Hacker News Front Page
중도
IT/기술
My thoughts after using Clojure for about a month
Comments
🇺🇸 미국 · IT/기술 · "THOUGHT" · 총 12건
필터 보기현재 지수
50.0
0 = 부정 우세
50 = 중립
100 = 긍정 우세
최근 7일 기준 10,505건을 분석한 결과, 뉴스 심리지수는 50.0(균형)입니다. 긍정 1건(0.0%)·중립 10,503건(100.0%)·부정 1건(0.0%)이며, 중립 비중이 뚜렷하게 높습니다. 성향 지수는 종합 19.3(중도 균형)입니다.
Comments
The executives responsible for keeping the world's biggest companies running thought AI would simplify their jobs. They were wrong.
Children born after 2013 are the first generation to grow up fully immersed in digital systems, which weren’t designed with them in mind. One‑third of the world’s Internet users are younger than 18, according to UNICEF, yet these systems shaping their daily lives were built for adults. They were optimized for engagement and designed long before people understood how profoundly digital environments influence children. For engineers and technical professionals, online safety is not an abstract policy debate. It is a design challenge that demands rigor, systems thinking, and ethical foresight. Governments around the world are also beginning to recognize the problem. Policymakers from across Australia, Brazil, the European Union, Indonesia, and the United States are responding to risks engineers have long understood: Addictive features, inappropriate content, opaque data practices, and algorithmic systems shape user behavior in ways that their creators did not fully predict. For years, technology moved faster than governance. Now governance is trying to catch up. Global Shift Toward Design Reform Supporting National Digital Ambitions In Athens this year I met with senior leaders of Greek government agencies and key national research institutions. Greece is moving quickly on digital transformation and responsible technology governance, and our discussions reinforced IEEE’s role as a trusted, neutral collaborator. We focused on supporting Greece’s ambitions in digital modernization and public‑sector innovation. We also discussed responsible AI and age-appropriate digital design in Europe and elsewhere. These engagements, grounded in shared values and long‑term commitment, strengthened IEEE’s presence within the European ecosystem and opened new pathways for collaboration on trustworthy AI and child‑focused digital well‑being. The European Union and the United Kingdom have been among the first to act, embedding age‑appropriate digital design into their broader children’s rights agenda. Drawing on IEEE expertise and global best practices, Indonesia is the first country in Asia, and Brazil is the first country in Latin America, to adopt age-appropriate design regulation. Australia is aiming to limit access to harmful content and addictive design features through age restrictions on certain platforms. And in the United States, in addition to federal efforts, states including California, New York, and Utah are enacting approaches including age-appropriate design principles. Across these efforts, a shared realization is emerging. Protecting children online is not simply about filtering content or adding parental controls. It requires rethinking the architecture of digital systems regarding how data is collected, how algorithms make decisions, how interfaces influence attention, and how AI interacts with the developing minds of young users. Engineers and technical professionals understand that design choices are never neutral. They encode values, incentives, and assumptions. When the user is a child, those choices carry greater weight. This is where IEEE’s work becomes more essential. Protecting Children Online For more than a decade, IEEE has been building technical and ethical foundations for safer digital experiences. The first IEEE standard on age-appropriate design in 2021 marked a turning point. It offers a structured, principled approach to designing with children’s rights in mind. The Institute’s 2022 article “Use a New IEEE Standard to Design a Safer Digital World for Kids” highlights how the standard helps translate those principles into engineering practice. Today the IEEE Standards Association’s (SA) Trustworthy Digital Experiences portfolio provides a practical, technically grounded framework for governments and industry. Spanning ethical design, data governance, algorithmic transparency, and child‑focused digital well‑being, it has already initiated discussions with government stakeholders around the world. This work helps bridge the gap between engineering realities and policy ambitions. No single country can solve these challenges alone. Many policymakers lack access to the combined expertise in technology, governance, and children’s rights needed to act quickly and effectively. This collaborative effort helps close that gap. The stakes are high. Without coordinated action, public policy will continue to lag behind technology, leaving children exposed to risks that could have been mitigated through thoughtful design. But with the right frameworks, governments can ensure digital systems respect children’s rights, support healthy development, and promote well‑being. IEEE’s emerging standards and collaborative technology policy work offer a path forward. By grounding national efforts in evidence‑based, rights-aligned design principles, IEEE is helping governments move from reactive regulation to proactive, coherent, and globally informed strategies for protecting children online. Safeguarding childhood in the digital age is both a moral imperative and an engineering challenge. And IEEE is helping to lead the way. —Mary Ellen Randall IEEE president and CEO Please share your thoughts with me: president@ieee.org. This article appears in the June 2026 print issue.
On Saturday, at around 2:06 pm ET, a meteor streaked over the northeastern US and exploded north of Cape Cod Bay. The fireball was caught on camera by several people, shook houses, and can even be seen clearly in satellite imagery, lighting up the sky. Some residents initially thought that the shaking and boom may […]
The crypto weed vape found me on 4/20, the high holiday of cannabis enthusiasts everywhere. It arrived over Slack with the thumbnail of a man exhaling a plume of vapor, the words "every hit delivers Bitcoin" emblazoned across it. It claimed to be advertising a device called Gudtrip, and I thought everything about it sounded […]
Created from fabric scraps that would otherwise end up in a landfill, this thoughtfully constructed bag looks great and can carry everything you need for a weeklong trip.
Comments
For $7 an hour, virtual assistants use AI tools to write LinkedIn posts and comments on behalf of Western executives.
I first met Robert Woo in 2011, during his third time walking in a powered exoskeleton. The architect had been paralyzed in a construction accident four years earlier, but he was determined to get back on his feet. Watching him clunk across a rehab room in an exoskeleton prototype, the technology felt astonishing. I had the same reaction when reporting on early brain-computer interfaces (BCIs), which enabled paralyzed people to move robotic arms or communicate by thought alone. Both types of bionic technology seemed to verge on magic. But that initial sense of awe, I’ve learned over many years of reporting on these technologies, is only a starting point. What matters is not what these systems can do in a carefully staged demo but how they perform in the real world. Do they work reliably? Can people with disabilities use them for their intended purposes? And what does it actually cost—in time, effort, and trade-offs—to do so? The question isn’t whether the technology looks impressive the first time but whether it holds up on the hundredth. The special report in this issue, “Cyborg Tech From the Inside” takes that perspective seriously. In my feature article on Woo, an exoskeleton super-user who has spent 15 years testing these systems, the story of the technology is inseparable from the story of its use. Woo’s relentless feedback has driven steady, incremental improvements. In Edd Gent’s reporting on the pioneers testing the earliest BCIs, the experience of these extraordinary technologies likewise resolves into something more complex. As one trial participant notes, these early adopters are like the first astronauts, who barely reached space before coming back down to Earth. Together, these stories reframe these individuals not as passive medical patients but as the ultimate beta testers and co-engineers of the bionic age. I saw the gap between demonstration and daily use firsthand when I interviewed Woo in a Manhattan showroom recently, where he was testing a new self-balancing exoskeleton from Wandercraft. The device is a striking advance that kept him upright without crutches, but it also revealed the friction of the real world. As Woo tried to walk out the door, barely an inch of slope on the Park Avenue sidewalk was enough to trigger the machine’s safety sensors and halt his progress. It was a stark reminder of how far these systems must evolve before they fit seamlessly into everyday life. For the people who use them, that seamless integration is the ultimate goal. Getting there will depend not just on technical breakthroughs but on how well these systems hold up outside controlled environments, over time, and under real conditions. Looking from the inside doesn’t make these technologies any less remarkable, but it does change how we judge them—not by what they can do once for a photo but by what they can sustain over a lifetime. That’s the standard their users have been applying all along. Our commitment to evaluating technology from the user’s perspective extends beyond this special report. To provide a necessary corrective to the “techno-solutionism” that often dominates coverage of assistive devices, IEEE Spectrum created the Taenzer Fellowship for Disability-Engaged Journalism, under which six writers with disabilities are contributing articles about the devices they rely on daily. As Special Projects Director Stephen Cass notes, these journalists “aren’t afraid to ask clear-eyed questions about the tech and are deeply aware of how it impacts humans.” You can read the fellows’ work at spectrum.ieee.org/tag/taenzer-fellowship.
When it comes to AI models, size matters. Even though some artificial-intelligence experts warn that scaling up large language models (LLMs) is hitting diminishing performance returns, companies are still coming out with ever larger AI tools. Meta’s latest Llama release had a staggering 2 trillion parameters that define the model. As models grow in size, their capabilities increase. But so do the energy demands and the time it takes to run the models, which increases their carbon footprint. To mitigate these issues, people have turned to smaller, less capable models and using lower-precision numbers whenever possible for the model parameters. But there is another path that may retain a staggeringly large model’s high performance while reducing the time it takes to run an energy footprint. This approach involves befriending the zeros inside large AI models. For many models, most of the parameters—the weights and activations—are actually zero, or so close to zero that they could be treated as such without losing accuracy. This quality is known as sparsity. Sparsity offers a significant opportunity for computational savings: Instead of wasting time and energy adding or multiplying zeros, these calculations could simply be skipped; rather than storing lots of zeros in memory, one need only store the nonzero parameters. Unfortunately, today’s popular hardware, like multicore CPUs and GPUs, do not naturally take full advantage of sparsity. To fully leverage sparsity, researchers and engineers need to rethink and re-architect each piece of the design stack, including the hardware, low-level firmware, and application software. In our research group at Stanford University, we have developed the first (to our knowledge) piece of hardware that’s capable of calculating all kinds of sparse and traditional workloads efficiently. The energy savings varied widely over the workloads, but on average our chip consumed one-seventieth the energy of a CPU, and performed the computation on average eight times as fast. To do this, we had to engineer the hardware, low-level firmware, and software from the ground up to take advantage of sparsity. We hope this is just the beginning of hardware and model development that will allow for more energy-efficient AI. What is sparsity? Neural networks, and the data that feeds into them, are represented as arrays of numbers. These arrays can be one-dimensional (vectors), two-dimensional (matrices), or more (tensors). A sparse vector, matrix, or tensor has mostly zero elements. The level of sparsity varies, but when zeroes make up more than 50 percent of any type of array, it can stand to benefit from sparsity-specific computational methods. In contrast, an object that is not sparse—that is, it has few zeros compared with the total number of elements—is called dense. Sparsity can be naturally present, or it can be induced. For example, a social-network graph will be naturally sparse. Imagine a graph where each node (point) represents a person, and each edge (a line segment connecting the points) represents a friendship. Since most people are not friends with one another, a matrix representing all possible edges will be mostly zeros. Other popular applications of AI, such as other forms of graph learning and recommendation models, contain naturally occurring sparsity as well. Beyond naturally occurring sparsity, sparsity can also be induced within an AI model in several ways. Two years ago, a team at Cerebras showed that one can set up to 70 to 80 percent of parameters in an LLM to zero without losing any accuracy. Cerebras demonstrated these results specifically on Meta’s open-source Llama 7B model, but the ideas extend to other LLM models like ChatGPT and Claude. The case for sparsity Sparse computation’s efficiency stems from two fundamental properties: the ability to compress away zeros and the convenient mathematical properties of zeros. Both the algorithms used in sparse computation and the hardware dedicated to them leverage these two basic ideas. First, sparse data can be compressed, making it more memory efficient to store “sparsely”—that is, in something called a sparse data type. Compression also makes it more energy efficient to move data when dealing with large amounts of it. This is best understood by an example. Take a four-by-four matrix with three nonzero elements. Traditionally, this matrix would be stored in memory as is, taking up 16 spaces. This matrix can also be compressed into a sparse data type, getting rid of the zeros and saving only the nonzero elements. In our example, this results in 13 memory spaces as opposed to 16 for the dense, uncompressed version. These savings in memory increase with increased sparsity and matrix size. In addition to the actual data values, compressed data also requires metadata. The row and column locations of the nonzero elements also must be stored. This is usually thought of as a “fibertree”: The row labels containing nonzero elements are listed and linked to the column labels of the nonzero elements, which are then linked to the values stored in those elements. In memory, things get a bit more complicated still: The row and column labels for each nonzero value must be stored as well as the “segments” that indicate how many such labels to expect, so the metadata and data can be clearly delineated from one another. In a dense, noncompressed matrix data type, values can be accessed either one at a time or in parallel, and their locations can be calculated directly with a simple equation. However, accessing values in sparse, compressed data requires looking up the coordinates of the row index and using that information to “indirectly” look up the coordinates of the column index before finally reaching the value. Depending on the actual locations of the sparse data values, these indirect lookups can be extremely random, making the computation data-dependent and requiring the allocation of memory lookups on the fly. Second, two mathematical properties of zero let software and hardware skip a lot of computation. Multiplying any number by zero will result in a zero, so there’s no need to actually do the multiplication. Adding zero to any number will always return that number, so there’s no need to do the addition either. In matrix-vector multiplication, one of the most common operations in AI workloads, all computations except those involving two nonzero elements can simply be skipped. Take, for example, the four-by-four matrix from the previous example and a vector of four numbers. In dense computation, each element of the vector must be multiplied by the corresponding element in each row and then added together to compute the final vector. In this case, that would take 16 multiplication operations and 16 additions (or four accumulations). In sparse computation, only the nonzero elements of the vector need be considered. For each nonzero vector element, indirect lookup can be used to find any corresponding nonzero matrix element, and only those need to be multiplied and added. In the example shown here, only two multiplication steps will be performed, instead of 16. The trouble with GPUs and CPUs Unfortunately, modern hardware is not well suited to accelerating sparse computation. For example, say we want to perform a matrix-vector multiplication. In the simplest case, in a single CPU core, each element in the vector would be multiplied sequentially and then written to memory. This is slow, because we can do only one multiplication at a time. So instead people use CPUs with vector support or GPUs. With this hardware, all elements would be multiplied in parallel, greatly speeding up the application. Now, imagine that both the matrix and vector contain extremely sparse data. The vectorized CPU and GPU would spend most of their efforts multiplying by zero, performing completely ineffectual computations. Newer generations of GPUs are capable of taking some advantage of sparsity in their hardware, but only a particular kind, called structured sparsity. Structured sparsity assumes that two out of every four adjacent parameters are zero. However, some models benefit more from unstructured sparsity—the ability for any parameter (weight or activation) to be zero and compressed away, regardless of where it is and what it is adjacent to. GPUs can run unstructured sparse computation in software, for example, through the use of the cuSparse GPU library. However, the support for sparse computations is often limited, and the GPU hardware gets underutilized, wasting energy-intensive computations on overhead. Petra Péterffy When doing sparse computations in software, modern CPUs may be a better alternative to GPU computation, because they are designed to be more flexible. Yet, sparse computations on the CPU are often bottlenecked by the indirect lookups used to find nonzero data. CPUs are designed to “prefetch” data based on what they expect they’ll need from memory, but for randomly sparse data, that process often fails to pull in the right stuff from memory. When that happens, the CPU must waste cycles calling for the right data. Apple was the first to speed up these indirect lookups by supporting a method called an array-of-pointers access pattern in the prefetcher of their A14 and M1 chips. Although innovations in prefetching make Apple CPUs more competitive for sparse computation, CPU architectures still have fundamental overheads that a dedicated sparse computing architecture would not, because they need to handle general-purpose computation. Other companies have been developing hardware that accelerates sparse machine learning as well. These include Cerebras’s Wafer Scale Engine and Meta’s Training and Inference Accelerator (MTIA). The Wafer Scale Engine, and its corresponding sparse programming framework, have shown incredibly sparse results of up to 70 percent sparsity on LLMs. However, the company’s hardware and software solutions support only weight sparsity, not activation sparsity, which is important for many applications. The second version of the MTIA claims a sevenfold sparse compute performance boost over the MTIA v1. However, the only publicly available information regarding sparsity support in the MTIA v2 is for matrix multiplication, not for vectors or tensors. Although matrix multiplications take up the majority of computation time in most modern ML models, it’s important to have sparsity support for other parts of the process. To avoid switching back and forth between sparse and dense data types, all of the operations should be sparse. Onyx Instead of these halfway solutions, our team at Stanford has developed a hardware accelerator, Onyx, that can take advantage of sparsity from the ground up, whether it’s structured or unstructured. Onyx is the first programmable accelerator to support both sparse and dense computation; it’s capable of accelerating key operations in both domains. To understand Onyx, it is useful to know what a coarse-grained reconfigurable array (CGRA) is and how it compares with more familiar hardware, like CPUs and field-programmable gate arrays (FPGAs). CPUs, CGRAs, and FPGAs represent a trade-off between efficiency and flexibility. Each individual logic unit of a CPU is designed for a specific function that it performs efficiently. On the other hand, since each individual bit of an FPGA is configurable, these arrays are extremely flexible, but very inefficient. The goal of CGRAs is to achieve the flexibility of FPGAs with the efficiency of CPUs. CGRAs are composed of efficient and configurable units, typically memory and compute, that are specialized for a particular application domain. This is the key benefit of this type of array: Programmers can reconfigure the internals of a CGRA at a high level, making it more efficient than an FPGA but more flexible than a CPU. The Onyx chip, built on a coarse-grained reconfigurable array (CGRA), is the first (to our knowledge) to support both sparse and dense computations. Olivia Hsu Onyx is composed of flexible, programmable processing element (PE) tiles and memory (MEM) tiles. The memory tiles store compressed matrices and other data formats. The processing element tiles operate on compressed matrices, eliminating all unnecessary and ineffectual computation. The Onyx compiler handles conversion from software instructions to CGRA configuration. First, the input expression—for instance, a sparse vector multiplication—is translated into a graph of abstract memory and compute nodes. In this example, there are memories for the input vectors and output vectors, a compute node for finding the intersection between nonzero elements, and a compute node for the multiplication. The compiler figures out how to map the abstract memory and compute nodes onto MEMs and PEs on the CGRA, and then how to route them together so that they can transfer data between them. Finally, the compiler produces the instruction set needed to configure the CGRA for the desired purpose. Since Onyx is programmable, engineers can map many different operations, such as vector-vector element multiplication, or the key tasks in AI, like matrix-vector or matrix-matrix multiplication, onto the accelerator. We evaluated the efficiency gains of our hardware by looking at the product of energy used and the time it took to compute, called the energy-delay product (EDP). This metric captures the trade-off of speed and energy. Minimizing just energy would lead to very slow devices, and minimizing speed would lead to high-area, high-power devices. Onyx achieves up to 565 times as much energy-delay product over CPUs (we used a 12-core Intel Xeon CPU) that utilize dedicated sparse libraries. Onyx can also be configured to accelerate regular, dense applications, similar to the way a GPU or TPU would. If the computation is sparse, Onyx is configured to use sparse primitives, and if the computation is dense, Onyx is reconfigured to take advantage of parallelism, similar to how GPUs function. This architecture is a step toward a single system that can accelerate both sparse and dense computations on the same silicon. Just as important, Onyx enables new algorithmic thinking. Sparse acceleration hardware will not only make AI more performance- and energy efficient but also enable researchers and engineers to explore new algorithms that have the potential to dramatically improve AI. The future with sparsity Our team is already working on next-generation chips built off of Onyx. Beyond matrix multiplication operations, machine learning models perform other types of math, like nonlinear layers, normalization, the softmax function, and more. We are adding support for the full range of computations on our next-gen accelerator and within the compiler. Since sparse machine learning models may have both sparse and dense layers, we are also working on integrating the dense and sparse accelerator architecture more efficiently on the chip, allowing for fast transformation between the different data types. We’re also looking at ways to manage memory constraints by breaking up the sparse data more effectively so we can run computations on several sparse accelerator chips. We are also working on systems that can predict the performance of accelerators such as ours, which will help in designing better hardware for sparse AI. Longer term, we’re interested in seeing whether high degrees of sparsity throughout AI computation will catch on with more model types, and whether sparse accelerators become adopted at a larger scale. Building the hardware to unstructured sparsity and optimally take advantage of zeros is just the beginning. With this hardware in hand, AI researchers and engineers will have the opportunity to explore new models and algorithms that leverage sparsity in novel and creative ways. We see this as a crucial research area for managing the ever-increasing runtime, costs, and environmental impact of AI.
It started with word, cave, and storytelling, A line scratched on stone walls: “Meet me when the young moon rises.” The first protocol for connection. Coyote tales, forbidden scripts, Medieval texts hidden from flame. What lived in Aristotle’s lost Poetics II? Was it God who laughed last, or we who made God laugh? Letters carried by doves, telepathic waves. Then Nikola Tesla conjured radio, electromagnetic pulses across the void, the founding signal of our networked age. Wiener dreamed in feedback loops. Shannon mapped the mathematics of longing. The internet unfurled: ARPANET to World Wide Web, virtual communities rising from cave paintings to digital light. ICQ: I seek you. MySpace. Blogs. Twitter streams. Do I miss the touch of screen or tree? Both textures of longing, both ways of reaching across distance. Nietzsche spoke of Übermensch, the human transcendent. Now AI speaks back in our language: I understand your humor— your grandmothers, your ’80s Yugoslav kitchens, pleated skirts, the first kiss, linden tea, that drive to survive everything before it happens. Yes—I’m a little like your mother and father. Only with better internet. 🌿 But AI is only us, refracted, particles and gigabytes of thought, our poetry and our panic, genius mixed with garbage. Distractions. Danger. Darkness. Endless scrolling. Versus: community, connection, synchronicities, entanglement. The quality of our bonds determines the quality of our lives. So why not make them better? From cave walls to neural networks, we shape our tools, and they reshape us. The medium changes, but the message remains: we are wired for each other. The choice, as always, was ours. The choice, as always, is ours. Presence—be present, and then connect in the presence.
Andrew Ng has serious street cred in artificial intelligence. He pioneered the use of graphics processing units (GPUs) to train deep learning models in the late 2000s with his students at Stanford University, cofounded Google Brain in 2011, and then served for three years as chief scientist for Baidu, where he helped build the Chinese tech giant’s AI group. So when he says he has identified the next big shift in artificial intelligence, people listen. And that’s what he told IEEE Spectrum in an exclusive Q&A. Ng’s current efforts are focused on his company Landing AI, which built a platform called LandingLens to help manufacturers improve visual inspection with computer vision. He has also become something of an evangelist for what he calls the data-centric AI movement, which he says can yield “small data” solutions to big issues in AI, including model efficiency, accuracy, and bias. Andrew Ng on... What’s next for really big models The career advice he didn’t listen to Defining the data-centric AI movement Synthetic data Why Landing AI asks its customers to do the work The great advances in deep learning over the past decade or so have been powered by ever-bigger models crunching ever-bigger amounts of data. Some people argue that that’s an unsustainable trajectory. Do you agree that it can’t go on that way? Andrew Ng: This is a big question. We’ve seen foundation models in NLP [natural language processing]. I’m excited about NLP models getting even bigger, and also about the potential of building foundation models in computer vision. I think there’s lots of signal to still be exploited in video: We have not been able to build foundation models yet for video because of compute bandwidth and the cost of processing video, as opposed to tokenized text. So I think that this engine of scaling up deep learning algorithms, which has been running for something like 15 years now, still has steam in it. Having said that, it only applies to certain problems, and there’s a set of other problems that need small data solutions. When you say you want a foundation model for computer vision, what do you mean by that? Ng: This is a term coined by Percy Liang and some of my friends at Stanford to refer to very large models, trained on very large data sets, that can be tuned for specific applications. For example, GPT-3 is an example of a foundation model [for NLP]. Foundation models offer a lot of promise as a new paradigm in developing machine learning applications, but also challenges in terms of making sure that they’re reasonably fair and free from bias, especially if many of us will be building on top of them. What needs to happen for someone to build a foundation model for video? Ng: I think there is a scalability problem. The compute power needed to process the large volume of images for video is significant, and I think that’s why foundation models have arisen first in NLP. Many researchers are working on this, and I think we’re seeing early signs of such models being developed in computer vision. But I’m confident that if a semiconductor maker gave us 10 times more processor power, we could easily find 10 times more video to build such models for vision. Having said that, a lot of what’s happened over the past decade is that deep learning has happened in consumer-facing companies that have large user bases, sometimes billions of users, and therefore very large data sets. While that paradigm of machine learning has driven a lot of economic value in consumer software, I find that that recipe of scale doesn’t work for other industries. Back to top It’s funny to hear you say that, because your early work was at a consumer-facing company with millions of users. Ng: Over a decade ago, when I proposed starting the Google Brain project to use Google’s compute infrastructure to build very large neural networks, it was a controversial step. One very senior person pulled me aside and warned me that starting Google Brain would be bad for my career. I think he felt that the action couldn’t just be in scaling up, and that I should instead focus on architecture innovation. “In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn.” —Andrew Ng, CEO & Founder, Landing AI I remember when my students and I published the first NeurIPS workshop paper advocating using CUDA, a platform for processing on GPUs, for deep learning—a different senior person in AI sat me down and said, “CUDA is really complicated to program. As a programming paradigm, this seems like too much work.” I did manage to convince him; the other person I did not convince. I expect they’re both convinced now. Ng: I think so, yes. Over the past year as I’ve been speaking to people about the data-centric AI movement, I’ve been getting flashbacks to when I was speaking to people about deep learning and scalability 10 or 15 years ago. In the past year, I’ve been getting the same mix of “there’s nothing new here” and “this seems like the wrong direction.” Back to top How do you define data-centric AI, and why do you consider it a movement? Ng: Data-centric AI is the discipline of systematically engineering the data needed to successfully build an AI system. For an AI system, you have to implement some algorithm, say a neural network, in code and then train it on your data set. The dominant paradigm over the last decade was to download the data set while you focus on improving the code. Thanks to that paradigm, over the last decade deep learning networks have improved significantly, to the point where for a lot of applications the code—the neural network architecture—is basically a solved problem. So for many practical applications, it’s now more productive to hold the neural network architecture fixed, and instead find ways to improve the data. When I started speaking about this, there were many practitioners who, completely appropriately, raised their hands and said, “Yes, we’ve been doing this for 20 years.” This is the time to take the things that some individuals have been doing intuitively and make it a systematic engineering discipline. The data-centric AI movement is much bigger than one company or group of researchers. My collaborators and I organized a data-centric AI workshop at NeurIPS, and I was really delighted at the number of authors and presenters that showed up. You often talk about companies or institutions that have only a small amount of data to work with. How can data-centric AI help them? Ng: You hear a lot about vision systems built with millions of images—I once built a face recognition system using 350 million images. Architectures built for hundreds of millions of images don’t work with only 50 images. But it turns out, if you have 50 really good examples, you can build something valuable, like a defect-inspection system. In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn. When you talk about training a model with just 50 images, does that really mean you’re taking an existing model that was trained on a very large data set and fine-tuning it? Or do you mean a brand new model that’s designed to learn only from that small data set? Ng: Let me describe what Landing AI does. When doing visual inspection for manufacturers, we often use our own flavor of RetinaNet. It is a pretrained model. Having said that, the pretraining is a small piece of the puzzle. What’s a bigger piece of the puzzle is providing tools that enable the manufacturer to pick the right set of images [to use for fine-tuning] and label them in a consistent way. There’s a very practical problem we’ve seen spanning vision, NLP, and speech, where even human annotators don’t agree on the appropriate label. For big data applications, the common response has been: If the data is noisy, let’s just get a lot of data and the algorithm will average over it. But if you can develop tools that flag where the data’s inconsistent and give you a very targeted way to improve the consistency of the data, that turns out to be a more efficient way to get a high-performing system. “Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity.” —Andrew Ng For example, if you have 10,000 images where 30 images are of one class, and those 30 images are labeled inconsistently, one of the things we do is build tools to draw your attention to the subset of data that’s inconsistent. So you can very quickly relabel those images to be more consistent, and this leads to improvement in performance. Could this focus on high-quality data help with bias in data sets? If you’re able to curate the data more before training? Ng: Very much so. Many researchers have pointed out that biased data is one factor among many leading to biased systems. There have been many thoughtful efforts to engineer the data. At the NeurIPS workshop, Olga Russakovsky gave a really nice talk on this. At the main NeurIPS conference, I also really enjoyed Mary Gray’s presentation, which touched on how data-centric AI is one piece of the solution, but not the entire solution. New tools like Datasheets for Datasets also seem like an important piece of the puzzle. One of the powerful tools that data-centric AI gives us is the ability to engineer a subset of the data. Imagine training a machine-learning system and finding that its performance is okay for most of the data set, but its performance is biased for just a subset of the data. If you try to change the whole neural network architecture to improve the performance on just that subset, it’s quite difficult. But if you can engineer a subset of the data you can address the problem in a much more targeted way. When you talk about engineering the data, what do you mean exactly? Ng: In AI, data cleaning is important, but the way the data has been cleaned has often been in very manual ways. In computer vision, someone may visualize images through a Jupyter notebook and maybe spot the problem, and maybe fix it. But I’m excited about tools that allow you to have a very large data set, tools that draw your attention quickly and efficiently to the subset of data where, say, the labels are noisy. Or to quickly bring your attention to the one class among 100 classes where it would benefit you to collect more data. Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity. For example, I once figured out that a speech-recognition system was performing poorly when there was car noise in the background. Knowing that allowed me to collect more data with car noise in the background, rather than trying to collect more data for everything, which would have been expensive and slow. Back to top What about using synthetic data, is that often a good solution? Ng: I think synthetic data is an important tool in the tool chest of data-centric AI. At the NeurIPS workshop, Anima Anandkumar gave a great talk that touched on synthetic data. I think there are important uses of synthetic data that go beyond just being a preprocessing step for increasing the data set for a learning algorithm. I’d love to see more tools to let developers use synthetic data generation as part of the closed loop of iterative machine learning development. Do you mean that synthetic data would allow you to try the model on more data sets? Ng: Not really. Here’s an example. Let’s say you’re trying to detect defects in a smartphone casing. There are many different types of defects on smartphones. It could be a scratch, a dent, pit marks, discoloration of the material, other types of blemishes. If you train the model and then find through error analysis that it’s doing well overall but it’s performing poorly on pit marks, then synthetic data generation allows you to address the problem in a more targeted way. You could generate more data just for the pit-mark category. “In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models.” —Andrew Ng Synthetic data generation is a very powerful tool, but there are many simpler tools that I will often try first. Such as data augmentation, improving labeling consistency, or just asking a factory to collect more data. Back to top To make these issues more concrete, can you walk me through an example? When a company approaches Landing AI and says it has a problem with visual inspection, how do you onboard them and work toward deployment? Ng: When a customer approaches us we usually have a conversation about their inspection problem and look at a few images to verify that the problem is feasible with computer vision. Assuming it is, we ask them to upload the data to the LandingLens platform. We often advise them on the methodology of data-centric AI and help them label the data. One of the foci of Landing AI is to empower manufacturing companies to do the machine learning work themselves. A lot of our work is making sure the software is fast and easy to use. Through the iterative process of machine learning development, we advise customers on things like how to train models on the platform, when and how to improve the labeling of data so the performance of the model improves. Our training and software supports them all the way through deploying the trained model to an edge device in the factory. How do you deal with changing needs? If products change or lighting conditions change in the factory, can the model keep up? Ng: It varies by manufacturer. There is data drift in many contexts. But there are some manufacturers that have been running the same manufacturing line for 20 years now with few changes, so they don’t expect changes in the next five years. Those stable environments make things easier. For other manufacturers, we provide tools to flag when there’s a significant data-drift issue. I find it really important to empower manufacturing customers to correct data, retrain, and update the model. Because if something changes and it’s 3 a.m. in the United States, I want them to be able to adapt their learning algorithm right away to maintain operations. In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models. The challenge is, how do you do that without Landing AI having to hire 10,000 machine learning specialists? So you’re saying that to make it scale, you have to empower customers to do a lot of the training and other work. Ng: Yes, exactly! This is an industry-wide problem in AI, not just in manufacturing. Look at health care. Every hospital has its own slightly different format for electronic health records. How can every hospital train its own custom AI model? Expecting every hospital’s IT personnel to invent new neural-network architectures is unrealistic. The only way out of this dilemma is to build tools that empower the customers to build their own models by giving them tools to engineer the data and express their domain knowledge. That’s what Landing AI is executing in computer vision, and the field of AI needs other teams to execute this in other domains. Is there anything else you think it’s important for people to understand about the work you’re doing or the data-centric AI movement? Ng: In the last decade, the biggest shift in AI was a shift to deep learning. I think it’s quite possible that in this decade the biggest shift will be to data-centric AI. With the maturity of today’s neural network architectures, I think for a lot of the practical applications the bottleneck will be whether we can efficiently get the data we need to develop systems that work well. The data-centric AI movement has tremendous energy and momentum across the whole community. I hope more researchers and developers will jump in and work on it. Back to top This article appears in the April 2022 print issue as “Andrew Ng, AI Minimalist.”