--- markdown_generated_at: "2026-11-04" title: "FAQs" url: "https://www.qualcomm.com/artificial-intelligence/what-is-ai-faq" --- Artificial Intelligence # FAQs ![](https://s7d1.scene7.com/is/image/dmqualcommprod/bg-blue-products-desktop-2?fmt=webp-alpha&qlt=85) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/ai-faq-desktop-1?$QC_Responsive$&fmt=png-alpha) ## What is artificial intelligence (AI)? Artificial intelligence (AI) refers to a broad set of technologies that enable software and machines to perform tasks that typically require human intelligence, such as understanding language, recognizing images, identifying patterns, and making decisions. AI is not a single technology. It includes machine learning, deep learning, generative AI, and agentic AI. Each of these builds on the others. - Machine learning learns patterns from data. - Deep learning uses neural networks with layers of interconnected nodes to process complex inputs like images and language. - Generative AI creates new content from natural inputs such as text, voice, images, or video. - Agentic AI enables systems to plan and act over time. Together, these technologies power applications ranging from recommendation and fraud detection to assistants, robots, devices, and autonomous systems. ## What is agentic AI? Agentic AI refers to systems that can reason, plan, and act over time rather than simply responding to a single request. It builds on generative AI by using inputs and outputs to drive ongoing tasks and decisions. These systems maintain context, determine next steps, and coordinate actions across tools or environments, adapting as conditions change. Agentic AI is well suited for multi-step workflows such as research, coding, operations, and customer support. ## How is agentic AI different from generative AI and traditional AI systems? Traditional AI systems (such as machine learning and deep learning) take a structured input, for example, an image, and produce a single output, such as a prediction. Generative AI creates new content by predicting patterns based on training data, enabling capabilities like text generation, image creation, or code completion. Agentic AI goes further by turning outputs into new inputs, maintaining context over time, reasoning across multiple steps, and coordinating actions across systems. On this page - [Systems](https://www.qualcomm.com/#systems) - [Capabilities](https://www.qualcomm.com/#capabilities) - [Connectivity](https://www.qualcomm.com/#connectivity) ## Systems ### How is AI evolving today? AI is evolving into systems that can reason and act, not just answer questions. At the same time, advances in hardware allow AI to run efficiently across devices, from wearables to data centers. This makes AI faster and more private by enabling it to run directly on devices without constant connectivity. Together, these shifts are making AI more responsive, scalable, and capable of operating continuously across environments. ### Why does it matter where AI runs? Where AI runs directly impacts performance. It affects how quickly results are delivered, how efficiently resources are used, and how reliable a system is. Time-sensitive tasks, such as responding to a voice command or analyzing a camera feed, often need to happen locally to minimize delay. Other tasks benefit from centralized processing and coordination. Balancing where AI runs allows systems to optimize for latency, privacy, reliability, and cost. ### What is distributed intelligence? **Distributed intelligence** is an approach to AI where processing and decision-making are spread across devices, edge systems, and the cloud, allowing them to work together as a coordinated system. This allows tasks to run where they are most effective, with time-sensitive work handled locally and larger-scale processing handled centrally. While it is often associated with hybrid AI, distributed intelligence specifically focuses on how these systems work together, not just where workloads are run. ### How does AI work across devices, edge systems, and the cloud? AI systems break tasks into parts and run each part in the most suitable location. Time-sensitive tasks, such as responding to a voice command or analyzing a camera feed, run locally to minimize delay. More complex processing, such as combining data from many sources or updating models, runs in the cloud where more computing power is available. These components continuously share data, allowing the system to operate as a single coordinated workflow. ### How is Qualcomm enabling distributed AI? Qualcomm Technologies provides a [unified software and hardware foundation](https://www.qualcomm.com/developer/artificial-intelligence) that allows AI workloads to move between environments based on latency, power, privacy, and performance needs. It supports [agentic AI systems](https://www.qualcomm.com/artificial-intelligence) that must operate in real time while coordinating intelligence across personal devices, shared systems, and centralized infrastructure. ### What are AI models and what limits their performance? An **AI model** is a system trained on data to recognize patterns and use them to make predictions, generate content, or act. Once trained, it can apply what it has learned to new inputs. Model performance depends on several factors, including data quality, computing resources, and speed. Large models can require significant memory and processing power, making them harder to run efficiently on devices. Maintaining consistent performance across devices, edge systems, and the cloud adds complexity. These challenges are addressed by distributing workloads across environments, so each task runs where it performs best. ### What is generative AI, and how does it fit into modern AI systems? **Generative AI** refers to systems that create new content, such as text, images, audio, or code, based on patterns learned from data. Unlike traditional approaches focused on prediction, generative AI produces original outputs and supports more flexible, interactive experiences. In modern systems, generative capabilities are often part of larger workflows, contributing to systems that can reason, act, and operate over time. ### What is an NPU, and how is it different from a CPU or GPU? **NPUs**, or neural processing units, are designed specifically for AI workloads and are optimized for efficient inference. **CPUs** (central processing units) are general-purpose processors that handle tasks sequentially and help coordinate overall system operations and agentic AI workloads. **GPUs** (graphics processing units) are designed for parallel processing. They are widely used for AI but typically consume more power than NPUs for AI computation. Because NPUs are designed for the types of computations used in neural networks, they deliver faster performance and greater energy efficiency for AI than CPUs or GPUs, making them especially well-suited for running AI directly on devices. ### What is AI bias, and how is it addressed in enterprise AI systems? **AI bias** occurs when a system produces unfair or inconsistent results because of patterns, gaps, or imbalances in training data, often shaped by human assumptions or subjective beliefs. Reducing these issues requires diverse training data, ongoing testing, and strong governance practices that monitor and improve systems over time to reduce risk and support trustworthy AI use. Enterprises combine technical safeguards with **governance frameworks** to help reduce risk, protect users, and support trustworthy AI deployment. ## Capabilities ### What is on-device AI, and why does it matter? **On‑device AI** runs models directly on a user’s device instead of sending data elsewhere. This improves responsiveness, protects local context and privacy, and allows systems to operate without constant connectivity. It is especially important for applications that require real-time interaction and continuous operation. ### What is edge AI, and when is it used? **Edge AI** refers to running AI models and inference at or near where data is generated. It is used when systems need to interact with the physical world in real time, such as in vehicles, industrial systems, or robotics. By processing data locally, edge AI enables faster responses, improves reliability, and supports continuous operation without relying on constant connectivity. ### Where is AI being used today? AI is already embedded in many everyday experiences. It powers personalized recommendations, enhances smartphone cameras, enables voice and facial recognition, and helps detect fraudulent activity. It is also widely used in medical imaging, real-time translation, industrial systems, and network optimization. Many of these applications operate in the background, making AI a seamless part of how devices, systems, and services work. ### What is Qualcomm doing in AI? Qualcomm has [invested in AI](https://www.qualcomm.com/research/artificial-intelligence) for more than a decade and is leading the transition to agent-centric computing. Qualcomm Technologies is scaling our performance-per-watt leadership across a pervasive footprint that includes smartphones, PCs, XR, automotive, industrial IoT, and 6G-ready network infrastructure. By combining AI, high-performance low-power computing, and connectivity, Qualcomm Technologies enables efficient, real-time intelligence across a wide range of industries and use cases. ## Connectivity ### How does 5G support AI? 5G enables faster data transfer, lower latency, and more reliable connectivity. This makes it easier to distribute AI workloads across devices and environments while maintaining real-time responsiveness. ### How is 5G different from Wi‑Fi for AI? 5G and Wi‑Fi both support AI workloads, but they are optimized for different environments. Wi‑Fi is best suited for fixed, localized environments, while 5G is designed for wide‑area coverage, mobility, and consistent performance across many devices. For AI systems that operate across locations or while moving, 5G provides stronger support. ### What role will 6G play in AI? 6G is expected to further enable AI across distributed environments by providing higher data rates, lower latency, and more intelligent networks. This will support more responsive, scalable, and continuously operating AI systems. [Learn more about the latest advances in 6G here.](https://www.qualcomm.com/research/6g)