AI Assistant Quest 3

AI Assistant Quest 3 — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

  • Color vision

    Color vision

    Color vision (CV), a feature of visual perception, is an ability to perceive differences between light composed of different frequencies independently of light intensity. Color perception is a part of the larger visual system and is mediated by a complex process between neurons that begins with differential stimulation of different types of photoreceptors by light entering the eye. Those photoreceptors then emit outputs that are propagated through many layers of neurons ultimately leading to higher cognitive functions in the brain. Color vision is found in many animals and is mediated by similar underlying mechanisms with common types of biological molecules and a complex history of the evolution of color vision within different animal taxa. In primates, color vision may have evolved under selective pressure for a variety of visual tasks including the foraging for nutritious young leaves, ripe fruit, and flowers, as well as detecting predator camouflage and emotional states in other primates. == Wavelength == Isaac Newton discovered that white light after being split into its component colors when passed through a dispersive prism could be recombined to make white light by passing them through a different prism. The visible light spectrum ranges from about 380 to 740 nanometers. Spectral colors (colors that are produced by a narrow band of wavelengths) such as red, orange, yellow, green, cyan, blue, and violet can be found in this range. These spectral colors do not refer to a single wavelength, but rather to a set of wavelengths: red, 625–740 nm; orange, 590–625 nm; yellow, 565–590 nm; green, 500–565 nm; cyan, 485–500 nm; blue, 450–485 nm; violet, 380–450 nm. Wavelengths longer or shorter than this range are called infrared or ultraviolet, respectively. Humans cannot generally see these wavelengths, but other animals may. === Hue detection === Sufficient differences in wavelength cause a difference in the perceived hue; the just-noticeable difference in wavelength varies from about 1 nm in the blue-green and yellow wavelengths to 10 nm and more in the longer red and shorter blue wavelengths. Although the human eye can distinguish up to a few hundred hues, when those pure spectral colors are mixed together or diluted with white light, the number of distinguishable chromaticities can be much higher. In very low light levels, vision is scotopic: light is detected by rod cells of the retina. Rods are maximally sensitive to wavelengths near 500 nm and play little, if any, role in color vision. In brighter light, such as daylight, vision is photopic: light is detected by cone cells which are responsible for color vision. Cones are sensitive to a range of wavelengths, but are most sensitive to wavelengths near 555 nm. Between these regions, mesopic vision comes into play and both rods and cones provide signals to the retinal ganglion cells. The shift in color perception from dim light to daylight gives rise to differences known as the Purkinje effect. The perception of "white" is formed by the entire spectrum of visible light, or by mixing colors of just a few wavelengths in animals with few types of color receptors. In humans, white light can be perceived by combining wavelengths such as red, green, and blue, or just a pair of complementary colors such as blue and yellow. === Non-spectral colors === There are a variety of colors in addition to spectral colors and their hues. These include grayscale colors, shades of colors obtained by mixing grayscale colors with spectral colors, violet-red colors, impossible colors, and metallic colors. Grayscale colors include white, gray, and black. Rods contain rhodopsin, which reacts to light intensity, providing grayscale coloring. Shades include colors such as pink or brown. Pink is obtained from mixing red and white. Brown may be obtained from mixing orange with gray or black. Navy is obtained from mixing blue and black. Violet-red colors include hues and shades of magenta. The light spectrum is a line on which violet is one end and the other is red, and yet we see hues of purple that connect those two colors. Impossible colors are a combination of cone responses that cannot be naturally produced. For example, medium cones cannot be activated completely on their own; if they were, we would see a 'hyper-green' color. == Dimensionality == Color vision is categorized foremost according to the dimensionality of the color gamut, which is defined by the number of primaries required to represent the color vision. This is generally equal to the number of photopsins expressed: a correlation that holds for vertebrates but not invertebrates. The common vertebrate ancestor possessed four photopsins (expressed in cones) plus rhodopsin (expressed in rods), so was tetrachromatic. However, many vertebrate lineages have lost one or many photopsin genes, leading to lower-dimension color vision. The dimensions of color vision range from 1-dimensional and up: == Physiology of color perception == Perception of color begins with specialized retinal cells known as cone cells. Cone cells contain different forms of opsin – a pigment protein – that have different spectral sensitivities. Humans contain three types, resulting in trichromatic color vision. Each individual cone contains pigments composed of opsin apoprotein covalently linked to a light-absorbing prosthetic group: either 11-cis-hydroretinal or, more rarely, 11-cis-dehydroretinal. The cones are conventionally labeled according to the ordering of the wavelengths of the peaks of their spectral sensitivities: short (S), medium (M), and long (L) cone types. These three types do not correspond well to particular colors as we know them. Rather, the perception of color is achieved by a complex process that starts with the differential output of these cells in the retina and which is finalized in the visual cortex and associative areas of the brain. For example, while the L cones have been referred to simply as red receptors, microspectrophotometry has shown that their peak sensitivity is in the greenish-yellow region of the spectrum. Similarly, the S cones and M cones do not directly correspond to blue and green, although they are often described as such. The RGB color model, therefore, is a convenient means for representing color but is not directly based on the types of cones in the human eye. The peak response of human cone cells varies, even among individuals with typical color vision; in some non-human species this polymorphic variation is even greater, and it may well be adaptive. === Theories === Two complementary theories of color vision are the trichromatic theory and the opponent process theory. The trichromatic theory, or Young–Helmholtz theory, proposed in the 19th century by Thomas Young and Hermann von Helmholtz, posits three types of cones preferentially sensitive to blue, green, and red, respectively. Others have suggested that the trichromatic theory is not specifically a theory of color vision but a theory of receptors for all vision, including color but not specific or limited to it. Equally, it has been suggested that the relationship between the phenomenal opponency described by Ewald Hering and the physiological opponent processes are not straightforward (see below), making of physiological opponency a mechanism that is relevant to the whole of vision, and not just to color vision alone. Hering proposed the opponent process theory in 1872. It states that the visual system interprets color in an antagonistic way: red vs. green, blue vs. yellow, black vs. white. Both theories are generally accepted as valid, describing different stages in visual physiology, visualized in the adjacent diagram. Green–magenta and blue–yellow are scales with mutually exclusive boundaries. In the same way that there cannot exist a "slightly negative" positive number, a single eye cannot perceive a bluish-yellow or a reddish-green. Although these two theories are both currently widely accepted theories, past and more recent work has led to criticism of the opponent process theory, stemming from a number of what are presented as discrepancies in the standard opponent process theory. For example, the phenomenon of an after-image of complementary color can be induced by fatiguing the cells responsible for color perception, by staring at a vibrant color for a length of time, and then looking at a white surface. This phenomenon of complementary colors shows that cyan, rather than green, is the complement of red, and that magenta, rather than red, is the complement of green. It therefore also shows that the reddish-green color supposed to be impossible by opponent process theory is actually the color yellow. Although this phenomenon is more readily explained by the trichromatic theory, explanations for the discrepancy may include alterations to the opponent process theory, such as redefining the opponent colors as red vs. cyan, to reflect this effect. Despite such criticis

    Read more →
  • Sriram Krishnan

    Sriram Krishnan

    Sriram Krishnan (born 1984) is a tech executive and White House official, currently serving as the Senior White House Policy Advisor on Artificial Intelligence. Krishnan was named a Time Person of the Year in 2025 as an "Architect of Artificial Intelligence." He was described in Time as providing the "wake-up call that we needed" to the other AI builders, leading to "a multiyear, $500 billion initiative dubbed Stargate" to push American-made AI, as well as numerous other AI initiatives. Also in December 2025, President Trump said of Krishnan, "without him, things on AI would not function well" and cited Krishnan as the leading figure behind the American executive order on AI. As the leader of the United States' policy team regarding artificial intelligence, Krishnan plays "a significant role in shaping the administration’s approach to AI and driving measures to advance federal adoption of AI." The role calls for removing barriers to AI adoption within the government, driving vendors toward solutions suitable for federal needs, designing sensible regulation of private-sector AI, and conducting "AI diplomacy". He has stated a policy goal of "reinvigorating US dominance in emerging technologies," including AI. He also represents the United States' interests in AI abroad, such as at the Paris AI Summit. He is one of the authors of the American "AI Action Plan" released in July, 2025, which he contends is necessary to win the "existential race with China" for AI supremacy. Krishnan, a U.S. citizen born in India, is also a venture capitalist, podcaster, product manager and author. Early in his career, he led product teams at Microsoft, Twitter, Yahoo!, Facebook, and Snap. In addition to his work as an investor and technologist, he and his wife, Aarthi Ramamurthy, rose to additional prominence in 2021 as podcast hosts. He served as a general partner at the venture capital firm Andreessen Horowitz and led its London office. In 2022, Krishnan announced that he was working with Elon Musk on the rebuilding of Twitter following Musk's acquisition of the company. On December 22, 2024, US president-elect Donald Trump announced that Krishnan would be Senior White House Policy Advisor on Artificial Intelligence in his incoming administration; in 2026 he joined the National Economic Council. == Early life and education == Krishnan was born in Chennai, India. He earned his Bachelor of Technology in Information Technology from SRM University (2001–2005), moved to the United States in 2007 to join Microsoft, and became a naturalized U.S. citizen in 2016. == Career == === Early career === In 2007, he began working at Microsoft where he served as a program manager for Visual Studio. At Facebook, Krishnan built the Facebook Audience Network, a competitive platform to Google's ad technologies. At Twitter, he led product and core user experience, driving a 20% annual user growth rate and launching a redesigned home page and events experience. === Andreessen Horowitz === Krishnan was appointed a general partner of American venture capital firm Andreessen Horowitz ("a16z") in February 2021. He was anticipated to serve consumer and social markets, however he has also theorized on the impact of "deep tech" on society. In 2023 he was appointed to lead the firm's London office, its first non-US location. The office is expected to serve Web3 investments as well as AI and other fields. Krishnan announced that he would leave the firm at the end of 2024. === Social media and AI === In 2022, various news media reported that Krishnan was assisting Elon Musk in the revamp of Twitter following Musk's takeover of the company. Additional reports named Krishnan as the leading candidate for the role of CEO of the newly private company. Krishnan penned a 2023 New York Times opinion column regarding social media, AI, and related fields. He predicted a rise in the number and diversity of online spaces due to decentralization and platforms like Farcaster, Bluesky and Mastodon. === Public office === In 2024, the Financial Times reported that Krishnan was active in international affairs, reintroducing Boris Johnson to Elon Musk, following Musk's nomination to the proposed Department of Government Efficiency. Krishnan was also reported as potentially leaving a16z at the end of the year to "be jumping into something I've wanted to spend [his] energy on," which was widely reported as being related to Musk's and Vivek Ramaswamy's work at DOGE. Others reported to be involved include Joe Lonsdale, Marc Andreesen, Bill Ackman, and Travis Kalanick. On December 22, 2024, US president-elect Donald Trump announced that he would be Senior White House Policy Advisor on Artificial Intelligence in his incoming administration. On February 6, 2025, Reuters reported that Krishnan would be accompanying Vice President Vance to the Paris AI Summit, a "major artificial intelligence" event later that month. Other members of the White House Office of Science and Technology Policy would also be joining the event with around 100 other countries to "focus on AI's potential." Krishnan joined a U.S. technology policy delegation to the Middle East in advance of President Trump's visit in May 2025. Conducting "AI diplomacy," Krishnan negotiated the spread of U.S. AI technologies with Crown Prince Mohammed bin Salman of Saudi Arabia, as well as other means to strengthen bilateral trade in artificial intelligence technologies. He explained that the goal of the diplomatic mission was that "we want American A.I. to spread." Krishnan, along with David Sacks and Michael Kratsios, were credited as authors of the American AI Action Plan released in July 2025. The plan is "the administration’s most significant policy directive" regarding artificial intelligence; it calls for financing to support the global spread of American AI models and a policy to enforce neutrality in models. The Washington Post referred to the plan as a "bold action to ensure that American AI remains at the cutting edge." The AI Action Plan is a continuation of prior efforts to reduce barriers to U.S. production of AI systems and the removal of rules that were considered to hinder such growth. Later in 2025, at the POLITICO AI & Tech Summit, Krishnan called national AI development "an existential race with China." He suggested that private companies are best positioned to create new models, quipping "let them cook." He further suggested that state-by-state regulation of AI technologies may hinder national AI competitiveness. Also in 2025, at the Axios AI+ Summit, Krishnan stated that the United States and China are in a race for AI supremacy, in which the winner will be judged by market share. Winning the race is a "business strategy" to Krishnan. Krishnan was named in the 2025 Time Person of the Year article as an "AI Architect". === The Aarthi and Sriram Show and other media === In early 2021, Krishnan and his wife, Aarthi Ramamurthy, launched a Clubhouse talk show that "focuses on organic conversations on anything from startups to venture capitalism and cryptocurrencies." An early appearance by Elon Musk on the Good Time Show was described as the first show that "broke Clubhouse" by rapidly exceeding the limit of 5,000 simultaneous users. The desire to interact with a larger community led to a variety of later innovations to allow streaming and replaying of Clubhouse chats. On that episode, Elon Musk grilled Robinhood CEO Vlad Tenev regarding the GameStop trading controversy. As of December 2021, the show had over 187,000 subscribers, plus 735,000 subscribers between Krishnan and Ramamurthy's personal Clubhouse accounts. Other guests have included Facebook CEO Mark Zuckerberg, Diane von Fürstenberg, Tony Hawk, MrBeast, and A.R. Rahman. In 2022, the Good Time Show moved to YouTube. It then evolved to a podcasting format under the name The Aarthi and Sriram Show, with both audio and video content. The Hollywood Reporter reported that the podcast had received more than 1 million downloads by early 2023. == Personal life == Krishnan is married to Aarthi Ramamurthy, co-host of The Aarthi and Sriram Show (formerly the Good Time Show) and a serial entrepreneur. They met in college in 2003 through a Yahoo! chat room related to a coding project and began dating in 2006 and eloped in 2010. == Awards == Time Person of the Year - 2025

    Read more →
  • Composite Capability/Preference Profiles

    Composite Capability/Preference Profiles

    Composite Capability/Preference Profiles (CC/PP) is a specification for defining capabilities and preferences of user agents (also known as "delivery context"). The delivery context can be used to guide the process of tailoring content for a user agent. CC/PP is a vocabulary extension of the Resource Description Framework (RDF). The CC/PP specification is maintained by the W3C's Ubiquitous Web Applications Working Group (UWAWG) Working Group. == History == Composite Capability/Preference Profiles (CC/PP): Structure and Vocabularies 1.0 became a W3C recommendation on 15 January 2004. A "Last-Call Working-Draft" of CC/PP 2.0 was issued in April 2007

    Read more →
  • Yale shooting problem

    Yale shooting problem

    The Yale shooting problem is a conundrum or scenario in formal situational logic on which early logical solutions to the frame problem fail. The name of this problem comes from a scenario proposed by its inventors, Steve Hanks and Drew McDermott, working at Yale University when they proposed it. In this scenario, Fred (later identified as a turkey) is initially alive and a gun is initially unloaded. Loading the gun, waiting for a moment, and then shooting the gun at Fred is expected to kill Fred. However, if inertia is formalized in logic by minimizing the changes in this situation, then it cannot be uniquely proved that Fred is dead after loading, waiting, and shooting. In one solution, Fred indeed dies; in another (also logically correct) solution, the gun becomes mysteriously unloaded and Fred survives. Technically, this scenario is described by two fluents (a fluent is a condition that can change truth value over time): a l i v e {\displaystyle alive} and l o a d e d {\displaystyle loaded} . Initially, the first condition is true and the second is false. Then, the gun is loaded, some time passes, and the gun is fired. Such problems can be formalized in logic by considering four time points 0 {\displaystyle 0} , 1 {\displaystyle 1} , 2 {\displaystyle 2} , and 3 {\displaystyle 3} , and turning every fluent such as a l i v e {\displaystyle alive} into a predicate a l i v e ( t ) {\displaystyle alive(t)} depending on time. A direct formalization of the statement of the Yale shooting problem in logic is the following one: a l i v e ( 0 ) {\displaystyle alive(0)} ¬ l o a d e d ( 0 ) {\displaystyle \neg loaded(0)} t r u e → l o a d e d ( 1 ) {\displaystyle true\rightarrow loaded(1)} l o a d e d ( 2 ) → ¬ a l i v e ( 3 ) {\displaystyle loaded(2)\rightarrow \neg alive(3)} The first two formulae represent the initial state. The third formula formalizes the effect of loading the gun at time 1 {\displaystyle 1} . The fourth formula formalizes the effect of shooting at Fred at time 2 {\displaystyle 2} . This is a simplified formalization in which action names are neglected and the effects of actions are directly specified for the time points in which the actions are executed. See situation calculus for details. The formulae above, while being direct formalizations of the known facts, do not suffice to correctly characterize the domain. Indeed, ¬ a l i v e ( 1 ) {\displaystyle \neg alive(1)} is consistent with all these formulae, although there is no reason to believe that Fred dies before the gun has been shot. The problem is that the formulae above only include the effects of actions, but do not specify that all fluents not changed by the actions remain the same. In other words, a formula a l i v e ( 0 ) ≡ a l i v e ( 1 ) {\displaystyle alive(0)\equiv alive(1)} must be added to formalize the implicit assumption that loading the gun only changes the value of l o a d e d {\displaystyle loaded} and not the value of a l i v e {\displaystyle alive} . The necessity of a large number of formulae stating the obvious fact that conditions do not change unless an action changes them is known as the frame problem. An early solution to the frame problem was based on minimizing the changes. In other words, the scenario is formalized by the formulae above (that specify only the effects of actions) and by the assumption that the changes in the fluents over time are as minimal as possible. The rationale is that the formulae above enforce all effect of actions to take place, while minimization should restrict the changes to exactly those due to the actions. In the Yale shooting scenario, one possible evaluation of the fluents in which the changes are minimized is the following one. This is the expected solution. It contains two fluent changes: l o a d e d {\displaystyle loaded} becomes true at time 1 and a l i v e {\displaystyle alive} becomes false at time 3. The following evaluation also satisfies all formulae above. In this evaluation, there are still two changes only: l o a d e d {\displaystyle loaded} becomes true at time 1 and false at time 2. As a result, this evaluation is considered a valid description of the evolution of the state, although there is no valid reason to explain l o a d e d {\displaystyle loaded} being false at time 2. The fact that minimization of changes leads to wrong solution is the motivation for the introduction of the Yale shooting problem. While the Yale shooting problem has been considered a severe obstacle to the use of logic for formalizing dynamical scenarios, solutions to it have been known since the late 1980s. One solution involves the use of predicate completion in the specification of actions: in this solution, the fact that shooting causes Fred to die is formalized by the preconditions: alive and loaded, and the effect is that alive changes value (since alive was true before, this corresponds to alive becoming false). By turning this implication into an if and only if statement, the effects of shooting are correctly formalized. (Predicate completion is more complicated when there is more than one implication involved.) A solution proposed by Erik Sandewall was to include a new condition of occlusion, which formalizes the “permission to change” for a fluent. The effect of an action that might change a fluent is therefore that the fluent has the new value, and that the occlusion is made (temporarily) true. What is minimized is not the set of changes, but the set of occlusions being true. Another constraint specifying that no fluent changes unless occlusion is true completes this solution. The Yale shooting scenario is also correctly formalized by the Reiter version of the situation calculus, the fluent calculus, and the action description languages. In 2005, the 1985 paper in which the Yale shooting scenario was first described received the AAAI Classic Paper award. In spite of being a solved problem, that example is still sometimes mentioned in recent research papers, where it is used as an illustrative example (e.g., for explaining the syntax of a new logic for reasoning about actions), rather than being presented as a problem.

    Read more →
  • Comparison of machine learning software

    Comparison of machine learning software

    The following tables are a comparison of machine learning software such as software frameworks, libraries, and computer programs used for machine learning. == Machine learning software == == Other comparisons == == Machine learning helper libraries and platforms == Apache OpenNLP — natural language processing toolkit CUDA — GPU computing platform used to accelerate machine learning and deep learning workloads Horovod — distributed training framework for deep learning Hugging Face Transformers — library of pretrained transformer models built on other machine learning frameworks Kubeflow — machine learning platform for Kubernetes Mallet — toolkit for natural language processing and text analysis NumPy — numerical computing library used in machine learning OpenCV — computer vision library with machine learning functions ONNX — open format for representing machine learning models pandas — data analysis and data preparation library used in machine learning PlaidML — tensor compiler and backend for machine learning frameworks Polars — Dataframe library used for machine learning data preparation and analysis PyArrow — columnar data library used in machine learning data processing ROOT (TMVA) — data analysis framework with machine learning tools SciPy — scientific computing and optimization library used in machine learning == Online development environments for machine learning == Google Colab — hosted Jupyter Notebook environment commonly used for machine learning and deep learning JupyterLab — notebook-based development environment for machine learning and data science Jupyter Notebook — interactive notebook environment used for machine learning and data science Kaggle — online data science and machine learning platform

    Read more →
  • LipNet

    LipNet

    LipNet is a deep neural network for audio-visual speech recognition (ASVR). It was created by University of Oxford researchers Yannis Assael, Brendan Shillingford, Shimon Whiteson, and Nando de Freitas. The researchers stated that could match mouth movements to text with 93 percent accuracy, though it was criticized for its test using a limited dataset of words and grammar. It was used in Nvidia's autonomous "backseat driver" prototype Co-Pilot.

    Read more →
  • Transparency in Frontier Artificial Intelligence Act

    Transparency in Frontier Artificial Intelligence Act

    The Transparency in Frontier Artificial Intelligence Act, also referred to as SB-53, is a 2025 California law which mandates increased transparency for companies building artificial intelligence. SB-53 is primarily focused on assessing and reducing potential catastrophic risks from AI, and is the first bill addressing such risks to be passed into law in America. The bill requires companies to create publicly accessible documents assessing potential "catastrophic risk[s]" from their AI models, as well as publishing documentation on how the model incorporates national and international safety standards. SB-53 also sets up whistleblower protections and procedures for alerting the government to a "critical safety incident". == History == SB-53 was preceded in 2024 by the unsuccessful Safe and Secure Innovation for Frontier Artificial Intelligence Models Act ("SB-1047"), a proposed bill authored by Senator Scott Wiener which was vetoed by Governor Gavin Newsom. Afterwords, Newsom created a "Joint California AI Policy Working Group" to provide recommendations for AI regulation, which guided the drafting of SB-53. Senator Scott Wiener introduced the bill on January 7, 2025, and after a series of amendments, SB-53 passed the Senate 29-8 on September 13. Governor Gavin Newsom approved the bill on September 25, passing it into law. == Provisions == SB-53 applies primarily to companies making at least $500 million in yearly gross revenue. It defines a “frontier model” as any AI trained with over 1026 FLOPS (including fine-tuning), including unreleased internal models. Both the financial and computational thresholds must be met before most of the law is applied, although the threshold can be lowered or otherwise updated by the California Department of Technology in an annual review starting in 2027. Most of the bill's provisions are focused on "catastrophic risks" from AI, which are defined as incidents in which a model contributes to more than 50 deaths or serious injuries, or causes more than one billion dollars ($1,000,000,000) in economic damage from AI-assisted acts (such as cyberattacks or the creation of biological weapons). The bill requires companies to provide publicly accessible safety frameworks for frontier AI models, describing how the company tests for catastrophic risk from its AI, and how it implements protections against such risks. This includes addressing the possibility that the AI may attempt to circumvent internal guardrails or oversight mechanisms. (Certain safety incidents, such as dangerously deceptive model behavior, physical injury, or death, must be reported to California Office of Emergency Services (OES) within 15 days, unless the incident poses imminent physical risk, in which case it must be reported immediately.) The company must follow its published framework, and if any changes are made, the framework should be updated within 30 days, and justification for said changes must also be made public. Additionally, all frontier companies are required to publish basic information about newly released frontier models (such as terms of service, supported languages, and intended use), although only large companies (making over $500 million annually) need to publish full safety frameworks. SB-53 also establishes various whistleblower protections for covered employees. Large companies must have anonymous whistleblowing channels in place which protect employees from retaliation from reporting risks to state or federal authorities if they have reasonable cause to believe that their employer is substantially risking public health and safety.

    Read more →
  • John F. Sowa

    John F. Sowa

    John Florian Sowa (born 1940) is an American computer scientist, an expert in artificial intelligence and computer design, and the inventor of conceptual graphs. == Biography == Sowa received a BS in mathematics from Massachusetts Institute of Technology in 1962, an MA in applied mathematics from Harvard University in 1966, and a PhD in computer science from the Vrije Universiteit Brussel in 1999 with a dissertation titled "Knowledge Representation: Logical, Philosophical, and Computational Foundations". Sowa spent most of his professional career at IBM, starting in 1962 at IBM's applied mathematics group. Over the decades he has researched and developed emerging fields of computer science from compilers, programming languages, and system architecture to artificial intelligence and knowledge representation. In the 1990s Sowa was associated with the IBM Educational Center in New York. Over the years he taught courses at the IBM Systems Research Institute, Binghamton University, Stanford University, the Linguistic Society of America and the Université du Québec à Montréal. He is a fellow of the Association for the Advancement of Artificial Intelligence. After early retirement at IBM, Sowa in 2001 cofounded VivoMind Intelligence, Inc. with Arun K. Majumdar. With this company he was developing data-mining and database technology, more specifically high-level "ontologies" for artificial intelligence and automated natural language understanding. Currently Sowa is working with Kyndi Inc., also founded by Majumdar. John Sowa is married to the philologist Cora Angier Sowa, and they live in Croton-on-Hudson, New York. == Work == Sowa's research interests since the 1970s were in the field of artificial intelligence, expert systems and database query linked to natural languages. In his work he combines ideas from numerous disciplines and eras modern and ancient, for example, applying ideas from Aristotle, the medieval scholastics to Alfred North Whitehead and including database schema theory, and incorporating the model of analogy of Islamic scholar Ibn Taymiyyah in his works. === Conceptual graph === Sowa invented conceptual graphs, a graphic notation for logic and natural language, based on the structures in semantic networks and on the existential graphs of Charles S. Peirce. He introduced the concept in the 1976 article "Conceptual graphs for a data base interface" in the IBM Journal of Research and Development. He elaborated upon it in the 1983 book Conceptual structures: information processing in mind and machine. In the 1980s, this theory had "been adopted by a number of research and development groups throughout the world. International conferences on conceptual structures (ICCS) have been held since 1993, following a series of conceptual graph workshops that began in 1986. === Sowa's law of standards === In 1991, Sowa first stated his Law of Standards: "Whenever a major organization develops a new system as an official standard for X, the primary result is the widespread adoption of some simpler system as a de facto standard for X." Like Gall's law, The Law of Standards is essentially an argument in favour of underspecification. Examples include: The introduction of PL/I resulting in COBOL and FORTRAN becoming the de facto standards for business and scientific programming respectively The introduction of Algol-68 resulting in Pascal becoming the de facto standard for academic programming The introduction of the Ada language resulting in C becoming the de facto standard for US Department of Defense programming The introduction of OS/2 resulting in Windows becoming the de facto standard for desktop OS The introduction of X.400 resulting in SMTP becoming the de facto standard for electronic mail The introduction of X.500 resulting in LDAP becoming the de facto standard for directory services == Publications == 1984. Conceptual Structures - Information Processing in Mind and Machine. The Systems Programming Series, Addison-Wesley 1991. Principles of Semantic Networks. Morgan Kaufmann. Mineau, Guy W; Moulin, Bernard; Sowa, John F, eds. (1993). Conceptual Graphs for Knowledge Representation. LNCS. Vol. 699. doi:10.1007/3-540-56979-0. ISBN 978-3-540-56979-4. S2CID 32275791. 1994. International Conference on Conceptual Structures (2nd : 1994 : College Park, Md.) Conceptual structures, current practices : Second International Conference on Conceptual Structures, ICCS'94, College Park, Maryland, USA, August 16–20, 1994 : proceedings. William M. Tepfenhart, Judith P. Dick, John F. Sowa, eds. Ellis, Gerard; Levinson, Robert; Rich, William; Sowa, John F, eds. (1995). Conceptual Structures: Applications, Implementation and Theory. LNCS. Vol. 954. doi:10.1007/3-540-60161-9. ISBN 978-3-540-60161-6. S2CID 27300281. Lukose, Dickson; Delugach, Harry; Keeler, Mary; Searle, Leroy; Sowa, John, eds. (1997). Conceptual Structures: Fulfilling Peirce's Dream. LNCS. Vol. 1257. doi:10.1007/BFb0027865. ISBN 3-540-63308-1. S2CID 1934069. 2000. Knowledge representation : logical, philosophical, and computational foundations, Brooks Cole Publishing Co., Pacific Grove Articles, a selection Sowa, J. F. (July 1976). "Conceptual Graphs for a Data Base Interface". IBM Journal of Research and Development. 20 (4): 336–357. doi:10.1147/rd.204.0336. Sowa, J. F.; Zachman, J. A. (1992). "Extending and formalizing the framework for information systems architecture". IBM Systems Journal. 31 (3): 590–616. doi:10.1147/sj.313.0590. 1992. "Conceptual Graph Summary"; In: T.E. Nagle et al. (Eds.). Conceptual Structures: Current Research and Practice. Chichester: Ellis Horwood. 1995. "Top-level ontological categories." in: International journal of human-computer studies. Vol. 43, Iss. 5–6, Nov. 1995, pp. 669–685 2006. "Semantic Networks". In: Encyclopedia of Cognitive Science.. John Wiley & Sons.

    Read more →
  • Meesho

    Meesho

    Meesho Limited (short for Meri shop, transl. My shop) is an Indian e-commerce company, headquartered in Bengaluru. Founded by Vidit Aatrey and Sanjeev Barnwal in December 2015, Meesho is an online marketplace in categories such as fashion, home and kitchen, beauty and personal care, electronics accessories, and daily use products. == History == Meesho Private Limited, formerly Fashnear Technologies Private Limited, was established by IIT Delhi graduates Vidit Aatrey and Sanjeev Barnwal in December, 2015 In 2016, the founders came up with the idea of re-establishing the platform as Meesho, one that would enable country-wide shipping for resellers with the use of social media sites as tools for marketing. In February 2019, the platform reported having around 209,000 users and about 1.2 million monthly orders, and in March 2020, it reported approximately 563,000 users and 3.1 million monthly orders. In 2021, the Meesho mobile application was ranked among the most downloaded shopping apps globally. In 2022, Meesho had about 120 million monthly users and about 910 million orders were made through the platform, with a gross merchandise value (GMV) of about $5 billion. According to report as of August 2023 Meesho delisted 42 lakh counterfeit listings and 10 lakh restricted products under its initiative Project Suraksha. During the same period, the platform blocked access for over 12,000 user accounts flagged for policy violations. The Court granted injunctive relief by directing domain registrars to suspend the infringing websites. Additionally, the Court ordered law enforcement authorities to initiate criminal investigations, freeze associated financial accounts against the identified offenders. In 2023, Meesho became the fastest shopping app to cross over 500 million downloads. In 2024, Meesho introduced Valmo, a logistics marketplace, to provide shipment services to sellers by aggregating multiple logistics providers. Meesho employs over 3,000 small businesses and 10-12 large firms for warehousing and sorting operations within its logistics framework. According to media reports, Valmo operating in approximately 15,000 pincodes in India with around 6,000 partners. It is reported to handle over 50% of Meesho's daily orders. In November 2024, Meesho introduced a generative AI-powered voice bot for customer support, managing approximately 60,000 calls daily in English and Hindi. According to media reports, the system resolves the majority of queries without human assistance, with only a small fraction of calls requiring manual intervention. According to media reports, in 2024, Meesho prevented over 22 million suspicious or potentially fraudulent transactions on its platform. The company initiated legal proceedings, resulting in the filing of twelve cases, including nine specifically targeting over forty individuals in the cities of Kolkata and Ranchi. The company filed a suit in the Delhi High Court for a permanent injunction against parties operating deceptive websites misappropriating its brand identity. Meesha went public through an initial public offering in December 2025, raising $603 million. It is listed on both the BSE and NSE. == Recognition == In 2023, Meesho was named one of the most influential companies of the year by Time (magazine).

    Read more →
  • Seeing AI

    Seeing AI

    Seeing AI is an artificial intelligence application developed by Microsoft for iOS. Seeing AI uses the device camera to identify people and objects, and then the app audibly describes those objects for visually impaired people. == Capabilities == Seeing AI is primarily used to describe short text, documents, products, people, currency scenery, colors, handwriting and light. The app can scan a barcode to describe a product and uses sounds to assist the user in focusing on the barcode. When the app describes people, it attempts to estimate the person's age, gender, and emotional status. Additionally, in a test run by German journalists in December 2019, Seeing AI apparently used some sort of facial recognition system to identify people on photographs by name. Some functions are performed on the device, however more complex functions such as describing a scene or recognizing handwriting require an Internet connection. In December 2017, Seeing AI introduced the ability for currency recognition for US and Canadian dollar, British pounds and Euros. In December 2019, Seeing AI added support for five more languages, Dutch, French, German, Japanese, Spanish. Seeing AI is available in 70 countries such as Brazil, Argentina, Australia, Canada, Egypt, Albania, Bhutan, etc. Supported on iPhone 5C, 5S and later best performance with iPhone 6S, SE and later models

    Read more →
  • GermaNet

    GermaNet

    GermaNet is a semantic network for the German language. It relates nouns, verbs, and adjectives semantically by grouping lexical units that express the same concept into synsets and by defining semantic relations between these synsets. GermaNet is free for academic use, after signing a license. GermaNet shares much in common with the English WordNet and can be viewed as an online thesaurus or a light-weight ontology. GermaNet has been developed and maintained at the University of Tübingen since 1997 within the research group for General and Computational Linguistics. It has been integrated into the EuroWordNet, a multilingual lexical-semantic database. == Database == === Contents === GermaNet partitions the lexical space into a set of concepts that are interlinked by semantic relations. A semantic concept is modeled by a synset. A synset is a set of words (called lexical units) where all the words are taken to have the same or almost the same meaning. Thus, a synset is a set of synonyms grouped under one definition, or "gloss". In addition to the gloss, synsets are labeled with their syntactic function and accompanied by example sentences for each distinct meaning in the synset. Just as in WordNet, for each word category the semantic space is divided into a number of semantic fields closely related to major nodes in the semantic network: Ort, or "location", Körper, or "body", etc. As of version 20.0 (release November 2025), GermaNet contains: Synsets: 179438 Lexical units: 231500 Literals: 216517 1.29 lexical units per synset Number of conceptual relations: 194367 Number of lexical relations: 13602 (synonymy excluded) Number of split compounds: 130901 Number of Interlingual Index (ILI) records: 28561 Number of Wiktionary sense descriptions: 29539 === Format === All GermaNet data is stored in a PostgreSQL relational database. The database schema follows the internal structure of GermaNet: there are tables to store synsets, lexical units, conceptual and lexical relations, etc. GermaNet data is distributed both in this database format and as XML files. In the XML data, two types of files, one for synsets and the other for relations, represent all data available in the GermaNet database. == Interfaces == There are software libraries and APIs available for Java and Python. These programs are distributed under free-software licenses and provide easy access to all information in various versions of GermaNet. GermaNet Rover is an on-line application that can be used to search for synsets in GermaNet, explore the data associated with them, and calculate the semantic similarity of pairs of synsets. It features visualizations of the hypernym relation and advanced filtering options for synset searching. == Licenses == GermaNet 20.0 (released November 2025) can be distributed under one of the following types of license agreements: Academic Research License Agreement: for the purpose of research at academic institutions. There is no license fee for academic use. Licenses are not given to individual students, and those seeking a license are required to talk to an academic advisor. Research and Development License Agreement: applies to non-academic institutions and research consortia. To be used strictly for technology development and internal research. Commercial License Agreement: applies to non-academic institutions and commercial enterprises. It permits technology development and internal research, as well as giving the non-exclusive right to distribute and market any derived product or service. == Alternatives == Open-de-WordNet is a freely available alternative to GermaNet which is compatible with WordNet. == Linguistic applications == GermaNet has been used for a variety of applications, including: semantic analysis shallow recognition of implicit document structure compound analysis analyzing sectional preferences word sense disambiguation

    Read more →
  • OpenAI Operator

    OpenAI Operator

    OpenAI Operator was an AI agent developed by OpenAI, capable of autonomously performing tasks through web browser interactions, including filling forms, placing online orders, scheduling appointments, and other repetitive browser-based tasks. It uses OpenAI's advanced models to expand practical automation capabilities for users in daily activities. Operator was launched on January 23, 2025. It was released as a limited-access research preview to ChatGPT Pro-tier subscribers in the United States on February 1, 2025, with future plans to broaden availability. Operator was deprecated after the release of ChatGPT agent, and shut down on August 31, 2025. == Performance and limitations == In benchmark assessments, Operator achieved notable success, scoring 38.1% on OSWorld benchmarks (OS-level tasks) and 58.1% on WebArena benchmarks (web interactions). However, it did not reach human-level accuracy and faced limitations with intricate user interfaces and extended workflows. == Safety and privacy == OpenAI emphasized privacy and safety measures within Operator, including stringent data protection protocols and built-in safety checks designed to prevent unauthorized sensitive actions or information misuse. == Availability == Initially, Operator was only available to ChatGPT Pro subscribers in the U.S., with plans for broader availability to Plus, Team, and Enterprise users in the future.

    Read more →
  • Path tracing

    Path tracing

    Path tracing is a rendering algorithm in computer graphics that simulates how light interacts with objects and participating media to generate realistic (physically plausible) images. It is based on earlier, more limited, ray tracing algorithms. Path tracing is used to create photorealistic images for artistic purposes, and for applications such as architectural rendering and product design. It is also used to render frames for animated films, and visual effects for film and television. Because it can be very accurate and unbiased, it is commonly used to generate reference images when testing the quality of other rendering algorithms. The technique uses the Monte Carlo method to compute estimates of global illumination and simulate the ways different materials reflect (or scatter), transmit, absorb, and emit light. It can incorporate simple modeling of the effects of aperture and lens (depth of field, and bokeh) and shutter speed (motion blur), or more realistic simulation of the optical components in a camera. The algorithm works by describing illumination in a scene using the rendering equation, or light transport equation, and finding an approximate solution using Monte Carlo integration. An inefficient (but accurate) version of the algorithm can be very simple, and involves tracing a ray from the camera, allowing this ray to bounce in random directions as it hits different objects in the scene, and computing the amount of light transmitted along the path to the camera whenever the path encounters a light source. This process is repeated many times for each pixel (each repetition, with generated path and transmitted light, is called a sample), and the results are averaged. One main difference between this algorithm and standard ray tracing is that a single unbranching path is traced each time, while "Whitted-style" or "Cook-style" ray tracing recursively samples branching paths (e.g. when light is both reflected and refracted by a glass object). More practical versions incorporate improvements such as quasi-Monte Carlo methods (techniques that distribute samples more evenly), importance sampling (take more samples of paths that are likely to transport more light), and next event estimation (allow a very limited form of branching, and sample additional paths that connect to the lights more directly). Because path tracing uses random samples there is noise in the final image, which decreases as more samples are taken. Images commonly require many thousands of samples per pixel (spp) to reduce noise to an acceptable level, and denoising techniques (e.g. based on neural networks) are often used. Denoising is usually necessary when path tracing is used for real-time rendering in video games, because relatively few samples can be taken. Many alternative algorithms for path tracing have been developed, although they do not always outperform more straightforward implementations. These include bidirectional path tracing (which traces paths forwards from the light source as well as backwards from the camera), Metropolis light transport, and ways of combining path tracing with photon mapping. Video games often use biased versions of path tracing to improve performance (e.g. limiting the number of bounces in each path). A family of techniques called ReSTIR has been developed that can help real-time path tracing by sharing data between nearby pixels and consecutive frames. == History == Like all ray tracing methods, path tracing is based on ray casting, which Arthur Appel used for computer graphics rendering in the late 1960s. In 1980, John Turner Whitted published a recursive ray tracing algorithm that allows rendering images of scenes containing mirrored surfaces and refractive transparent objects. In 1984, Cook et al. described a form of ray tracing called distributed ray tracing, which uses Monte Carlo integration to render effects such as depth of field, motion blur, reflection from rough surfaces, and area lights. The same year, the radiosity method (not a ray tracing method) was published, which was the first physically based method for rendering diffuse global illumination. In 1986, Jim Kajiya published a paper exploring how to use distributed ray tracing to render physically-based global illumination, and this paper also introduced and named the method called "path tracing". Path tracing and other distributed ray tracing techniques were further refined in the late 1980s and early 1990s by researchers such as James Arvo and Peter Shirley, and by Greg Ward in the open source Radiance software. Despite being theoretically able to render any lighting, the original form of path tracing can sometimes be very inefficient (or noisy) for rendering light that is reflected or refracted before illuminating a visible surface, including diffuse global illumination where light enters an area through narrow gaps, because it traces paths only from the camera. To address this, variations of path tracing that trace paths from both the camera and from light sources, called bidirectional path tracing, were published in 1993 by Eric Lafortune and Yves Willems, and in 1997 by Eric Veach and Leonidas Guibas. In 1997 Veach and Guibas also published an alternative method called Metropolis light transport, which combines bidirectional path tracing with the Metropolis method. Veach's lengthy Ph.D. dissertation described both techniques, along with the theoretical background of path tracing; later, the book Physically Based Rendering (which won an Academy Award for Technical Achievement in 2014) helped to make information about path tracing more widely available. Path tracing requires tracing a large number of paths of light in order to produce an image with a visually acceptable amount of noise. This made path tracing very slow on computers available in the 1980s and 1990s, and noise remained a problem when trying to reproduce the style of earlier computer graphics animated films. Most animated films produced until around 2010, by studios such as Pixar, used rasterization-based rendering, with ray tracing used selectively for reflections (and later for precomputed or cached global illumination). However the speed of computers rapidly increased during the 1990s. Blue Sky Studios pioneered using Monte Carlo ray tracing for global illumination in animation, including in the 1998 short film "Bunny", but they did not disclose the precise techniques used. Path tracing gradually become more practical for film production in the early 2000s. The Arnold renderer, developed by Marcos Fajardo, was used by Sony Pictures Imageworks to produce the feature-length film Monster House, released in 2006. Pixar rewrote their RenderMan software to use path tracing, and released their first feature-length path-traced film Finding Dory in 2016. Although path tracing still had a large computational cost, animation studios discovered that less human labor was required when using it, for example because global illumination no longer needed to be faked by manually placing lights. The amount of noise present in path traced images still caused difficulties, particularly when rendering motion blur (which was used extensively by earlier animated films) but denoising techniques were developed to address this. New techniques were also needed for rendering hair and fur, and to handle the extremely large scenes sometimes required by films. Renderers such as Arnold, and Disney's Hyperion, originally only used CPUs for rendering, but as GPUs became more capable (and APIs such as CUDA, OpenCL, and OptiX were released) researchers and developers began adapting algorithms and implementations to use GPUs. GPUs can dramatically reduce rendering time: for example using a high-end GPU to accelerate portions of the rendering code can make it over 30 times faster than using only a high-end CPU. == Description == Kajiya's 1986 paper defined a recursive integral equation called the rendering equation, which describes a simplified form of light transport. Using Monte Carlo integration for the integral on the right side of the equation leads fairly directly to the path tracing algorithm: I ( x , x ′ ) = g ( x , x ′ ) [ ϵ ( x , x ′ ) + ∫ S ρ ( x , x ′ , x ″ ) I ( x ′ , x ″ ) d x ″ ] {\displaystyle I(x,x')=g(x,x')\left[\epsilon (x,x')+\int _{S}\rho (x,x',x'')I(x',x'')dx''\right]} This expresses I(x,x'), the light arriving at point x from point x', as the product of a geometry term, g(x,x'), which is 0 if there is something blocking the light between the two points and 1 otherwise, and the amount of light leaving point x' and traveling towards x. The light leaving point x' is the sum of the light emitted by the surface at x', and the integral of the light arriving at x' from all other points in the scene (the integration domain S) and being reflected towards x. The factor ρ(x,x',x''), which calculates how much light is reflected, must take into account the angles at which the light is arriving and leaving, and

    Read more →
  • Geoffrey Hinton

    Geoffrey Hinton

    Geoffrey Everest Hinton (born 6 December 1947) is a British-Canadian computer scientist, cognitive scientist, cognitive psychologist and Nobel Prize laureate known for his work on artificial neural networks, which earned him the title "the Godfather of AI". He is University Professor Emeritus at the University of Toronto. From 2013 to 2023, he divided his time working for Google Brain and the University of Toronto before publicly announcing his departure from Google in May 2023, citing concerns about the many risks of artificial intelligence (AI) technology. In 2017, he co-founded and became the chief scientific advisor of the Vector Institute in Toronto. With David Rumelhart and Ronald J. Williams, Hinton was co-author of a highly cited paper published in 1986 that popularised the backpropagation algorithm for training multi-layer neural networks, although they were not the first to propose the approach. Hinton is viewed as a leading figure in the deep learning community. The image-recognition milestone of the AlexNet designed in collaboration with his students Alex Krizhevsky and Ilya Sutskever for the ImageNet challenge 2012 was a breakthrough in the field of computer vision. Hinton received the 2018 Turing Award, together with Yoshua Bengio and Yann LeCun for their work on deep learning. They are sometimes referred to as the "Godfathers of Deep Learning" and have continued to give public talks together. He was also awarded, along with John Hopfield, the 2024 Nobel Prize in Physics for "foundational discoveries and inventions that enable machine learning with artificial neural networks". In May 2023, Hinton announced his resignation from Google to be able to "freely speak out about the risks of AI". He has voiced concerns about deliberate misuse by malicious actors, technological unemployment, and existential risk from artificial general intelligence. He noted that establishing safety guidelines will require cooperation among those competing in use of AI in order to avoid the worst outcomes. After receiving the Nobel Prize, he called for urgent research into AI safety to figure out how to control AI systems smarter than humans. == Education == Hinton was born on 6 December 1947 in Wimbledon in the United Kingdom and was educated at Clifton College in Bristol. In 1967, he matriculated as an undergraduate student at King's College, Cambridge and, after switching between different fields such as natural sciences, history of art, and philosophy, eventually graduated with a Bachelor of Arts in experimental psychology in 1970. He spent a year apprenticing carpentry before returning to academic studies. From 1972 to 1975, he continued his study at the University of Edinburgh, where he was awarded a PhD in artificial intelligence in 1978 for research supervised by Christopher Longuet-Higgins, who favored the symbolic AI approach over the neural network approach. == Career == After his PhD, Hinton initially worked at the University of Sussex and at the MRC Applied Psychology Unit. After having difficulty getting funding in Britain, he worked in the US at the University of California, San Diego, and Carnegie Mellon University. He was the founding director of the Gatsby Charitable Foundation Computational Neuroscience Unit at University College London. He is currently University Professor Emeritus in the Department of Computer Science at the University of Toronto, where he has been affiliated since 1987. Upon arrival in Canada, Geoffrey Hinton was appointed at the Canadian Institute for Advanced Research (CIFAR) in 1987 as a Fellow in CIFAR's first research program, Artificial Intelligence, Robotics & Society. In 2004, Hinton and collaborators successfully proposed the launch of a new program at CIFAR, "Neural Computation and Adaptive Perception" (NCAP), which today is named "Learning in Machines & Brains". Hinton would go on to lead NCAP for ten years. Among the members of the program are Yoshua Bengio and Yann LeCun, with whom Hinton would go on to win the ACM A.M. Turing Award in 2018. All three Turing winners continue to be members of the CIFAR Learning in Machines & Brains program. Hinton taught a free online course on Neural Networks on the education platform Coursera in 2012. He co-founded DNNresearch Inc. in 2012 with his two graduate students, Alex Krizhevsky and Ilya Sutskever, at the University of Toronto's department of computer science. In March 2013, Google acquired DNNresearch Inc. for $44 million, and Hinton planned to "divide his time between his university research and his work at Google". In May 2023, Hinton publicly announced his resignation from Google. He explained his decision, saying he wanted to "freely speak out about the risks of AI" and added that part of him now regrets his life's work. Notable former PhD students and postdoctoral researchers from his group include Peter Dayan, Sam Roweis, Max Welling, Richard Zemel, Brendan Frey, Radford M. Neal, Yee Whye Teh, Ruslan Salakhutdinov, Ilya Sutskever, Yann LeCun, Alex Graves, Zoubin Ghahramani, and Peter Fitzhugh Brown. == Research == Hinton's research concerns the use of neural networks for machine learning, memory, perception, and symbol processing. He has written or co-written more than 200 peer-reviewed publications. In the 1980s, Hinton was part of the "Parallel Distributed Processing" group at Carnegie Mellon University, which included notable scientists like Terrence Sejnowski, Francis Crick, David Rumelhart, and James McClelland. This group favoured the connectionist approach during the AI winter. Their findings were published in a two-volume set. The connectionist approach adopted by Hinton suggests that capabilities in areas like logic and grammar can be encoded into the parameters of neural networks, and that neural networks can learn them from data. Symbolists on the other side advocated for explicitly programming knowledge and rules into AI systems. In 1985, Hinton co-invented Boltzmann machines with David Ackley and Terry Sejnowski. His other contributions to neural network research include distributed representations, time delay neural network, mixtures of experts, Helmholtz machines and product of experts. An accessible introduction to Geoffrey Hinton's research can be found in his articles in Scientific American in September 1992 and October 1993. In 1995, Hinton and colleagues proposed the wake-sleep algorithm, involving a neural network with separate pathways for recognition and generation, being trained with alternating "wake" and "sleep" phases. In 2007, Hinton coauthored an unsupervised learning paper titled Unsupervised learning of image transformations. In 2008, he developed the visualization method t-SNE with Laurens van der Maaten.While Hinton was a postdoc at UC San Diego, David Rumelhart, Hinton and Ronald J. Williams applied the backpropagation algorithm to multi-layer neural networks. Their experiments showed that such networks can learn useful internal representations of data. In a 2018 interview, Hinton said that "David Rumelhart came up with the basic idea of backpropagation, so it's his invention." Although this work was important in popularising backpropagation, it was not the first to suggest the approach. Reverse-mode automatic differentiation, of which backpropagation is a special case, was proposed by Seppo Linnainmaa in 1970, and Paul Werbos proposed to use it to train neural networks in 1974. In 2017, Hinton co-authored two open-access research papers about capsule neural networks, extending the concept of "capsule" introduced by Hinton in 2011. The architecture aims to better model part-whole relationships within objects in visual data. In 2021, Hinton presented GLOM, a speculative architecture idea also aiming to improve image understanding by modeling part-whole relationships in neural networks. In 2021, Hinton co-authored a widely cited paper proposing a framework for contrastive learning in computer vision. The technique involves pulling together representations of augmented versions of the same image, and pushing apart dissimilar representations. At the 2022 Conference on Neural Information Processing Systems (NeurIPS), Hinton introduced a new learning algorithm for neural networks that he calls the "Forward-Forward" algorithm. The idea is to replace the traditional forward-backwards passes of backpropagation with two forward passes, one with positive (i.e. real) data and the other with negative data that could be generated solely by the network. The Forward-Forward algorithm is well-suited for what Hinton calls "mortal computation", where the knowledge learned is not transferable to other systems and thus dies with the hardware, as can be the case for certain analog computers used for machine learning. == Honours and awards == Hinton is a Fellow of the US Association for the Advancement of Artificial Intelligence (FAAAI) since 1990. He was elected a Fellow of the Royal Society of Canada (FRSC) in 1996, and then a

    Read more →
  • AlphaFold

    AlphaFold

    AlphaFold is an artificial intelligence (AI) program developed by DeepMind, a subsidiary of Alphabet, which performs predictions of protein structure. It is designed using deep learning techniques. AlphaFold 1 (2018) placed first in the overall rankings of the 13th Critical Assessment of Structure Prediction (CASP) in December 2018. It was particularly successful at predicting the most accurate structures for targets rated as most difficult by the competition organizers, where no existing template structures were available from proteins with partially similar sequences. AlphaFold 2 (2020) repeated this placement in the CASP14 competition in November 2020. It achieved a level of accuracy much higher than any other entry. It scored above 90 on CASP's global distance test (GDT) for approximately two-thirds of the proteins, a test measuring the similarity between a computationally predicted structure and the experimentally determined structure, where 100 represents a complete match. The inclusion of metagenomic data has improved the quality of the prediction of multiple sequence alignments. One of the biggest sources of the training data was the custom-built Big Fantastic Database of 65,983,866 protein families, represented as multiple sequence alignments and Hidden Markov models, covering 2,204,359,010 protein sequences from reference databases, metagenomes, and metatranscriptomes. AlphaFold 2's results at CASP14 were described as "astounding" and "transformational". However, some researchers noted that the accuracy was insufficient for a third of its predictions, and that it did not reveal the underlying mechanism or rules of protein folding for the protein folding problem, which remains unsolved. Despite this, the technical achievement was widely recognized. On 15 July 2021, the AlphaFold 2 paper was published in Nature as an advance access publication alongside open source software and a searchable database of species proteomes. As of November 2025, the paper had been cited nearly 43,000 times. AlphaFold 3 was announced on 8 May 2024. It can predict the structure of complexes created by proteins with DNA, RNA, various ligands, and ions. The new prediction method shows a minimum 50% improvement in accuracy for protein interactions with other molecules compared to existing methods. Demis Hassabis and John Jumper shared one half of the 2024 Nobel Prize in Chemistry, awarded "for protein structure prediction," while the other half went to David Baker "for computational protein design." Hassabis and Jumper had previously won the Breakthrough Prize in Life Sciences and the Albert Lasker Award for Basic Medical Research in 2023 for their leadership of the AlphaFold project. == Background == Proteins consist of chains of amino acids which spontaneously fold to form the three dimensional (3-D) structures of the proteins. The 3-D structure is crucial to understanding the biological function of the protein. Protein structures can be determined experimentally through techniques such as X-ray crystallography, cryo-electron microscopy and nuclear magnetic resonance (NMR), which are all expensive and time-consuming. Such efforts, using the experimental methods, have identified the structures of about 170,000 proteins over the last 60 years, while there are over 200 million known proteins across all life forms. Over the years, researchers have applied numerous computational methods to predict the 3D structures of proteins from their amino acid sequences, accuracy of such methods in best possible scenario is close to experimental techniques (NMR) by the use of homology modeling based on molecular evolution. CASP, which was launched in 1994 to challenge the scientific community to produce their best protein structure predictions, found that GDT scores of only about 40 out of 100 can be achieved for the most difficult proteins by 2016. AlphaFold started competing in the 2018 CASP using an artificial intelligence (AI) deep learning technique. == Algorithm == DeepMind is known to have trained the program on over 170,000 protein structures from the Protein Data Bank, a public repository of protein sequences and structures. The program uses a form of attention network, a deep learning technique that focuses on having the AI identify parts of a larger problem, then piece it together to obtain the overall solution. The overall training was conducted on processing power between 100 and 200 GPUs. === AlphaFold 1 (2018) === AlphaFold 1 (2018) was built on work developed by various teams in the 2010s, work that looked at the large databases of related protein sequences now available from many different organisms (most without known 3D structures), to try to find changes at different residues (peptides) that appeared to be correlated, even though the residues were not consecutive in the main chain. Such correlations suggest that the residues may be close to each other physically, even though not close in the sequence, allowing a contact map to be estimated. Building on recent work prior to 2018, AlphaFold 1 extended this by estimating a probability distribution for the distances between residues, effectively transforming the contact map into a distance map. It also used more advanced learning methods than previously to develop the inference. The code was not made publicly available, except to run on sequences of proteins in the 2018 CASP competition. === AlphaFold 2 (2020) === The 2020 version of the program (AlphaFold 2, 2020) is significantly different from the original version that won CASP 13 in 2018, according to the team at DeepMind. AlphaFold 1 used a number of separately trained modules to produce a guide potential, which was then combined with a physics-based energy potential. AlphaFold 2 replaced this with a system of interconnected sub-networks, forming a single, differentiable, end-to-end model based on pattern recognition. This model was trained in an integrated manner. After the neural network's prediction converges, a final refinement step applies local physical constraints using energy minimization based on the AMBER force field. This step only slightly adjusts the predicted structure. A key part of the 2020 system are two modules, believed to be based on a transformer design, which are used to progressively refine a vector of information for each relationship (or "edge" in graph-theory terminology) between an amino acid residue of the protein and another amino acid residue (these relationships are represented by the array shown in green); and between each amino acid position and each different sequences in the input sequence alignment (these relationships are represented by the array shown in red). Internally these refinement transformations contain layers that have the effect of bringing relevant data together and filtering out irrelevant data (the "attention mechanism") for these relationships, in a context-dependent way, learned from training data. These transformations are iterated, the updated information output by one step becoming the input of the next, with the sharpened residue/residue information feeding into the update of the residue/sequence information, and then the improved residue/sequence information feeding into the update of the residue/residue information. As the iteration progresses, according to one report, the "attention algorithm ... mimics the way a person might assemble a jigsaw puzzle: first connecting pieces in small clumps—in this case clusters of amino acids—and then searching for ways to join the clumps in a larger whole." The output of these iterations then informs the final structure prediction module, which also uses transformers, and is itself then iterated. In an example presented by DeepMind, the structure prediction module achieved a correct topology for the target protein on its first iteration, scored as having a GDT_TS of 78, but with a large number (90%) of stereochemical violations – i.e. unphysical bond angles or lengths. With subsequent iterations the number of stereochemical violations fell. By the third iteration the GDT_TS of the prediction was approaching 90, and by the eighth iteration the number of stereochemical violations was approaching zero. The training data was originally restricted to single peptide chains. However, the October 2021 update, named AlphaFold-Multimer, included protein complexes in its training data. DeepMind stated this update succeeded about 70% of the time at accurately predicting protein-protein interactions. === AlphaFold 3 (2024) === Announced on 8 May 2024, AlphaFold 3 was co-developed by Google DeepMind and Isomorphic Labs, both subsidiaries of Alphabet. AlphaFold 3 is not limited to proteins, as it can also predict the structures of protein complexes with DNA, RNA, post-translational modifications and selected ligands and ions. AlphaFold 3 introduces the "Pairformer," a deep learning architecture inspired by the transformer, which is considered similar to, but si

    Read more →