AI Content Generation Tools

AI Content Generation Tools — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

  • Gapo

    Gapo

    Gapo is a Vietnamese social networking service based in Hanoi, Vietnam. Users are able to create a personal profile and share text, photos and videos with others on the platform. Users can also use Gapo for live streaming, instant messaging, blogging, and online payments. Gapo was launched in July 2019 by Hà Trung Kiên and Duong Vi Khoa. == History == Gapo was founded in response to calls for Vietnam's Communist-led government to produce a domestic alternative to social media giants like Facebook and Google. Gapo officially launched on July 23, 2019 at an event in Hanoi. The company received 500 billion đồng (US$22 million) in funding from technology corporation G-Group to be utilized in the first phase of development. They also partnered with Sony Music Entertainment to provide music content to its services. == Features == Gapo features a news feed for posting content, livestreaming, instant messaging, and blogging. It also allows users to pay online and access public services. == Reception == Within two days of launch, Gapo received about 200,000 registrations. By September 2019, the user base increased to one million. Upon launch, Gapo experienced significant technical difficulties. Users complained about the inability to sign up for a new account and said that certain functions were not available for use at launch. This issue caused Gapo to temporarily suspend their services in order to perform upgrades and bug fixes. Gapo relaunched the next day, though many users reported that the access speed decreased. The mobile app also received mixed reviews from users in both the App Store and the Google Play Store, with an average rating of 3.1 and 3.5, respectively. Most users found the app to be a knockoff of Facebook, although some users praised the app for being locally developed. === Expert opinions on platform viability === Le Hong Hiep of the ISEAS - Yusof Ishak Institute was doubtful that a Vietnamese-owned social network service could be as powerful as a foreign-based service, stating that Vietnam might not be able to develop a viable social media network to compete with the likes of Facebook or Google. Others, like blogger Ann Chi, said that, due to local players complying with local censorship policy, there is a chance that locals might not trust Gapo and other local services in light of possible surveillance. Regarding the targeted user base figure for the end of 2019 and 2021, experts cautioned that the company might need an additional trillion đồng of funding to reach its planned user base targets. In response, the company stated that Gapo was never meant to compete with Facebook, but instead noted that the main difference between Gapo and Facebook is that Gapo provides a personalized user experience through customization. == Censorship == Gapo has the right to censor posts and news that are deemed offensive and inaccurate by users or not approved by the censorship curators.

    Read more →
  • Danilo McGarry

    Danilo McGarry

    Danilo McGarry (born 1985) is a British tech executive, writer, and speaker who has led AI initiatives in finance and healthcare. == Early life and education == Danilo McGarry was born in 1985. He received a Bachelor of Science (BSc) with honors in Business Management from the University of Bath. == Career == McGarry began his career in technology and financial services, with positions at companies including Motorola, JPMorgan Chase, and BNP Paribas. He later joined the Royal Bank of Canada (RBC) as an analyst and later became a director, where he led transformation initiatives involving robotic process automation (RPA) in the bank's capital markets operations. McGarry subsequently moved into leadership roles focused on AI. At Citigroup, he served as Head of Artificial Intelligence and Machine Learning, where he launched an AI-driven robotics and automation initiative. At UnitedHealth Group (UHG), he held a senior role in the company's automation program, which utilized a large fleet of software robots in its healthcare operations. In December 2019, McGarry was appointed Global Head of AI & Automation at Alter Domus, a multinational financial services firm. In this role, he established a new AI and automation department. He left the firm in late 2023 to establish his businesses. In 2025, the Chartered Institute of Personnel and Development (CIPD) appointed him as its strategic adviser on artificial intelligence.

    Read more →
  • ACROSS Project

    ACROSS Project

    ACROSS is a Singular Strategic R&D Project led by Treelogic funded by the Spanish Ministry of Industry, Tourism and Trade activities in the field of Robotics and Cognitive Computing over an execution time-frame from 2009 to 2011. ACROSS project involves a number higher than 100 researchers from 13 Spanish entities. == ACROSS project objectives == ACROSS modifies the design of social robotics, blocked in providing predefined services, going further by means of intelligent systems. These systems are able to self-reconfigure and modify their behavior autonomously through the capacity for understanding, learning and software remote access. In order to provide an open framework for collaboration between universities, research centers and the Administration, ACROSS develops Open Source Services available to everybody. == Three application domains == ACROSS works in three application domains: Autonomous living: robots are used as technological tools to help handicapped person into daily tasks. Psycho-Affective Disorders (autism): robots are used to mitigate cognitive disorders. Marketing: robots are used to interact with humans in a recreational approach. == Consortium == Treelogic Alimerka Bizintek Universitat Politécnica de Catalunya University of Deusto European Centre for Soft Computing Fatronik - Tecnalia Fundació Hospital Comarcal Sant Antoni Abat Fundación Pública Andaluza para la Gestión de la Investigación en Salud de Sevilla, "Virgen del Rocío" University Hospitals m-BOT Omicron Electronic Universidad de Extremadura - RoboLab Verbio Technologies

    Read more →
  • Jensen Huang

    Jensen Huang

    Jen-Hsun "Jensen" Huang (Chinese: 黃仁勳; Wade–Giles: Huáng Jén-hsūn; Tâi-lô: N̂g Jîn-hun; born February 17, 1963) is a Taiwanese and American business executive and electrical engineer who is the founder, president, and CEO of Nvidia, the world's most valuable company. As of 2026, Forbes estimates his net worth at over US$200 billion, making him the seventh-wealthiest individual in the world. The son of Taiwanese immigrants, Huang spent his childhood in Taiwan and Thailand before moving to the United States, where he was a student in Kentucky and Oregon. After earning a master's degree from Stanford University, Huang launched Nvidia in 1993 from a Denny's restaurant in San Jose, California, at age 30 and has remained its president and CEO ever since. He led the company out of near-bankruptcy during the 1990s and oversaw its expansion into GPU production, high-performance computing, and artificial intelligence (AI). Under Huang, Nvidia experienced rapid growth during the AI boom, becoming the first company to reach a market capitalization of over $5 trillion in October 2025. In 2021 and 2024, Time magazine included Huang in their list of the most influential people. In 2025, he was named as one of the "Architects of AI" for Time's Person of the Year. == Early life and education == Huang was born in Taipei, Taiwan, on February 17, 1963, and moved to the southern city of Tainan as a child. He is the younger of two sons of Huang Hsing-tai, a chemical engineer at an oil refinery, and Lo Tsai-hsiu, a schoolteacher. They were a middle-class Taiwanese family that relocated often, and were native speakers of Taiwanese Hokkien. Each day, Jensen's mother randomly selected 10 words from the dictionary to teach her sons English. When he was five years old, Huang's family moved to Thailand to support his father's refinery career and remained there for approximately four years. He attended Ruamrudee International School while in Bangkok. In the late 1960s, Hsing-tai traveled from Taiwan to New York City to train under an air conditioning company and, after returning home, resolved to send his sons to the United States. At age nine, Jensen, despite not yet being able to speak English fluently, was sent by his parents to live in the United States. He and his older brother moved in 1973 to live with an uncle in Tacoma, Washington, escaping widespread social unrest in Thailand. Both Huang's aunt and uncle were recent immigrants to Washington state; they accidentally enrolled him and his brother in the Oneida Baptist Institute, a religious reform academy in Kentucky for troubled youth, mistakenly believing it to be a prestigious boarding school. In order to afford the academy's tuition, Jensen's parents sold nearly all their possessions. When he was 10 years old, Huang lived with his older brother in the Oneida boys' dormitory. Each student was expected to work every day, and his brother was assigned to perform manual labor on a nearby tobacco farm. Because he was too young to attend classes at the reform academy, Huang was educated at a separate public school—the Oneida Elementary school in Oneida, Kentucky—arriving as "an undersized Asian immigrant with long hair and heavily accented English" and was frequently bullied and beaten. In Oneida, Huang cleaned toilets every day, learned to play table-tennis, joined the swimming team, and appeared in Sports Illustrated at age 14. He taught his illiterate roommate, a "17-year-old covered in tattoos and knife scars," how to read in exchange for being taught how to bench press. In 2002, Huang said he remembered his life in Kentucky "more vividly than just about any other". Two years after Huang arrived in Oneida, his parents moved to the United States and settled in Beaverton, Oregon, after which the brothers withdrew from school in Kentucky to live back with them. As a teenager, Huang attended Aloha High School in Aloha, Oregon, where he excelled academically. He skipped two grades, graduated at age 16, and became a nationally ranked table-tennis player in addition to being a member of its mathematics, computer, and science clubs. In 1977, the school purchased an Apple II computer. Huang used the machine to play Super Star Trek, a text-based game, and to program in BASIC, creating his own version of Snake. Beginning at age 15, Huang got his first job working the graveyard shift at a local Denny's restaurant as a dishwasher, busboy, and waiter from 1978 to 1983. After high school, he chose to enroll at Oregon State University due to its low in-state tuition. He studied electrical engineering and graduated in 1984 with a bachelor's degree with highest honors. Huang later recalled, "I was the youngest kid in school, in class" and the only student who "looked like a child". Years later, while working as a microchip designer in Silicon Valley, he concurrently pursued graduate night classes at Stanford University, where he earned a master's degree in electrical engineering in 1992. == AMD and LSI Logic == After graduating from college, Huang was a microchip designer in Silicon Valley. He was recruited for positions at Texas Instruments, Advanced Micro Devices (AMD), and LSI Logic, ultimately choosing the California-based AMD due to already being familiar with the company. Huang designed AMD microprocessors while simultaneously attending Stanford and raising his two children. However, when he heard of new chip design processes at LSI Logic, Huang left AMD to assume a role as a technical officer at the LSI Corporation, working under a startup company, Sun Microsystems, where he met engineers Chris Malachowsky and Curtis Priem. LSI was in contract with Sun Microsystems and had introduced Huang to Malachowsky and Priem, who were working on a new graphics accelerator card. While the three produced the card's manufacturing process, the relationship between Malachowsky and Priem became strained as the two disputed the chip's design, leading to infighting; according to Malachowsky, they "broke every tool that LSI Logic had in their standard portfolio". In 1989, Huang, Malachowsky, and Priem finalized the accelerator, which they called the "GX graphics engine". GX was a widespread financial success; the sales of the graphics engine contributed to Sun Microsystem's revenue increasing from $262 million in 1987 to $656 million in 1990, and Huang was promoted to be the director of LSI's CoreWare, a division that manufactured chips for hardware vendors. == Nvidia == === Founding (1993) === When business began to slow for Sun Microsystems after 1990, Huang, along with Priem and Malachowsky, each resigned their jobs to pursue a venture together in making graphics chips for PC games. They initially named their new company "NVision" until Huang suggested that the company be named "Nvidia" based on the Latin word invidia, as Priem wanted competitors to turn "green with envy". They eventually dropped the "i" to honor the NV1 chip that they were then developing. The three met frequently in 1992 at a Denny's roadside diner in East San Jose to formulate a business plan. Huang chose for them to meet at Denny's due to his prior work experience at the restaurant chain and because it was "quieter than home and had cheap coffee". The three founded the company during one meeting at a breakfast booth at the diner. To formally incorporate the company, Huang found a lawyer, James Gaither of Cooley Godward, who demanded the $200 in cash in Huang's pockets to capitalize the company. After that meeting, Huang went back to Priem and Malachowsky to ask each of them for $200 for their respective shares of the company, which meant that Nvidia's initial capital was $600. On April 5, 1993, Huang personally signed Nvidia's original articles of incorporation into effect. Although he left LSI, Huang remained in good standing with the company and was able to secure funding for Nvidia from LSI's CEO, Wilfred Corrigan, who introduced Huang to venture capitalist Don Valentine. An account cited how Huang's presentation pitch went badly. Valentine, the leader of Sequoia Capital, chose to invest in Nvidia through Corrigan's support, as did Sutter Hill Ventures. The funding enabled Nvidia to begin development efforts toward its first chip and to begin paying wages for its employees. By the first day of operation, Huang was made Nvidia's president and CEO. Even though Huang, at age 30, was younger than Priem and Malachowsky, both Priem and Malachowsky believed that he was prepared to be CEO. According to Priem, "we basically deferred to Jensen on day one" and told Huang, "you're in charge of running the company—all the stuff Chris and I don't know how to do". === President and CEO (1993–present) === As of 2024, Huang has been Nvidia's chief executive for over three decades, a tenure described by The Wall Street Journal as "almost unheard of in fast-moving Silicon Valley". He owns 3.6% of Nvidia's stock, which went public in 1999. He earned US$24.6 million as CEO i

    Read more →
  • Glossary of machine vision

    Glossary of machine vision

    The following are common definitions related to the machine vision field. General related fields Machine vision Computer vision Image processing Signal processing == 0-9 == 1394. FireWire is Apple Inc.'s brand name for the IEEE 1394 interface. It is also known as i.Link (Sony's name) or IEEE 1394 (although the 1394 standard also defines a backplane interface). It is a personal computer (and digital audio/digital video) serial bus interface standard, offering high-speed communications and isochronous real-time data services. 1D. One-dimensional. 2D computer graphics. The computer-based generation of digital images—mostly from two-dimensional models (such as 2D geometric models, text, and digital images) and by techniques specific to them. 3D computer graphics. 3D computer graphics are different from 2D computer graphics in that a three-dimensional representation of geometric data is stored in the computer for the purposes of performing calculations and rendering 2D images. Such images may be for later display or for real-time viewing. Despite these differences, 3D computer graphics rely on many of the same algorithms as 2D computer vector graphics in the wire frame model and 2D computer raster graphics in the final rendered display. In computer graphics software, the distinction between 2D and 3D is occasionally blurred; 2D applications may use 3D techniques to achieve effects such as lighting, and primarily 3D may use 2D rendering techniques. 3D scanner. This is a device that analyzes a real-world object or environment to collect data on its shape and possibly color. The collected data can then be used to construct digital, three dimensional models useful for a wide variety of applications. == A == Aberration. Optically, defocus refers to a translation along the optical axis away from the plane or surface of best focus. In general, defocus reduces the sharpness and contrast of the image. What should be sharp, high-contrast edges in a scene become gradual transitions. Algebraic distance or algebraic error. The algebraic distance from a point xi to a curve or surface defined by f ( x , β ) = 0 {\displaystyle f(x,\beta )=0} is the value of f ( x i , β ) {\displaystyle f(x_{i},\beta )} , i.e. the residual in the least squares problem with data point (xi, 0) and model function f. This term is mainly used in computer vision.[1][2] Aperture. In context of photography or machine vision, aperture refers to the diameter of the aperture stop of a photographic lens. The aperture stop can be adjusted to control the amount of light reaching the film or image sensor. aspect ratio (image). The aspect ratio of an image is its displayed width divided by its height (usually expressed as "x:y"). Angular resolution. Describes the resolving power of any image forming device such as an optical or radio telescope, a microscope, a camera, or an eye. Automated optical inspection. == B == Barcode. A barcode (also bar code) is a machine-readable representation of information in a visual format on a surface. Blob discovery. Inspecting an image for discrete blobs of connected pixels (e.g. a black hole in a grey object) as image landmarks. These blobs frequently represent optical targets for machining, robotic capture, or manufacturing failure. Bitmap. A raster graphics image, digital image, or bitmap, is a data file or structure representing a generally rectangular grid of pixels, or points of color, on a computer monitor, paper, or other display device. == C == Camera. A camera is a device used to take pictures, either singly or in sequence. A camera that takes pictures singly is sometimes called a photo camera to distinguish it from a video camera. Camera Link. Camera Link is a serial communication protocol designed for computer vision applications based on the National Semiconductor interface Channel-link. It was designed for the purpose of standardizing scientific and industrial video products including cameras, cables and frame grabbers. The standard is maintained and administered by the Automated Imaging Association, or AIA, the global machine vision industry's trade group. Charge-coupled device. A charge-coupled device (CCD) is a sensor for recording images, consisting of an integrated circuit containing an array of linked, or coupled, capacitors. CCD sensors and cameras tend to be more sensitive, less noisy, and more expensive than CMOS sensors and cameras. CIE 1931 Color Space. In the study of the perception of color, one of the first mathematically defined color spaces was the CIE XYZ color space (also known as CIE 1931 color space), created by the International Commission on Illumination (CIE) in 1931. CMOS. CMOS ("see-moss")stands for complementary metal-oxide semiconductor, is a major class of integrated circuits. CMOS imaging sensors for machine vision are cheaper than CCD sensors but more noisy. CoaXPress. CoaXPress (CXP) is an asymmetric high speed serial communication standard over coaxial cable. CoaXPress combines high speed image data, low speed camera control and power over a single coaxial cable. The standard is maintained by JIIA, the Japan Industrial Imaging Association. Color. The perception of the frequency (or wavelength) of light, and can be compared to how pitch (or a musical note) is the perception of the frequency or wavelength of sound. Color blindness. Also known as color vision deficiency, in humans is the inability to perceive differences between some or all colors that other people can distinguish Color temperature. "White light" is commonly described by its color temperature. A traditional incandescent light source's color temperature is determined by comparing its hue with a theoretical, heated black-body radiator. The lamp's color temperature is the temperature in kelvins at which the heated black-body radiator matches the hue of the lamp. Color vision. CV is the capacity of an organism or machine to distinguish objects based on the wavelengths (or frequencies) of the light they reflect or emit. computer vision. The study and application of methods which allow computers to "understand" image content. Contrast. In visual perception, contrast is the difference in visual properties that makes an object (or its representation in an image) distinguishable from other objects and the background. C-Mount. Standardized adapter for optical lenses on CCD - cameras. C-Mount lenses have a back focal distance 17.5 mm vs. 12.5 mm for "CS-mount" lenses. A C-Mount lens can be used on a CS-Mount camera through the use of a 5 mm extension adapter. C-mount is a 1" diameter, 32 threads per inch mounting thread (1"-32UN-2A.) CS-Mount. Same as C-Mount but the focal point is 5 mm shorter. A CS-Mount lens will not work on a C-Mount camera. CS-mount is a 1" diameter, 32 threads per inch mounting thread. == D == Data matrix. A two dimensional Barcode. Depth of field. In optics, particularly photography and machine vision, the depth of field (DOF) is the distance in front of and behind the subject which appears to be in focus. Depth perception. DP is the visual ability to perceive the world in three dimensions. It is a trait common to many higher animals. Depth perception allows the beholder to accurately gauge the distance to an object. Diaphragm. In optics, a diaphragm is a thin opaque structure with an opening (aperture) at its centre. The role of the diaphragm is to stop the passage of light, except for the light passing through the aperture. == E == Edge detection. ED marks the points in a digital image at which the luminous intensity changes sharply. It also marks the points of luminous intensity changes of an object or spatial-taxon silhouette. Electromagnetic interference. Radio Frequency Interference (RFI) is electromagnetic radiation which is emitted by electrical circuits carrying rapidly changing signals, as a by-product of their normal operation, and which causes unwanted signals (interference or noise) to be induced in other circuits. == F == FireWire. FireWire (also known as i. Link or IEEE 1394) is a personal computer (and digital audio/video) serial bus interface standard, offering high-speed communications. It is often used as an interface for industrial cameras. Fixed-pattern noise. Flat-field correction. Frame grabber. An electronic device that captures individual, digital still frames from an analog video signal or a digital video stream. Fringe Projection Technique. 3D data acquisition technique employing projector displaying fringe pattern on a surface of measured piece, and one or more cameras recording image(s). Field of view. The field of view (FOV) is the part which can be seen by the machine vision system at one moment. The field of view depends from the lens of the system and from the working distance between object and camera. Focus. An image, or image point or region, is said to be in focus if light from object points is converged about as well as possible in the image; conversely, it is out of focus if light is not w

    Read more →
  • Reason maintenance

    Reason maintenance

    Reason maintenance is a knowledge representation approach to efficient handling of inferred information that is explicitly stored. Reason maintenance distinguishes between base facts, which can be defeated, and derived facts. As such it differs from belief revision which, in its basic form, assumes that all facts are equally important. Reason maintenance was originally developed as a technique for implementing problem solvers. It encompasses a variety of techniques that share a common architecture: two components—a reasoner and a reason maintenance system—communicate with each other via an interface. The reasoner uses the reason maintenance system to record its inferences and justifications of ("reasons" for) the inferences. The reasoner also informs the reason maintenance system which are the currently valid base facts (assumptions). The reason maintenance system uses the information to compute the truth value of the stored derived facts and to restore consistency if an inconsistency is derived. == Truth maintenance system == A truth maintenance system, or TMS, is a knowledge representation method for representing both beliefs and their dependencies and an algorithm called the "truth maintenance algorithm" that manipulates and maintains the dependencies. The name truth maintenance is due to the ability of these systems to restore consistency. A truth maintenance system maintains consistency between old believed knowledge and current believed knowledge in the knowledge base (KB) through revision. If the current believed statements contradict the knowledge in the KB, then the KB is updated with the new knowledge. It may happen that the same data will again be believed, and the previous knowledge will be required in the KB. If the previous data are not present, but may be required for new inference. But if the previous knowledge was in the KB, then no retracing of the same knowledge is needed. The use of TMS avoids such retracing; it keeps track of the contradictory data with the help of a dependency record. This record reflects the retractions and additions which makes the inference engine (IE) aware of its current belief set. == Algorithm == Each statement having at least one valid justification is made a part of the current belief set. When a contradiction is found, the statement(s) responsible for the contradiction are identified and the records are appropriately updated. This process is called dependency-directed backtracking. The TMS algorithm maintains the records in the form of a dependency network. Each node in the network is an entry in the KB (a premise, antecedent, or inference rule etc.) Each arc of the network represent the inference steps through which the node was derived. A premise is a fundamental belief which is assumed to be true. They do not need justifications. The set of premises are the basis from which justifications for all other nodes will be derived. == Justification == There are two types of justification for a node. They are: Support list [SL] Conditional proof (CP) == Examples == Many kinds of truth maintenance systems exist. Two major types are single-context and multi-context truth maintenance. In single context systems, consistency is maintained among all facts in memory (KB) and relates to the notion of consistency found in classical logic. Multi-context systems support paraconsistency by allowing consistency to be relevant to a subset of facts in memory, a context, according to the history of logical inference. This is achieved by tagging each fact or deduction with its logical history. Multi-agent truth maintenance systems perform truth maintenance across multiple memories, often located on different machines. de Kleer's assumption-based truth maintenance system (ATMS, 1986) was utilized in systems based upon KEE on the Lisp Machine. The first multi-agent TMS was created by Mason and Johnson. It was a multi-context system. Bridgeland and Huhns created the first single-context multi-agent system.

    Read more →
  • Vivification

    Vivification

    Vivification is an operation on a description logic knowledge base to improve performance of a semantic reasoner. Vivification replaces a disjunction of concepts C 1 ⊔ C 2 … ⊔ C n {\displaystyle C_{1}\sqcup C_{2}\ldots \sqcup C_{n}} by the least common subsumer of the concepts C 1 , C 2 , … C n {\displaystyle C_{1},C_{2},\ldots C_{n}} . The goal of this operation is to improve the performance of the reasoner by replacing a complex set of concepts with a single concept which subsumes the original concepts. For example, consider the example given in (Cohen 92): Suppose we have the concept PIANIST(Jill) ∨ ORGANIST(Jill) {\displaystyle {\textrm {PIANIST(Jill)}}\vee {\textrm {ORGANIST(Jill)}}} . This concept can be vivified into a simpler concept KEYBOARD-PLAYER(Jill) {\displaystyle {\textrm {KEYBOARD-PLAYER(Jill)}}} . This summarization leads to an approximation that may not be exactly equivalent to the original. == An approximation == Knowledge base vivification is not necessarily exact. If the reasoner is operating under the open world assumption we may get surprising results. In the previous example, if we replace the disjunction with the vivified concept, we will arrive at a surprising results. First, we find that the reasoner will no longer classify Jill as either a pianist or an organist. Even though ORGANIST {\displaystyle {\textrm {ORGANIST}}} and PIANIST {\displaystyle {\textrm {PIANIST}}} are the only two sub-classes, under the OWA we can no longer classify Jill as playing one or the other. The reason is that there may be another keyboard instrument (e.g. a harpsichord) that Jill plays but which does not have a specific subclass.

    Read more →
  • Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

    Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

    The Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, or SB 1047, was a failed 2024 California bill intended to "mitigate the risk of catastrophic harms from AI models so advanced that they are not yet known to exist". Specifically, the bill would have applied to models which cost more than $100 million to train and were trained using a quantity of computing power greater than 1026 integer or floating-point operations. SB 1047 would have applied to all AI companies doing business in California—the location of the company would not matter. The bill would have created protections for whistleblowers and required developers to perform risk assessments of their models prior to release, with guidance from the Government Operations Agency. It would also have established CalCompute, a University of California public cloud computing cluster for startups, researchers and community groups. == Background == The rapid increase in capabilities of AI systems in the 2020s, including the release of ChatGPT in November 2022, caused some researchers and members of the public to become concerned about the existential risks associated with increasingly powerful AI systems. Hundreds of tech executives and AI researchers, including two of the so-called "Godfathers of AI", Geoffrey Hinton and Yoshua Bengio, signed a statement in May 2023 calling for the mitigation of the "risk of extinction from AI" to be a global priority alongside "pandemics and nuclear war". However, the plausibility of these risks is still widely debated. Strong regulation of AI has been criticized for purportedly causing regulatory capture by large AI companies like OpenAI, a phenomenon in which regulation advances the interest of larger companies at the expense of smaller competition and the public in general, although OpenAI ended up opposing the bill. Other advocates of AI regulation aim to prevent bias and privacy violations, rather than existential risks. For example, some experts who view existential concerns as overblown and unrealistic view them as a distraction from near-term harms of AI like discriminatory automated decision making. In the face of existential concerns, technology companies have made voluntary commitments to conduct safety testing, for example at the AI Safety Summit and AI Seoul Summit. In 2023, not long before the bill was proposed, Governor Newsom of California and President Biden issued executive orders on artificial intelligence. State Senator Wiener said SB 1047 draws heavily on the Biden executive order, and is motivated by the absence of unified federal legislation on AI safety. Historically, California has passed regulation on several tech issues itself, including consumer privacy and net neutrality, in the absence of action by Congress. == History == === Proposal and voting === The bill was authored by State Senator Scott Wiener. Wiener first proposed AI legislation for California through an intent bill called SB 294, the Safety in Artificial Intelligence Act, in September 2023. On February 7, 2024, Wiener introduced SB 1047. On May 21, SB 1047 passed the Senate 32–1. The bill was significantly amended by Wiener on August 15, 2024, in response to industry advice. Amendments included adding clarifications, and removing the creation of a "Frontier Model Division" and the penalty of perjury. On August 28, the bill passed the State Assembly 48–16. Then, due to the amendments, the bill was once again voted on by the Senate, passing 30–9. === Veto by governor === On September 29, Governor Gavin Newsom vetoed the bill. The deadline for California lawmakers to overrule Newsom's veto was November 30, 2024. Newsom cited concerns over the bill's regulatory framework targeting only large AI models based on their computational size, while not taking into account whether the models are deployed in high-risk environments. Newsom emphasized that this approach could create a false sense of security, overlooking smaller models that might present equally significant risks. He acknowledged the need for AI safety protocols but stressed the importance of adaptability in regulation as AI technology continues to evolve rapidly. Governor Newsom also committed to working with technology experts, federal partners, and research institutions, including the Carnegie Endowment for International Peace, led by former California Supreme Court Justice Mariano-Florentino Cuéllar; and Stanford University's Human-Centered AI (HAI) Institute, led by Dr. Fei-Fei Li. He announced plans to collaborate with these entities to advance responsible AI development, aiming to protect the public while fostering innovation. == Provisions == SB 1047 would have covered AI models with training compute over 1026 integer or floating-point operations and a cost of over $100 million. If a covered model is fine-tuned using more than $10 million, the resulting model would also have been covered. The bill would have defined critical harms with respect to four categories: Creation or use of a chemical, biological, radiological, or nuclear weapon Cyberattacks on critical infrastructure causing mass casualties or at least $500 million of damage Autonomous crimes causing mass casualties or at least $500 million of damage Other harms of comparable severity Developers would have needed to create a "safety and security protocol" before training covered models. Before deployment, they would have submitted a statement of compliance, confirming they took reasonable care to take measures to prevent covered models that pose an unreasonable risk of critical harms. The statement would have included risk assessments and descriptions of their compliance process. These rules would have applied to both covered models and their derivatives, including post-training modifications, with annual third-party audits required starting in 2026. Safeguards to reduce risk included the ability to shut down the model, which has been variously described as a "kill switch" and "circuit breaker". Whistleblowing provisions would have protected employees who report safety problems and incidents. Additionally, SB 1047 would have created a public cloud computing cluster called CalCompute, associated with the University of California, to support startups, researchers, and community groups that lack large-scale computing resources. === Compliance and supervision === SB 1047 would have required developers, beginning January 1, 2026, to annually retain a third-party auditor to perform an independent audit of compliance with the requirements of the bill, as provided. The Government Operations Agency would have reviewed the results of safety tests and incidents, and issue guidance, standards, and best practices. The bill would have created a Board of Frontier Models to supervise the application of the bill by the Government Operations Agency. It is would be composed of 9 members. == Reception == === Subjects of debate === Proponents of the bill described its provisions as simple and narrowly focused, with Sen. Scott Weiner describing it as a "light-touch, basic safety bill". This was disputed by critics of the bill, who described the bill's language as vague and criticized it as consolidating power in the largest AI companies at the expense of smaller ones. Proponents, in turn, argued that the bill only applies to models trained using more than 1026 FLOPS and with over $100 million, or fine-tuned with more than $10 million, and that the threshold could be increased if needed. The penalty of perjury was also a subject of debate, and was eventually removed through an amendment. The scope of the "kill switch" requirement was also reduced, following concerns from open-source developers. The use of the term "reasonable assurance" in the bill was also controversial, and it was eventually amended to "reasonable care". Critics then argued that "reasonable care" imposed an excessive burden by requiring confidence that models could not be used to cause catastrophic harm; proponents claimed that the standard did not require certainty and that it already applied to AI developers under existing law. === Support and opposition === Individual supporters of the bill included Turing Award recipients Yoshua Bengio and Geoffrey Hinton, Elon Musk, Bill de Blasio, Kevin Esvelt, Dan Hendrycks, Vitalik Buterin, OpenAI whistleblowers Daniel Kokotajlo and William Saunders, Lawrence Lessig, Sneha Revanur, Stuart Russell, Jan Leike, actors Mark Ruffalo, Sean Astin, and Rosie Perez, Scott Aaronson, and Max Tegmark. Over 120 Hollywood celebrities, including Mark Hamill, Jane Fonda, and J. J. Abrams, also signed a statement in support of the bill. Max Tegmark likened the bill's focus on holding companies responsible for the harms caused by their models to the FDA requiring clinical trials before a company can release a drug to the market. Organizations sponsoring the bill included the Center for AI Safety, Economic Security California and Encode. The la

    Read more →
  • Graph cuts in computer vision and artificial intelligence

    Graph cuts in computer vision and artificial intelligence

    As applied in the field of computer vision, graph cut optimization can be employed to efficiently solve a wide variety of low-level computer vision problems (early vision), such as image smoothing, the stereo correspondence problem, image segmentation, object co-segmentation, numerous military applications (eg Automatic target recognition) and many other problems that can be formulated in terms of energy minimization (eg Climate Science and Environmental modelling). Graph cut techniques are now increasingly being used in combination with more general spatial Artificial intelligence techniques (eg to enforce structure in Large language model output to sharpen tumour boundaries and similarly for various Augmented reality, Self-driving car, Robotics, Google Maps applications etc). Many of these energy minimization problems can be approximated by solving a maximum flow problem in a graph (and thus, by the max-flow min-cut theorem, define a minimal cut of the graph). Under most formulations of such problems in computer vision, the minimum energy solution corresponds to the maximum a posteriori estimate of a solution. Although many computer vision algorithms involve cutting a graph (e.g. normalized cuts), the term "graph cuts" is applied specifically to those models which employ a max-flow/min-cut optimization (other graph cutting algorithms may be considered as graph partitioning algorithms). "Binary" problems (such as denoising a binary image) can be solved exactly using this approach; problems where pixels can be labeled with more than two different labels (such as stereo correspondence, or denoising of a grayscale image) cannot be solved exactly, but solutions produced are usually near the global optimum. == History == The foundational theory of graph cuts in computer vision was first developed by Margaret Greig, Bruce Porteous and Allan Seheult (GPS) of Durham University in a now legendary discussion contribution to Julian Besag's 1986 paper and a more detailed follow on paper in 1989. In the Bayesian statistical context of smoothing noisy images, using a Markov random field as the image prior distribution, they showed with a mathematically beautiful proof how the maximum a posteriori estimate of a binary image can be obtained exactly by maximizing the flow through an associated image network, or graph, involving the introduction of a source and sink and Log-likelihood ratios. The problem was shown to be efficiently solvable exactly, an unexpected result as the problem was believed to be computationally intractable (NP hard). GPS also addressed the computational cost of the max-flow algorithm on large graphs, a significant concern at the time. They proposed a partitioning algorithm (see Section 4 of GPS) involving the recursive amalgamation of non-overlapping blocks, or tiles, which gave a 12X increase in speed. This approach recursively solved and amalgamated independent sub-graphs until the whole graph was solved. While contemporaries like Geman and Geman had advocated Parallel computing in the context of Simulated annealing, the GPS blocking strategy offered a deterministic structure amenable to parallelisation and anticipated modern artificial intelligence design across multiple GPUs. However, until recently, this aspect of the paper was largely ignored and subsequent research focused on Serial computer global search trees, such as the Boykov-Kolmogorov algorithm. Although the general k {\displaystyle k} -colour problem is NP hard for k > 2 , {\displaystyle k>2,} the GPS approach has turned out to have very wide applicability in general computer vision problems. This was first demonstrated by Boykov, Veksler and Zabih who, in a seminal paper published more than 10 years after the original GPS paper, and in other important works, lit the blue touch paper for the general adoption of graph cut techniques in computer vision. They showed that, for general problems, the GPS approach can be applied iteratively to sequences of binary problems, using their now ubiquitous alpha-expansion algorithm, yielding near optimal solutions. Prior to these results, approximate local optimisation techniques such as simulated annealing (as proposed by the Geman brothers) or iterated conditional modes (a type of greedy algorithm suggested by Julian Besag) were used to solve such image smoothing problems. Building on these advancements, GPS graph cut optimization was subsequently adapted for interactive image segmentation, most notably through the "GrabCut" algorithm introduced by Carsten Rother, Vladimir Kolmogorov, and Andrew Blake of Microsoft Research, Cambridge. GrabCut extended earlier interactive graph cut methods by replacing monochrome image histograms with Gaussian mixture models to estimate colour distributions, and by employing an iterative GPS energy minimisation scheme. This approach significantly simplified user interaction, requiring only a rough bounding box around the target object rather than detailed user-drawn strokes, and it quickly became a standard tool in both academic research and commercial image editing software. The GPS paper connected and bridged profound ideas from Mathematical statistics (Bayes' theorem, Markov random field), Physics (Ising model), Optimisation (Energy function) and Computer science (Network flow problem) and led the move away from approximate local and slow optimisation approaches (eg simulated annealing) to more powerful exact, or near exact, faster global optimisation techniques. It is now recognised as seminal as it was well ahead of its time and, in particular, was published years before the computing power revolution of Moore's law and GPUs. Significantly, GPS was published in a mathematical statistics (rather than a computer vision) journal, and this led to it being overlooked by the computer vision community for many years. It is unofficially known as "The Velvet Underground" paper of computer vision (ie although very few computer vision people read the paper [bought the record], those that did, most importantly Boykov, Veksler and Zabih, started new and important research [formed a band]). This is confirmed by GPS' very large amplification ratio (2nd order citations/first order citations), estimated at well in excess of 100. Despite the foundational nature of the GPS work, formal recognition from the computer vision community has predominantly gone to the researchers who followed to extend and popularise the graph cut method. For example, Boykov, Veksler and Zabih deservedly received a Helmholtz Prize from the ICCV in 2011. This prize recognises ICCV papers from 10 or more years earlier that have had a significant impact on computer vision research. In 2011, Couprie et al. proposed a general image segmentation framework, called the "Power Watershed", that minimized a real-valued indicator function from [0,1] over a graph, constrained by user seeds (or unary terms) set to 0 or 1, in which the minimization of the indicator function over the graph is optimized with respect to an exponent p {\displaystyle p} . When p = 1 {\displaystyle p=1} , the Power Watershed is optimized by graph cuts, when p = 0 {\displaystyle p=0} the Power Watershed is optimized by shortest paths, p = 2 {\displaystyle p=2} is optimized by the random walker algorithm and p = ∞ {\displaystyle p=\infty } is optimized by the watershed algorithm. In this way, the Power Watershed may be viewed as a generalization of graph cuts that provides a straightforward connection with other energy optimization segmentation/clustering algorithms. == Binary segmentation of images == === Notation === Image: x ∈ { R , G , B } N {\displaystyle x\in \{R,G,B\}^{N}} Output: Segmentation (also called opacity) S ∈ R N {\displaystyle S\in R^{N}} (soft segmentation). For hard segmentation S ∈ { 0 for background , 1 for foreground/object to be detected } N {\displaystyle S\in \{0{\text{ for background}},1{\text{ for foreground/object to be detected}}\}^{N}} Energy function: E ( x , S , C , λ ) {\displaystyle E(x,S,C,\lambda )} where C is the color parameter and λ is the coherence parameter. E ( x , S , C , λ ) = E c o l o r + E c o h e r e n c e {\displaystyle E(x,S,C,\lambda )=E_{\rm {color}}+E_{\rm {coherence}}} Optimization: The segmentation can be estimated as a global minimum over S: arg ⁡ min S E ( x , S , C , λ ) {\displaystyle {\arg \min }_{S}E(x,S,C,\lambda )} === Existing methods === Standard Graph cuts: optimize energy function over the segmentation (unknown S value). Iterated Graph cuts: First step optimizes over the color parameters using K-means. Second step performs the usual graph cuts algorithm. These 2 steps are repeated recursively until convergence Dynamic graph cuts:Allows to re-run the algorithm much faster after modifying the problem (e.g. after new seeds have been added by a user). === Energy function === Pr ( x ∣ S ) = K − E {\displaystyle \Pr(x\mid S)=K^{-E}} where the energy E {\displaystyle E} is composed of two different mod

    Read more →
  • WordNet

    WordNet

    WordNet is a lexical database of semantic relations between words that links words into semantic relations including synonyms, hyponyms, and meronyms. The synonyms are grouped into synsets with short definitions and usage examples. It can thus be seen as a combination and extension of a dictionary and thesaurus. Its primary use is in automatic text analysis and artificial intelligence applications. It was first created in the English language and the English WordNet database and software tools have been released under a BSD style license and are freely available for download. The latest official release from Princeton was released in 2011. Princeton currently has no plans to release any new versions due to staffing and funding issues. New versions are still being released annually through the Open English WordNet website. Until about 2024 an online version was previously available through wordnet.princeton.edu. That version of WordNet has been deprecated, but a new online version is available at en-word.net. There are now WordNets in more than 200 languages. == History and team members == WordNet was first created in 1985, in English only, in the Cognitive Science Laboratory of Princeton University under the direction of psychology professor George Armitage Miller. It was later directed by Christiane Fellbaum. The project was initially funded by the U.S. Office of Naval Research, and later also by other U.S. government agencies including the DARPA, the National Science Foundation, the Disruptive Technology Office (formerly the Advanced Research and Development Activity) and REFLEX. George Miller and Christiane Fellbaum received the 2006 Antonio Zampolli Prize for their work with WordNet. The Global WordNet Association is a non-commercial organization that provides a platform for discussing, sharing and connecting WordNets for all languages in the world. Christiane Fellbaum and Piek Th.J.M. Vossen are its co-presidents. == Database contents == The database contains 155,327 words organized in 175,979 synsets for a total of 207,016 word-sense pairs; in compressed form, it is about 12 megabytes in size. It includes the lexical categories nouns, verbs, adjectives and adverbs but ignores prepositions, determiners and other function words. Words from the same lexical category that are roughly synonymous are grouped into synsets, which include simplex words as well as collocations like "eat out" and "car pool." The different senses of a polysemous word form are assigned to different synsets. A synset's meaning is further clarified with a short defining gloss and one or more usage examples. An example adjective synset is: good, right, ripe – (most suitable or right for a particular purpose; "a good time to plant tomatoes"; "the right time to act"; "the time is ripe for great sociological changes") All synsets are connected by means of semantic relations. These relations, which are not all shared by all lexical categories, include: Nouns hypernym: Y is a hypernym of X if every X is a (kind of) Y (canine is a hypernym of dog) hyponym: Y is a hyponym of X if every Y is a (kind of) X (dog is a hyponym of canine) coordinate term: Y is a coordinate term of X if X and Y share a hypernym (wolf is a coordinate term of dog, and dog is a coordinate term of wolf) holonym: Y is a holonym of X if X is a part of Y (building is a holonym of window) meronym: Y is a meronym of X if Y is a part of X (window is a meronym of building) Verbs hypernym: the verb Y is a hypernym of the verb X if the activity X is a (kind of) Y (to perceive is an hypernym of to listen) troponym: the verb Y is a troponym of the verb X if the activity Y is doing X in some manner (to lisp is a troponym of to talk) entailment: the verb Y is entailed by the verb X if by doing X you must be doing Y (to sleep is entailed by to snore) coordinate term: the verb Y is a coordinate term of the verb X if X and Y share a hypernym (to lisp is a coordinate term of to yell, and to yell is a coordinate term of to lisp) These semantic relations hold among all members of the linked synsets. Individual synset members (words) can also be connected with lexical relations. For example, (one sense of) the noun "director" is linked to (one sense of) the verb "direct" from which it is derived via a "morphosemantic" link. The morphology functions of the software distributed with the database try to deduce the lemma or stem form of a word from the user's input. Irregular forms are stored in a list, and looking up "ate" will return "eat," for example. == Knowledge structure == Both nouns and verbs are organized into hierarchies, defined by hypernym or IS A relationships. For instance, one sense of the word dog is found following hypernym hierarchy; the words at the same level represent synset members. Each set of synonyms has a unique index. At the top level, these hierarchies are organized into 25 beginner "trees" for nouns and 15 for verbs (called lexicographic files at a maintenance level). All are linked to a unique beginner synset, "entity". Noun hierarchies are far deeper than verb hierarchies. Adjectives are not organized into hierarchical trees. Instead, two "central" antonyms such as "hot" and "cold" form binary poles, while 'satellite' synonyms such as "steaming" and "chilly" connect to their respective poles via a "similarity" relations. The adjectives can be visualized in this way as "dumbbells" rather than as "trees". == Psycholinguistic aspects == The initial goal of the WordNet project was to build a lexical database that would be consistent with theories of human semantic memory developed in the late 1960s. Psychological experiments indicated that speakers organized their knowledge of concepts in an economic, hierarchical fashion. Retrieval time required to access conceptual knowledge seemed to be directly related to the number of hierarchies the speaker needed to "traverse" to access the knowledge. Thus, speakers could more quickly verify that canaries can sing because a canary is a songbird, but required slightly more time to verify that canaries can fly (where they had to access the concept "bird" on the superordinate level) and even more time to verify canaries have skin (requiring look-up across multiple levels of hyponymy, up to "animal"). While such psycholinguistic experiments and the underlying theories have been subject to criticism, some of WordNet's organization is consistent with experimental evidence. For example, anomic aphasia selectively affects speakers' ability to produce words from a specific semantic category, a WordNet hierarchy. Antonymous adjectives (WordNet's central adjectives in the dumbbell structure) are found to co-occur far more frequently than chance, a fact that has been found to hold for many languages. == As a lexical ontology == WordNet is sometimes called an ontology, a persistent claim that its creators do not make. The hypernym/hyponym relationships among the noun synsets can be interpreted as specialization relations among conceptual categories. In other words, WordNet can be interpreted and used as a lexical ontology in the computer science sense. However, such an ontology should be corrected before being used, because it contains hundreds of basic semantic inconsistencies; for example there are, (i) common specializations for exclusive categories and (ii) redundancies in the specialization hierarchy. Furthermore, transforming WordNet into a lexical ontology usable for knowledge representation should normally also involve (i) distinguishing the specialization relations into subtypeOf and instanceOf relations, and (ii) associating intuitive unique identifiers to each category. Although such corrections and transformations have been performed and documented as part of the integration of WordNet 1.7 into the cooperatively updatable knowledge base of WebKB-2, most projects claiming to reuse WordNet for knowledge-based applications (typically, knowledge-oriented information retrieval) simply reuse it directly. WordNet has also been converted to a formal specification, by means of a hybrid bottom-up top-down methodology to automatically extract association relations from it and interpret these associations in terms of a set of conceptual relations, formally defined in the DOLCE foundational ontology. In most works that claim to have integrated WordNet into ontologies, the content of WordNet has not simply been corrected when it seemed necessary; instead, it has been heavily reinterpreted and updated whenever suitable. This was the case when, for example, the top-level ontology of WordNet was restructured according to the OntoClean-based approach, or when it was used as a primary source for constructing the lower classes of the SENSUS ontology. == Limitations == The most widely discussed limitation of WordNet (and related resources like ImageNet) is that some of the semantic relations are more suited to concrete concepts than to abstract concepts. For example,

    Read more →
  • OpenVINO

    OpenVINO

    OpenVINO is an open-source software toolkit developed by Intel for optimizing and deploying deep learning models. It supports several popular model formats and categories, such as large language models, computer vision, and generative AI. OpenVINO is optimized for Intel hardware, but offers support for ARM/ARM64 processors. It sees great use in AI Sound Processing drivers when tied with Intel's Gaussian & Neural Accelerator (GNA). Based in C++, it extends API support for C and Python, as well as Node.js (in early preview). OpenVINO is cross-platform and free for use under Apache License 2.0. == Workflow == The simplest OpenVINO usage involves obtaining a model and running it as is. Yet for the best results, a more complete workflow is suggested: obtain a model in one of supported frameworks, convert the model to OpenVINO IR using the OpenVINO Converter tool, optimize the model, using training-time or post-training options provided by OpenVINO's NNCF. execute inference, using OpenVINO Runtime by specifying one of several inference modes. == OpenVINO model format == OpenVINO IR is the default format used to run inference. It is saved as a set of two files, .bin and .xml, containing weights and topology, respectively. It is obtained by converting a model from one of the supported frameworks, using the application's API or a dedicated converter. Models of the supported formats may also be used for inference directly, without prior conversion to OpenVINO IR. Such an approach is more convenient but offers fewer optimization options and lower performance, since the conversion is performed automatically before inference. Some pre-converted models can be found in the Hugging Face repository. The supported model formats are: PyTorch TensorFlow TensorFlow Lite ONNX (including formats that may be serialized to ONNX) PaddlePaddle JAX/Flax == OS support == OpenVINO runs on Windows, Linux and MacOS.

    Read more →
  • Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

    Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

    The Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, or SB 1047, was a failed 2024 California bill intended to "mitigate the risk of catastrophic harms from AI models so advanced that they are not yet known to exist". Specifically, the bill would have applied to models which cost more than $100 million to train and were trained using a quantity of computing power greater than 1026 integer or floating-point operations. SB 1047 would have applied to all AI companies doing business in California—the location of the company would not matter. The bill would have created protections for whistleblowers and required developers to perform risk assessments of their models prior to release, with guidance from the Government Operations Agency. It would also have established CalCompute, a University of California public cloud computing cluster for startups, researchers and community groups. == Background == The rapid increase in capabilities of AI systems in the 2020s, including the release of ChatGPT in November 2022, caused some researchers and members of the public to become concerned about the existential risks associated with increasingly powerful AI systems. Hundreds of tech executives and AI researchers, including two of the so-called "Godfathers of AI", Geoffrey Hinton and Yoshua Bengio, signed a statement in May 2023 calling for the mitigation of the "risk of extinction from AI" to be a global priority alongside "pandemics and nuclear war". However, the plausibility of these risks is still widely debated. Strong regulation of AI has been criticized for purportedly causing regulatory capture by large AI companies like OpenAI, a phenomenon in which regulation advances the interest of larger companies at the expense of smaller competition and the public in general, although OpenAI ended up opposing the bill. Other advocates of AI regulation aim to prevent bias and privacy violations, rather than existential risks. For example, some experts who view existential concerns as overblown and unrealistic view them as a distraction from near-term harms of AI like discriminatory automated decision making. In the face of existential concerns, technology companies have made voluntary commitments to conduct safety testing, for example at the AI Safety Summit and AI Seoul Summit. In 2023, not long before the bill was proposed, Governor Newsom of California and President Biden issued executive orders on artificial intelligence. State Senator Wiener said SB 1047 draws heavily on the Biden executive order, and is motivated by the absence of unified federal legislation on AI safety. Historically, California has passed regulation on several tech issues itself, including consumer privacy and net neutrality, in the absence of action by Congress. == History == === Proposal and voting === The bill was authored by State Senator Scott Wiener. Wiener first proposed AI legislation for California through an intent bill called SB 294, the Safety in Artificial Intelligence Act, in September 2023. On February 7, 2024, Wiener introduced SB 1047. On May 21, SB 1047 passed the Senate 32–1. The bill was significantly amended by Wiener on August 15, 2024, in response to industry advice. Amendments included adding clarifications, and removing the creation of a "Frontier Model Division" and the penalty of perjury. On August 28, the bill passed the State Assembly 48–16. Then, due to the amendments, the bill was once again voted on by the Senate, passing 30–9. === Veto by governor === On September 29, Governor Gavin Newsom vetoed the bill. The deadline for California lawmakers to overrule Newsom's veto was November 30, 2024. Newsom cited concerns over the bill's regulatory framework targeting only large AI models based on their computational size, while not taking into account whether the models are deployed in high-risk environments. Newsom emphasized that this approach could create a false sense of security, overlooking smaller models that might present equally significant risks. He acknowledged the need for AI safety protocols but stressed the importance of adaptability in regulation as AI technology continues to evolve rapidly. Governor Newsom also committed to working with technology experts, federal partners, and research institutions, including the Carnegie Endowment for International Peace, led by former California Supreme Court Justice Mariano-Florentino Cuéllar; and Stanford University's Human-Centered AI (HAI) Institute, led by Dr. Fei-Fei Li. He announced plans to collaborate with these entities to advance responsible AI development, aiming to protect the public while fostering innovation. == Provisions == SB 1047 would have covered AI models with training compute over 1026 integer or floating-point operations and a cost of over $100 million. If a covered model is fine-tuned using more than $10 million, the resulting model would also have been covered. The bill would have defined critical harms with respect to four categories: Creation or use of a chemical, biological, radiological, or nuclear weapon Cyberattacks on critical infrastructure causing mass casualties or at least $500 million of damage Autonomous crimes causing mass casualties or at least $500 million of damage Other harms of comparable severity Developers would have needed to create a "safety and security protocol" before training covered models. Before deployment, they would have submitted a statement of compliance, confirming they took reasonable care to take measures to prevent covered models that pose an unreasonable risk of critical harms. The statement would have included risk assessments and descriptions of their compliance process. These rules would have applied to both covered models and their derivatives, including post-training modifications, with annual third-party audits required starting in 2026. Safeguards to reduce risk included the ability to shut down the model, which has been variously described as a "kill switch" and "circuit breaker". Whistleblowing provisions would have protected employees who report safety problems and incidents. Additionally, SB 1047 would have created a public cloud computing cluster called CalCompute, associated with the University of California, to support startups, researchers, and community groups that lack large-scale computing resources. === Compliance and supervision === SB 1047 would have required developers, beginning January 1, 2026, to annually retain a third-party auditor to perform an independent audit of compliance with the requirements of the bill, as provided. The Government Operations Agency would have reviewed the results of safety tests and incidents, and issue guidance, standards, and best practices. The bill would have created a Board of Frontier Models to supervise the application of the bill by the Government Operations Agency. It is would be composed of 9 members. == Reception == === Subjects of debate === Proponents of the bill described its provisions as simple and narrowly focused, with Sen. Scott Weiner describing it as a "light-touch, basic safety bill". This was disputed by critics of the bill, who described the bill's language as vague and criticized it as consolidating power in the largest AI companies at the expense of smaller ones. Proponents, in turn, argued that the bill only applies to models trained using more than 1026 FLOPS and with over $100 million, or fine-tuned with more than $10 million, and that the threshold could be increased if needed. The penalty of perjury was also a subject of debate, and was eventually removed through an amendment. The scope of the "kill switch" requirement was also reduced, following concerns from open-source developers. The use of the term "reasonable assurance" in the bill was also controversial, and it was eventually amended to "reasonable care". Critics then argued that "reasonable care" imposed an excessive burden by requiring confidence that models could not be used to cause catastrophic harm; proponents claimed that the standard did not require certainty and that it already applied to AI developers under existing law. === Support and opposition === Individual supporters of the bill included Turing Award recipients Yoshua Bengio and Geoffrey Hinton, Elon Musk, Bill de Blasio, Kevin Esvelt, Dan Hendrycks, Vitalik Buterin, OpenAI whistleblowers Daniel Kokotajlo and William Saunders, Lawrence Lessig, Sneha Revanur, Stuart Russell, Jan Leike, actors Mark Ruffalo, Sean Astin, and Rosie Perez, Scott Aaronson, and Max Tegmark. Over 120 Hollywood celebrities, including Mark Hamill, Jane Fonda, and J. J. Abrams, also signed a statement in support of the bill. Max Tegmark likened the bill's focus on holding companies responsible for the harms caused by their models to the FDA requiring clinical trials before a company can release a drug to the market. Organizations sponsoring the bill included the Center for AI Safety, Economic Security California and Encode. The la

    Read more →
  • Language model benchmark

    Language model benchmark

    A language model benchmark is a standardized test designed to evaluate the performance of language models on various natural language processing tasks. These tests are intended for comparing different models' capabilities in areas such as language understanding, generation, and reasoning. Benchmarks generally consist of a dataset and corresponding evaluation metrics. The dataset provides text samples and annotations, while the metrics measure a model's performance on tasks like answering questions, text classification, and machine translation. These benchmarks are developed and maintained by academic institutions, research organizations, and industry players to track progress in the field. In addition to accuracy, the metrics can include throughput, energy efficiency, bias, trust, and sustainability. == Overview == === Types === Benchmarks may be described by the following adjectives, not mutually exclusive: Classical: These tasks are studied in natural language processing, even before the advent of deep learning. Examples include the Penn Treebank for testing syntactic and semantic parsing, as well as bilingual translation benchmarked by BLEU scores. Question answering: These tasks have a text question and a text answer, often multiple-choice. They can be open-book or closed-book. Open-book QA resembles reading comprehension questions, with relevant passages included as annotation in the question, in which the answer appears. Closed-book QA includes no relevant passages. Closed-book QA is also called open-domain question-answering. Before the era of large language models, open-book QA was more common, and understood as testing information retrieval methods. Closed-book QA became common since GPT-2 as a method to measure knowledge stored within model parameters. Omnibus: An omnibus benchmark combines many benchmarks, often previously published. It is intended as an all-in-one benchmarking solution. Reasoning: These tasks are usually in the question-answering format, but are intended to be more difficult than standard question answering. Multimodal: These tasks require processing not only text, but also other modalities, such as images and sound. Examples include OCR and transcription. Agency: These tasks are for a language-model–based software agent that operates a computer for a user, such as editing images, browsing the web, etc. Adversarial: A benchmark is "adversarial" if the items in the benchmark are picked specifically so that certain models do badly on them. Adversarial benchmarks are often constructed after state of the art (SOTA) models have saturated (achieved 100% performance) a benchmark, to renew the benchmark. A benchmark is "adversarial" only at a certain moment in time, since what is adversarial may cease to be adversarial as newer SOTA models appear. Public/Private: A benchmark might be partly or entirely private, meaning that some or all of the questions are not publicly available. The idea is that if a question is publicly available, then it might be used for training, which would be "training on the test set" and invalidate the result of the benchmark. Usually, only the guardians of the benchmark have access to the private subsets, and to score a model on such a benchmark, one must send the model weights, or provide API access, to the guardians. The boundary between a benchmark and a dataset is not sharp. Generally, a dataset contains three "splits": training, test, and validation. Both the test and validation splits are essentially benchmarks. In general, a benchmark is distinguished from a test/validation dataset in that a benchmark is typically intended to be used to measure the performance of many different models that are not trained specifically for doing well on the benchmark, while a test/validation set is intended to be used to measure the performance of models trained specifically on the corresponding training set. In other words, a benchmark may be thought of as a test/validation set without a corresponding training set. Conversely, certain benchmarks may be used as a training set, such as the English Gigaword or the One Billion Word Benchmark, which in modern language is just the negative log-likelihood loss on a pretraining set with 1 billion words. Indeed, the distinction between benchmark and dataset in language models became sharper after the rise of the pretraining paradigm, whereby a model is first trained on massive, unlabeled datasets to learn general language patterns, syntax, and knowledge (pretraining), and the base model is then adapted to specific, downstream tasks using smaller, labeled datasets (fine-tuning). === Lifecycle === Generally, the life cycle of a benchmark consists of the following steps: Inception: A benchmark is published. It can be simply given as a demonstration of the power of a new model (implicitly) that others then picked up as a benchmark, or as a benchmark that others are encouraged to use (explicitly). Growth: More papers and models use the benchmark, and the performance on the benchmark grows. Maturity, degeneration or deprecation: A benchmark may be saturated, after which researchers move on to other benchmarks. Progress on the benchmark may also be neglected as the field moves to focus on other benchmarks. Renewal: A saturated benchmark can be upgraded to make it no longer saturated, allowing further progress. === Construction === Like datasets, benchmarks are typically constructed by several methods, individually or in combination: Web scraping: Ready-made question-answer pairs may be scraped online, such as from websites that teach mathematics and programming. Conversion: Items may be constructed programmatically from scraped web content, such as by blanking out named entities from sentences, and asking the model to fill in the blank. This was used for making the CNN/Daily Mail Reading Comprehension Task. Crowd sourcing: Items may be constructed by paying people to write them, such as on Amazon Mechanical Turk. This was used for making the MCTest. === Evaluation === Generally, benchmarks are fully automated. This limits the questions that can be asked. For example, with mathematical questions, "proving a claim" would be difficult to automatically check, while "calculate an answer with a unique integer answer" would be automatically checkable. With programming tasks, the answer can generally be checked by running unit tests, with an upper limit on runtime. The benchmark scores are of the following kinds: For multiple choice or cloze questions, common scores are accuracy (frequency of correct answer), precision, recall, F1 score, etc. pass@n: The model is given n {\displaystyle n} attempts to solve each problem. If any attempt is correct, the model earns a point. The pass@n score is the model's average score over all problems. k@n: The model makes n {\displaystyle n} attempts to solve each problem, but only k {\displaystyle k} attempts out of them are selected for submission. If any submission is correct, the model earns a point. The k@n score is the model's average score over all problems. cons@n: The model is given n {\displaystyle n} attempts to solve each problem. If the most common answer is correct, the model earns a point. The cons@n score is the model's average score over all problems. Here "cons" stands for "consensus" or "majority voting". The pass@n score can be estimated more accurately by making N > n {\displaystyle N>n} attempts, and use the unbiased estimator 1 − ( N − c n ) ( N n ) {\displaystyle 1-{\frac {\binom {N-c}{n}}{\binom {N}{n}}}} , where c {\displaystyle c} is the number of correct attempts. For less well-formed tasks, where the output can be any sentence, there are the following commonly used scores including BLEU ROUGE, METEOR, NIST, word error rate, LEPOR, CIDEr, and SPICE. === Issues === error: Some benchmark answers may be wrong. ambiguity: Some benchmark questions may be ambiguously worded. subjective: Some benchmark questions may not have an objective answer at all. This problem generally prevents creative writing benchmarks. Similarly, this prevents benchmarking writing proofs in natural language, though benchmarking proofs in a formal language is possible. open-ended: Some benchmark questions may not have a single answer of a fixed size. This problem generally prevents programming benchmarks from using more natural tasks such as "write a program for X", and instead uses tasks such as "write a function that implements specification X". inter-annotator agreement: Some benchmark questions may be not fully objective, such that even people would not agree with 100% on what the answer should be. This is common in natural language processing tasks, such as syntactic annotation. shortcut: Some benchmark questions may be easily solved by an "unintended" shortcut. For example, in the SNLI benchmark, having a negative word like "not" in the second sentence is a strong signal for the "Contradiction" category, regardless of what the se

    Read more →
  • Public First Action

    Public First Action

    Public First Action is a 501(c)(4) nonprofit organization focused on United States public policy related to artificial intelligence. Public First Action is a bipartisan group that advocates for AI transparency, safeguards, and export controls on advanced AI chips. The organization is aligned with the political action committees Jobs and Democracy, Defending Our Values and Public First. == History == Public First Action was formed in 2025 by former Congressmen Brad Carson, a Democrat, and Chris Stewart, a Republican, to advocate for federal, state, and local regulations related to AI. The group's formation followed the founding of a super PAC network, Leading the Future, which advocates for deregulation of the AI industry and faster development of the new technology. Public First Action supports measures that would increase transparency at frontier AI companies and impose export controls on advanced AI chips, in addition to opposing the preemption of state-level AI laws. In February 2026, Public First Action received $20 million from the AI company Anthropic. That same month, the group announced plans to support 30 to 50 Democrats and Republicans in state and federal races, with Public First Action and aligned super PACs launching advertisements in Nebraska, Tennessee, and other states. In one ad, Public First Action touted Senator Marsha Blackburn for her work on child online safety. As of 2026, the group plans to raise between $50 and $75 million for public oversight of AI and related reforms. == Organization == === Leadership and funding === Public First Action is led by Carson and Stewart. The group has raised nearly $50 million in funding with a goal of raising $75 million during the 2026 midterms. Anthropic has contributed $20 million to the group. === Structure === Public First Action is aligned with three political action committees: "Jobs and Democracy", which supports Democratic candidates; "Defending Our Values", which supports Republican candidates; and "Public First", which supports both Republicans and Democrats.

    Read more →
  • Journal of Experimental and Theoretical Artificial Intelligence

    Journal of Experimental and Theoretical Artificial Intelligence

    The Journal of Experimental and Theoretical Artificial Intelligence is a quarterly peer-reviewed scientific journal published by Taylor and Francis. It covers all aspects of artificial intelligence and was established in 1989. The editor-in-chief is Eric Dietrich (Binghamton University), the deputy editors-in-chief are Li Pheng Khoo (School of Mechanical & Aerospace Engineering, Nanyang Technological University) and Antonio Lieto (Department of Computer Science, University of Turin). == Abstracting and indexing == The journal is abstracted and indexed in: According to the Journal Citation Reports, the journal has a 2020/2021 impact factor of 2.340 .

    Read more →