pureXML is the native XML storage feature in the IBM Db2 data server. pureXML provides query languages, storage technologies, indexing technologies, and other features to support XML data. The word pure in pureXML was chosen to indicate that Db2 natively stores and natively processes XML data in its inherent hierarchical structure, as opposed to treating XML data as plain text or converting it into a relational format. == Technical information == Db2 includes two distinct storage mechanisms: one for efficiently managing traditional SQL data types, and another for managing XML data. The underlying storage mechanism is transparent to users and applications; they simply use SQL (including SQL with XML extensions or SQL/XML) or XQuery to work with the data. XML data is stored in columns of Db2 tables that have the XML data type. XML data is stored in a parsed format that reflects the hierarchical nature of the original XML data. As such, pureXML uses trees and nodes as its model for storing and processing XML data. If you instruct Db2 to validate XML data against an XML schema prior to storage, Db2 annotates all nodes in the XML hierarchy with information about the schema types; otherwise, it will annotate the nodes with default type information. Upon storage, Db2 preserves the internal structure of XML data, converting its tag names and other information into integer values. Doing so helps conserve disk space and also improves the performance of queries that use navigational expressions. However, users aren't aware of this internal representation. Finally, Db2 automatically splits XML nodes across multiple database pages, as needed. XML schemas specify which XML elements are valid, in what order these elements should appear in XML data, which XML data types are associated with each element, and so on. pureXML allows you to validate the cells in a column of XML data against no schema, one schema, or multiple schemas. pureXML also provides tools to support evolving XML schemas. IBM has enhanced its programming language interfaces to support access to its XML data. These enhancements span Java (JDBC), C (embedded SQL and call-level interface), COBOL (embedded SQL), PHP, and Microsoft's .NET Framework (through the DB2.NET provider). == History == pureXML was first included in the DB2 9 for Linux, Unix, and Microsoft Windows release, which was codenamed Viper, in June 2006. It was available on DB2 9 for z/OS in March 2007. In October 2007, IBM released DB2 9.5 with improved XML data transaction performance and improved storage savings. In June 2009, IBM released DB2 9.7 with XML supported for database-partitioned, range-partitioned, and multi-dimensionally clustered tables as well as compression of XML data and indices. == Competition == Db2 is a hybrid data server—it offers data management for traditional relational data, as well as providing native XML data management. Other vendors that offer data management for both relational data and native XML storage include Oracle with its 11g product and Microsoft with its SQL Server product. pureXML also competes with native XML databases like BaseX, eXist, MarkLogic or Sedna. == Books == IBM International Technical Support Organization (ITSO) has published the following books, which are available in print or as free e-books: DB2 9: pureXML Overview and Fast Start DB2 9 pureXML Guide The following books are also available for purchase: DB2 pureXML Cookbook: Master the Power of IBM Hybrid Data Server == Education and training == The following pureXML classroom and online courses are available from IBM Education: Query and Manage XML Data with DB2 9. IBM course CG130. Classroom. Duration: 4 days. Query XML Data with DB2 9. IBM course CG100. Classroom. Duration: 2 days (first 2 days of CG130). Managing XML Data in DB2 9. IBM course CG160. Classroom. Duration: 2 days (last 2 days of CG130). DB2 pureXML. IBM Course CT140. Self-paced study plus Live Virtual Classroom.
Color quantization
In computer graphics, color quantization or color image quantization is quantization applied to color spaces; it is a process that reduces the number of distinct colors used in an image, usually with the intention that the new image should be as visually similar as possible to the original image. Computer algorithms to perform color quantization on bitmaps have been studied since the 1970s. Color quantization is critical for displaying images with many colors on devices that can only display a limited number of colors, usually due to memory limitations, and enables efficient compression of certain types of images. The name "color quantization" is primarily used in computer graphics research literature; in applications, terms such as optimized palette generation, optimal palette generation, or decreasing color depth are used. Some of these are misleading, as the palettes generated by standard algorithms are not necessarily the best possible. == Algorithms == Most standard techniques treat color quantization as a problem of clustering points in three-dimensional space, where the points represent colors found in the original image and the three axes represent the three color channels. Almost any three-dimensional clustering algorithm can be applied to color quantization, and vice versa. After the clusters are located, typically the points in each cluster are averaged to obtain the representative color that all colors in that cluster are mapped to. The three color channels are usually red, green, and blue, but another popular choice is the Lab color space, in which Euclidean distance is more consistent with perceptual difference. The most popular algorithm by far for color quantization, invented by Paul Heckbert in 1979, is the median cut algorithm. Many variations on this scheme are in use. Before this time, most color quantization was done using the population algorithm or population method, which essentially constructs a histogram of equal-sized ranges and assigns colors to the ranges containing the most points. A more modern popular method is clustering using octrees, first conceived by Gervautz and Purgathofer and improved by Xerox PARC researcher Dan Bloomberg. If the palette is fixed, as is often the case in real-time color quantization systems such as those used in operating systems, color quantization is usually done using the "straight-line distance" or "nearest color" algorithm, which simply takes each color in the original image and finds the closest palette entry, where distance is determined by the distance between the two corresponding points in three-dimensional space. In other words, if the colors are ( r 1 , g 1 , b 1 ) {\displaystyle (r_{1},g_{1},b_{1})} and ( r 2 , g 2 , b 2 ) {\displaystyle (r_{2},g_{2},b_{2})} , we want to minimize the Euclidean distance: ( r 1 − r 2 ) 2 + ( g 1 − g 2 ) 2 + ( b 1 − b 2 ) 2 . {\displaystyle {\sqrt {(r_{1}-r_{2})^{2}+(g_{1}-g_{2})^{2}+(b_{1}-b_{2})^{2}}}.} This effectively decomposes the color cube into a Voronoi diagram, where the palette entries are the points and a cell contains all colors mapping to a single palette entry. There are efficient algorithms from computational geometry for computing Voronoi diagrams and determining which region a given point falls in; in practice, indexed palettes are so small that these are usually overkill. Color quantization is frequently combined with dithering, which can eliminate unpleasant artifacts such as banding that appear when quantizing smooth gradients and give the appearance of a larger number of colors. Some modern schemes for color quantization attempt to combine palette selection with dithering in one stage, rather than perform them independently. A number of other much less frequently used methods have been invented that use entirely different approaches. The Local K-means algorithm, conceived by Oleg Verevka in 1995, is designed for use in windowing systems where a core set of "reserved colors" is fixed for use by the system and many images with different color schemes might be displayed simultaneously. It is a post-clustering scheme that makes an initial guess at the palette and then iteratively refines it. In the early days of color quantization, the k-means clustering algorithm was deemed unsuitable because of its high computational requirements and sensitivity to initialization. In 2011, M. Emre Celebi reinvestigated the performance of k-means as a color quantizer. He demonstrated that an efficient implementation of k-means outperforms a large number of color quantization methods. The high-quality but slow NeuQuant algorithm reduces images to 256 colors by training a Kohonen neural network "which self-organises through learning to match the distribution of colours in an input image. Taking the position in RGB-space of each neuron gives a high-quality colour map in which adjacent colours are similar." It is particularly advantageous for images with gradients. Finally, one of the newer methods is spatial color quantization, conceived by Puzicha, Held, Ketterer, Buhmann, and Fellner of the University of Bonn, which combines dithering with palette generation and a simplified model of human perception to produce visually impressive results even for very small numbers of colors. It does not treat palette selection strictly as a clustering problem, in that the colors of nearby pixels in the original image also affect the color of a pixel. See sample images. == History and applications == In the early days of PCs, it was common for video adapters to support only 2, 4, 16, or (eventually) 256 colors due to video memory limitations; they preferred to dedicate the video memory to having more pixels (higher resolution) rather than more colors. Color quantization helped to justify this tradeoff by making it possible to display many high color images in 16- and 256-color modes with limited visual degradation. Many operating systems automatically perform quantization and dithering when viewing high color images in a 256 color video mode, which was important when video devices limited to 256 color modes were dominant. Modern computers can now display millions of colors at once, far more than can be distinguished by the human eye, limiting this application primarily to mobile devices and legacy hardware. Nowadays, color quantization is mainly used in GIF and PNG images. GIF, for a long time the most popular lossless and animated bitmap format on the World Wide Web, only supports up to 256 colors, necessitating quantization for many images. Some early web browsers constrained images to use a specific palette known as the web colors, leading to severe degradation in quality compared to optimized palettes. PNG images support 24-bit color, but can often be made much smaller in filesize without much visual degradation by application of color quantization, since PNG files use fewer bits per pixel for palettized images. The infinite number of colors available through the lens of a camera is impossible to display on a computer screen; thus converting any photograph to a digital representation necessarily involves some quantization. Practically speaking, 24-bit color is sufficiently rich to represent almost all colors perceivable by humans with sufficiently small error as to be visually identical (if presented faithfully), within the available color space. However, the digitization of color, either in a camera detector or on a screen, necessarily limits the available color space. Consequently there are many colors that may be impossible to reproduce, regardless of how many bits are used to represent the color. For example, it is impossible in typical RGB color spaces (common on computer monitors) to reproduce the full range of green colors that the human eye is capable of perceiving. With the few colors available on early computers, different quantization algorithms produced very different-looking output images. As a result, a lot of time was spent on writing sophisticated algorithms to be more lifelike. === Quantization for image compression === Many image file formats support indexed color. A whole-image palette typically selects 256 "representative" colors for the entire image, where each pixel references any one of the colors in the palette, as in the GIF and PNG file formats. A block palette typically selects 2 or 4 colors for each block of 4x4 pixels, used in BTC, CCC, S2TC, and S3TC. === Editor support === Many bitmap graphics editors contain built-in support for color quantization, and will automatically perform it when converting an image with many colors to an image format with fewer colors. Most of these implementations allow the user to set exactly the number of desired colors. Examples of such support include: Photoshop's Mode→Indexed Color function supplies a number of quantization algorithms ranging from the fixed Windows system and Web palettes to the proprietary Local and Global algorithms for generating palettes suited to a particu
PagedAttention
PagedAttention is an attention algorithm for efficient serving of large language models (LLMs). It was introduced in 2023 by Woosuk Kwon and colleagues in the paper Efficient Memory Management for Large Language Model Serving with PagedAttention, alongside the vLLM serving engine. The method stores the key–value cache used during autoregressive decoding in fixed-size blocks that can be mapped to non-contiguous physical memory, borrowing ideas from virtual memory, paging, and operating system design. == Background == In transformer inference, the key–value cache grows with sequence length and the number of concurrent requests. Kwon et al. argued that earlier serving systems typically reserved contiguous cache regions in advance, which caused reserved space, internal fragmentation, and external fragmentation. In their experiments, the paper reported that the effective memory utilization of previous systems could fall as low as 20.4%. == Description == PagedAttention partitions the cache of each sequence into fixed-size KV blocks. A request's cache is represented as a sequence of logical blocks, while a block table maps those logical blocks to physical GPU-memory blocks. As a result, neighboring logical blocks do not need to be contiguous in physical memory, and new blocks can be allocated on demand as generation proceeds. The design also makes it easier to share cache state across related decoding paths. In vLLM, physical blocks can be reference-counted and shared among requests or branches, with block-granularity copy-on-write used when a shared block must be modified. The original paper applied this design to parallel sampling, beam search, and prompts with shared prefixes. == Mathematical formulation == For a query token i {\displaystyle i} in causal self-attention, the standard attention output can be written as a i j = exp ( q i ⊤ k j / d ) ∑ t = 1 i exp ( q i ⊤ k t / d ) , o i = ∑ j = 1 i a i j v j {\displaystyle a_{ij}={\frac {\exp(\mathbf {q} _{i}^{\top }\mathbf {k} _{j}/{\sqrt {d}})}{\sum _{t=1}^{i}\exp(\mathbf {q} _{i}^{\top }\mathbf {k} _{t}/{\sqrt {d}})}},\;\mathbf {o} _{i}=\sum _{j=1}^{i}a_{ij}\mathbf {v} _{j}} where q i {\displaystyle \mathbf {q} _{i}} , k j {\displaystyle \mathbf {k} _{j}} , and v j {\displaystyle \mathbf {v} _{j}} are the query, key, and value vectors, and d {\displaystyle d} is the attention dimension. If the cache is partitioned into blocks of size B {\displaystyle B} , the key and value blocks may be written as K j = ( k ( j − 1 ) B + 1 , … , k j B ) , V j = ( v ( j − 1 ) B + 1 , … , v j B ) {\displaystyle \mathbf {K} _{j}=(\mathbf {k} _{(j-1)B+1},\ldots ,\mathbf {k} _{jB}),\;\mathbf {V} _{j}=(\mathbf {v} _{(j-1)B+1},\ldots ,\mathbf {v} _{jB})} PagedAttention then performs the computation blockwise: A i j = exp ( q i ⊤ K j / d ) ∑ t = 1 ⌈ i / B ⌉ exp ( q i ⊤ K t / d ) , o i = ∑ j = 1 ⌈ i / B ⌉ V j A i j ⊤ {\displaystyle \mathbf {A} _{ij}={\frac {\exp(\mathbf {q} _{i}^{\top }\mathbf {K} _{j}/{\sqrt {d}})}{\sum _{t=1}^{\lceil i/B\rceil }\exp(\mathbf {q} _{i}^{\top }\mathbf {K} _{t}/{\sqrt {d}})}},\;\mathbf {o} _{i}=\sum _{j=1}^{\lceil i/B\rceil }\mathbf {V} _{j}\mathbf {A} _{ij}^{\top }} where A i j {\displaystyle \mathbf {A} _{ij}} is the vector of attention scores for the j {\displaystyle j} -th KV block. In the formulation given by Kwon et al., this preserves the causal attention calculation while allowing the key and value blocks to reside in non-contiguous physical memory. == Performance and use == The vLLM paper reported that, on its evaluated workloads, the use of PagedAttention and the associated memory-management design improved serving throughput by 2–4× over the compared baselines, including FasterTransformer and Orca, while preserving model outputs. In experiments on OPT-13B with the Alpaca trace, the paper also reported memory savings of 6.1–9.8% for parallel sampling and 37.6–55.2% for beam search through KV-block sharing. A 2024 survey of LLM serving systems described PagedAttention as having become an industry norm in LLM serving frameworks, citing support in TGI, vLLM, and TensorRT-LLM. == Limitations and alternatives == Subsequent work has described trade-offs in the approach. The 2025 vAttention paper argued that PagedAttention requires attention kernels to be rewritten to support paging and increases software complexity, portability issues, redundancy, and execution overhead, proposing instead a memory manager that keeps the cache contiguous in virtual memory while relying on demand paging for physical allocation. === vAttention === Unlike PagedAttention, vAttention does not introduce a different attention rule; it retains the standard attention computation Attention ( q i , K , V ) = softmax ( q i K ⊤ s c a l e ) V . {\displaystyle \operatorname {Attention} (q_{i},K,V)=\operatorname {softmax} \left({\frac {q_{i}K^{\top }}{\mathrm {scale} }}\right)V.} In the notation of Prabhu et al., the key and value tensors for a request seen so far are K , V ∈ R L ′ × ( H × D ) {\displaystyle K,V\in \mathbb {R} ^{L'\times (H\times D)}} , where L ′ {\displaystyle L'} is the context length seen so far, H {\displaystyle H} is the number of KV heads on a worker, and D {\displaystyle D} is the dimension of each KV head. In systems prior to PagedAttention, the K cache (or V cache) at each layer of a worker is typically allocated as a 4D tensor of shape [ B , L , H , D ] , {\displaystyle [B,L,H,D],} where B {\displaystyle B} is batch size and L {\displaystyle L} is the maximum context length supported by the model. vAttention preserves this contiguous virtual-memory view while deferring physical-memory allocation to runtime. A serving framework maintains separate K and V tensors for each layer, so vAttention reserves 2 N {\displaystyle 2N} virtual-memory buffers on a worker, where N {\displaystyle N} is the number of layers managed by that worker. The maximum size of one virtual-memory buffer is B S = B × S , {\displaystyle BS=B\times S,} where S {\displaystyle S} is the maximum size of a single request's per-layer K cache (or V cache) on a worker. The paper defines S = L × H × D × P , {\displaystyle S=L\times H\times D\times P,} where P {\displaystyle P} is the number of bytes needed to store one element. In this formulation, vAttention keeps the KV cache contiguous in virtual memory and relies on demand paging for physical allocation, rather than modifying the attention kernel to operate over non-contiguous KV-cache blocks.
Compute (machine learning)
In machine learning and deep learning, compute is the amount of computing power or computational resources required to train machine learning models and large language models. More broadly, compute is the computational power or resources necessary for a computer or computer program to function. == Definition == Compute is commonly defined as the amount of computing power or computational resources required to train machine learning and large language models. The term "compute" has also been more broadly applied to cloud computing, referencing processing power, memory, networking, storage, and other resources required for the computation of any program. Compute is measured in petaflop/s-days and is used to document AI training. A petaflop/s-day (pfs-day) consists of performing 1015 neural net operations per second for one day, or a total of about 1020 operations. The compute-time product serves as a mental convenience, similar to kilowatt-hour for energy. An amount of compute is meant to give an idea of the number of actual operations performed. == History == In a 2018 analysis titled "AI and compute", artificial intelligence company OpenAI introduced the concept of compute. OpenAI identified two eras of training AI systems in terms of compute-usage. From 1959 to 2012, compute roughly followed Moore’s law. Between 2012 and 2018, the amount of compute used in the largest AI training runs increased exponentially, growing by more than 300,000 times — roughly doubling every 3.4 months. By comparison, Moore’s Law doubled every two years over the same period. One of the largest models, released in 2020, used 600,000 times more computing power than the 2012 model. After 2020, compute growth began to slow down, with the compute needed for the largest AI models continuing to slow down in 2023. The notion of compute has become increasingly used from the mid-2020s onwards. == Compute growth and AI progress == Larger AI models trained on more data and using more computational resources, tend to perform better. This happens even if the algorithms themselves remain unchanged. As early as 2018, OpenAI noted the exponential increase in compute to be have a key role in AI progress. OpenAI considers three factors drive the advance of AI: algorithmic innovation, data, and the amount of compute available for training. AI models with more compute not only improve in the tasks they were trained on but can develop emergent abilities. Incremental improvements can lead to more abrupt leaps in capabilities. AI provider SpaceXAI said in 2026 that their AI progress is driven by compute and used it a key metric in the AI training of its supercomputer Colossus, the which contains 1 million GPUs. Anthropic has a contract of $1.25 billion per month with SpaceXAI to buy all the compute capacity at Colossus 1 data center. === Criticism and policy === Increasing, promoting or constraining progress in artificial intelligence has often be done via controlling the amount of compute. Policymarkers have enacted policies and provided support to make compute resources more accessible to domestic AI researchers. In a January 2022 report, the Center for Security and Emerging Technology (CSET) suggested to institutions that increasingly powerful and generalizable AI (AGI) will likely require other strategies than maximizing compute. Some AI researchers are also concerned that government might exclusively focus on scaling compute instead of other strategies. The CSET has reported on the various bottlenecks which could explain why deep learning needs for compute have slow down: training is expensive and training extremely large models generates traffic jams across many processors that are difficult to manage. there is a limited supply of AI chips (see AI chip memory shortage). CSET advances that the main resource is human capital, specifically talented researchers — according to a 2023 published survey of more than 400 AI researchers, academic and private sector workers. The survey found that AI researchers are not primarily or exclusively constrained by compute access. However, both academic and industry AI researchers equally report concerns that insufficient compute could prevent them from contributing meaningfully to AI research in the future. High compute users are more concerned about compute access. When asked about which resource provided by the government would be the most useful to them, some AI researchers select compute, other prefer grant funding. For this goal, CSET advised policymakers to ensure that even researchers with smaller budgets could effectively contribute to AI research. Other proposed strategies include using contemporary AI algorithms, managing modern AI infrastructure or focusing on interdisciplinary work between the AI field and other fields of computer science. A 2024 study on compute access found that academic-only AI research teams often have less compute intensive research topics, especially foundation models, compared to industry AI labs. As a consequence, academia is likely to play a smaller role in advancing such techniques. The researchers suggest nationally-sponsored computing infrastructure as well as open science initiatives to boost academic compute access. === Data === A 2022 study found that current large language models are significantly under-trained, a consequence of focusing on scaling language models whilst keeping the amount of training data constant. By training over 400 language models of various parameter and token size, they found that "for compute-optimal training", the model size and the number of training tokens should ideally be scaled equally: for every doubling of model size the number of training tokens should also be doubled.
Isotropic position
In the fields of machine learning, the theory of computation, and random matrix theory, a probability distribution over vectors is said to be in isotropic position if its covariance matrix is proportional to the identity matrix. == Formal definitions == Let D {\textstyle D} be a distribution over vectors in the vector space R n {\textstyle \mathbb {R} ^{n}} . Then D {\textstyle D} is in isotropic position if, for vector v {\textstyle v} sampled from the distribution, E v v T = I d . {\displaystyle \mathbb {E} \,vv^{\mathsf {T}}=\mathrm {Id} .} A set of vectors is said to be in isotropic position if the uniform distribution over that set is in isotropic position. In particular, every orthonormal set of vectors is isotropic. As a related definition, a convex body K {\textstyle K} in R n {\textstyle \mathbb {R} ^{n}} is called isotropic if it has volume | K | = 1 {\textstyle |K|=1} , center of mass at the origin, and there is a constant α > 0 {\textstyle \alpha >0} such that ∫ K ⟨ x , y ⟩ 2 d x = α 2 | y | 2 , {\displaystyle \int _{K}\langle x,y\rangle ^{2}dx=\alpha ^{2}|y|^{2},} for all vectors y {\textstyle y} in R n {\textstyle \mathbb {R} ^{n}} ; here | ⋅ | {\textstyle |\cdot |} stands for the standard Euclidean norm.
Visual temporal attention
Visual temporal attention is a special case of visual attention that involves directing attention to specific instant of time. Similar to its spatial counterpart visual spatial attention, these attention modules have been widely implemented in video analytics in computer vision to provide enhanced performance and human interpretable explanation of deep learning models. As visual spatial attention mechanism allows human and/or computer vision systems to focus more on semantically more substantial regions in space, visual temporal attention modules enable machine learning algorithms to emphasize more on critical video frames in video analytics tasks, such as human action recognition. In convolutional neural network-based systems, the prioritization introduced by the attention mechanism is regularly implemented as a linear weighting layer with parameters determined by labeled training data. == Application in Action Recognition == Recent video segmentation algorithms often exploits both spatial and temporal attention mechanisms. Research in human action recognition has accelerated significantly since the introduction of powerful tools such as Convolutional Neural Networks (CNNs). However, effective methods for incorporation of temporal information into CNNs are still being actively explored. Motivated by the popular recurrent attention models in natural language processing, the Attention-aware Temporal Weighted CNN (ATW CNN) is proposed in videos, which embeds a visual attention model into a temporal weighted multi-stream CNN. This attention model is implemented as temporal weighting and it effectively boosts the recognition performance of video representations. Besides, each stream in the proposed ATW CNN framework is capable of end-to-end training, with both network parameters and temporal weights optimized by stochastic gradient descent (SGD) with back-propagation. Experimental results show that the ATW CNN attention mechanism contributes substantially to the performance gains with the more discriminative snippets by focusing on more relevant video segments. == Literature == Seibold VC, Balke J and Rolke B (2023): Temporal attention. Front. Cognit. 2:1168320. doi: 10.3389/fcogn.2023.1168320.
Artificial intelligence in education
Artificial intelligence in education (often abbreviated as AIEd) is a subfield of educational technology that studies how to use artificial intelligence to create learning environments. Considerations in the field include data-driven decision-making, AI ethics, data privacy and AI literacy. Concerns include the potential for cheating, over-reliance, equity of access, reduced critical thinking, and the perpetuation of misinformation and bias. == History == Efforts to integrate AI into educational contexts have often followed technological advancement in the history of artificial intelligence. In the 1960s, educators and researchers began developing computer-based instruction systems, such as PLATO, developed by the University of Illinois. In the 1970s and 1980s, intelligent tutoring systems (ITS) were being adapted for classroom instruction. The International Artificial Intelligence in Education Society was founded in 1993. Coinciding with the AI boom of the 2020s, the use of large language models in the global north has been promoted and funded by venture capital and big tech. Companies creating AI services have targeted students and educational institutions as customers. Similarly, pre-AI boom educational companies have expanded their use of AI technologies. These commercial incentives for AIEd use may be related to a potential AI bubble. In the U.S., bipartisan support of AI development in K-12 education has been expressed, but specific implementations and best practices remain contentious. == Theory == AIEd applies theory from education studies, machine learning, and related fields. A 2019 review of the previous decade of studies found that most research prioritized technological design over pedagogical integration. Ouyang and Jiao (2021) propose three paradigms for AI in education, which follow roughly from least to most learner-centered and from requiring least to most technical complexity from the AI systems: AI-directed, learner-as-recipient: AIEd systems present a pre-set curriculum based on statistical patterns that do not adjust to learner's feedback. AI-supported, learner-as-collaborator: Systems that incorporate responsiveness to learner's feedback through, for example, natural language processing, wherein AI can support knowledge construction. AI-empowered, learner-as-leader: This model seeks to position AI as a supplement to human intelligence wherein learners take agency and AI provides consistent and actionable feedback. Some scholars place AI in education within a socio-technical framework. This positions AI alongside other emerging educational technologies, such as computing, the internet, and social media. The framework of Tsao, Heinrichs and Camit (2025) draws on new materialism and posthumanism, specifically Donna Haraway's concept of sympoiesis (making-with). This perspective views learning as an entanglement of human and non-human actors (students, teachers, and AI algorithms), where knowledge is co-composed in contact zones between human context and algorithmic prediction. AI agents have been trained on biased datasets, and thus continue to perpetuate societal biases. Since LLMs were created to produce human-like text, algorithmic bias can be introduced and reproduced. AI's data processing and monitoring reinforce neoliberal approaches to education rather than addressing inequalities. == Applications == Uses of generative AI chatbots in education have included assessment and feedback, machine translations, proof-reading exam question generation and copy editing, or as virtual assistants. Emotional AI in education is the study and development of systems that can detect learners' emotions or provide emotional support in learning. == Usage == === Schools and educators === Following the release of ChatGPT in November 2022, some schools and large school districts blocked access to the site and issued warnings that the use of such tools would be seen as cheating. Governmental and non-governmental organizations such as UNESCO, Article 4 of the European Union's AI Act, and the U.S. Department of Education have published reports advocating for specific AIEd approaches. National higher-education bodies have also published guidance on generative AI, including Ireland's Higher Education Authority, which issued a policy framework for higher education teaching and learning in December 2025. In 2024, UNESCO released updated global guidance for generative AI in education, emphasizing ethical use, teacher training, and data protection to ensure responsible integration of AI tools in learning environments. According to Taso (2025), policy implementation in higher education is interpreted and enacted differently by various organizations. These decentralized policies can lead to inconsistent enforcement and confusion among students regarding what constitutes acceptable use, with the burden of ethical navigation falling on individual teachers and students. AI integration in classrooms has created new forms of invisible labour for educators, who must navigate ambiguous policies, redesign assessments to be AI-resilient, and adjudicate potential academic integrity violations. The use of AI detection tools has also been criticised for creating an adversarial relationship between students and institutions, where students may be falsely accused of misconduct based on probabilistic software. AIEd advocates say that efforts should be made towards increasing global accessibility and training educators to serve underprivileged areas. === Students === Reliance on generative AI has been linked with reduced academic self-esteem and performance, and heightened learned helplessness. Algorithm errors and hallucinations are common flaws in AI agents, making them less trustworthy and reliable. According to a 2025 survey from Inside Higher Ed, 85% of higher education students use generative AI technology in some way, with 25% using AI to complete assignments for them. The most common reason cited for using AI to cheat was pressure to get high grades. 97% of students wanted some form of action from schools on the threat to academic integrity caused by AI, with the most popular options being clearer policies and more education about ethical uses of AI. In September 2025, The Atlantic published an op-ed from a high school senior arguing that the normalization of AI cheating was eroding critical thinking, academic integrity, creativity, and the shared student experience.