difficulty scales non-linearly with the reliability and complexity required. Getting a cool demo output with a clever prompt is one thing. Building a feature that consistently works for thousands of users on diverse inputs while being secure and maintainable? That’s a whole different ballgame. The Real “Secret”? It’s Just Good Engineering. If there’s any “secret” to building effective applications with LLMs, it’s not a prompt string. It’s integrating the model into a well-architected system. This involves: Data Pipelines: Getting the right data to the model (for RAG, fine-tuning, etc.). Orchestration Frameworks: Using tools like LangChain, LlamaIndex, or building custom workflows to sequence model calls, tool use, and data retrieval. Evaluation: Developing robust methods to quantitatively measure the quality of LLM output beyond just eyeballing it. This is hard. Guardrails: Implementing safety checks, moderation, and input validation *outside* the LLM call itself. Fallback Mechanisms: What happens when the model gives a bad answer or fails? Your application needs graceful degradation. Version Control and Testing: Treating prompts and the surrounding logic with the same rigor as any other production code. Prompt engineering is a critical *skill*, part of the overall toolkit. It’s like knowing how to write effective SQL queries. Essential for database interaction, but it doesn’t mean you can build a scalable web application with just SQL. You need application code, infrastructure, frontend, etc. Wrapping Up So, Google’s whitepaper and similar resources offer valuable best practices for interacting with LLMs. They formalize common-sense approaches to communication and leverage observed model behaviors like few-shot learning and step-by-step processing. If you’re just starting out, or using LLMs for simple tasks, mastering these techniques will absolutely improve your results. But if you’re a developer, an AI practitioner, or a technical founder looking to build robust, reliable applications powered by LLMs, understand this: prompt engineering is table stakes. It’s necessary, but far from sufficient. The real challenge, the actual “secrets” if you want to call them that, lie in the surrounding engineering — the data management, the orchestration, the evaluation, the guardrails, and the sheer hard work of building a system that accounts for the LLM’s inherent unpredictability and limitations. Don’t get fixated on finding the perfect prompt string. Focus on building a resilient system around it. That’s where the real progress happens. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI Towards AI - Medium Share this post

Wireless mics fail when they rely too much on perfect conditions. BOYAMIC 2 fixes that by making every part of the system self-contained. Each transmitter records on its own. Each receiver controls levels, backups, and signal without needing an app. Noise is filtered in real time. Recording keeps going even if the connection drops. Designer: BOYAMIC There’s no need for a separate recorder or post-edit rescue. The unit handles gain shifts, background interference, and voice clarity without user intervention. Everything shows on screen. Adjustments happen through physical controls. Files are saved directly to internal memory. This system is built to capture clean audio without depending on external gear. It records immediately, adapts instantly, and stores everything without breaking the workflow. Industrial Design and Physical Form Each transmitter is small but solid. It’s 40 millimeters tall with a ridged surface that helps with grip and alignment. The finish reduces glare and makes handling easier. You can clip it or use the built-in magnet. Placement is quick, and it stays put. The record button is recessed, so you won’t hit it by mistake. An LED shows when it’s active. The mic capsule stays exposed but protected, avoiding interference from hands or clothing. Nothing sticks out or gets in the way.   The receiver is built around a screen and a knob. The 1.1-inch display shows battery, signal, gain, and status. The knob adjusts volume and selects settings. It works fast, without touchscreen lag. You can see and feel every change. Connections are spaced cleanly. One side has a USB-C port. The other has a 3.5 mm jack. A plug-in port supports USB-C or Lightning. The mount is fixed and locks into rigs without shifting. The charging case holds two transmitters and one receiver. Each has its own slot with magnetic contacts. Drop them in, close the lid, and they stay in place. LEDs on the case show power levels. There are no loose parts, exposed pins, or extra steps. Every shape and control supports fast setup and clear operation. You can press, turn, mount, and move without second-guessing. The design doesn’t try to be invisible; it stays readable, durable, and direct. Signal Processing and Audio Control BOYAMIC 2 uses onboard AI to separate voice from background noise. The system was trained on over 700,000 real-world sound samples. It filters traffic, crowds, wind, and mechanical hum in real time. Depending on the environment, you can toggle between strong and weak noise reduction. Both modes work directly from the transmitter or through the receiver. The mic uses a 6mm condenser capsule with a 48 kHz sample rate and 24-bit depth. The signal-to-noise ratio reaches 90 dB. Two low-cut filter options, at 75 Hz and 150 Hz, handle low-end rumble. These are effective against HVAC, engine hum, or low vibration. Gain is managed with automatic control. The system boosts quiet voices and pulls back when sound gets too loud. Built-in limiters stop clipping during spikes. A safety track records a second copy at -12 dB for backup. This makes it harder to lose a usable take even when volume jumps suddenly. Each setting is adjustable on screen. You don’t need a mobile app to access basic controls. Everything runs live and updates immediately. There are no delays or sync problems during capture. Recording and Storage Each transmitter records internally without needing the receiver. Files are saved in 32-bit float or 24-bit WAV formats. Internal storage is 8 GB. That gives you about ten hours of float audio or fifteen hours of 24-bit. When full, the system loops and overwrites older files. Recording continues even if the connection drops. Every session is split into timestamped chunks for fast transfer. You can plug the transmitter into any USB-C port and drag the files directly. No software is needed. This setup protects against signal loss, battery drops, or app crashes. The mic stays live, and the recording stays intact. Each transmitter runs for up to nine hours without noise cancellation or recording. With both features on, the runtime is closer to six hours. The receiver runs for about fifteen hours. The charging case holds enough power to recharge all three units twice. The system uses 2.4 GHz digital transmission. Its range can reach up to 300 meters in open areas. With walls or obstacles, it drops to around 60 meters. Latency stays at 25 milliseconds, even at long distances. You get reliable sync and stable audio across open ground or indoor spaces. Charging is handled through the included case or by direct USB-C. Each device takes under two hours to recharge fully. Compatibility and Multi-Device Support The system supports cameras, smartphones, and computers. USB-C and Lightning adapters are included. A 3.5 mm TRS cable connects the receiver to most cameras or mixers. While recording, you can charge your phone through the receiver, which is useful for long mobile shoots. One transmitter can send audio to up to four receivers at once, which helps with multi-angle setups or backup channels. The receiver also supports stereo, mono, and safety track modes. Based on your workflow, you choose how audio is split or merged. Settings can be changed from the receiver screen or through the BOYA app. The app adds firmware updates, custom EQ profiles, and gain presets for different camera brands. But the core controls don’t depend on it.The post BOYAMIC 2 Rebuilds Mobile Audio with AI and Onboard Capture first appeared on Yanko Design.