This article's tone or style may not reflect the encyclopedic tone used on Wikipedia.(May 2022) |
Company type | Division |
---|---|
Industry | Artificial intelligence |
Founded | December 11, 2015 |
Founders | |
Headquarters | Astor Place, New York City, New York, U.S. |
Products | LLaMA |
Owner | Meta Platforms |
Website | ai |
This article is part of a series about |
Meta Platforms |
---|
Products and services |
People |
Business |
Part of a series on |
Artificial intelligence |
---|
Meta AI is an American company owned by Meta (formerly Facebook). The company develops artificial intelligence and augmented and artificial reality technologies. Meta AI deems itself an academic research laboratory, focused on generating knowledge for the AI community, and should not be confused with Meta's Applied Machine Learning (AML) team, which focuses on the practical applications of its products.
The laboratory was founded as Facebook Artificial Intelligence Research (FAIR) with locations at the headquarters in Menlo Park, California, London, United Kingdom, and a new laboratory in Manhattan. FAIR was officially announced in September 2013. [1] FAIR was first directed by New York University's Yann LeCun, a deep learning professor and Turing Award winner. [2] Working with NYU's Center for Data Science, FAIR's initial goal was to research data science, machine learning, and artificial intelligence and to "understand intelligence, to discover its fundamental principles, and to make machines significantly more intelligent". [3] Research at FAIR pioneered the technology that led to face recognition, tagging in photographs, and personalized feed recommendation. [4] Vladimir Vapnik, a pioneer in statistical learning, joined FAIR [5] in 2014. Vapnik is the co-inventor of the support-vector machine and one of the developers of the Vapnik–Chervonenkis theory.
FAIR opened a research center in Paris, France in 2015, [6] and subsequently launched smaller satellite research labs in Seattle, Pittsburgh, Tel Aviv, Montreal and London. [7] In 2016, FAIR partnered with Google, Amazon, IBM, and Microsoft in creating the Partnership on Artificial Intelligence to Benefit People and Society, an organization with a focus on open licensed research, supporting ethical and efficient research practices, and discussing fairness, inclusivity, and transparency.
In 2018, Jérôme Pesenti, former CTO of IBM's big data group, assumed the role of president of FAIR, while LeCun stepped down to serve as chief AI scientist. [8] In 2018, FAIR was placed 25th in the AI Research Rankings 2019, which ranked the top global organizations leading AI research. [9] FAIR quickly rose to eighth position in 2019, [10] and maintained eighth position in the 2020 rank. [11] FAIR had approximately 200 staff in 2018, and had the goal to double that number by 2020. [12]
FAIR's initial work included research in learning-model enabled memory networks, self-supervised learning and generative adversarial networks, text classification and translation, as well as computer vision. [3] FAIR released Torch deep-learning modules as well as PyTorch in 2017, an open-source machine learning framework, [3] which was subsequently used in several deep learning technologies, such as Tesla's autopilot [13] and Uber's Pyro. [14] Also in 2017, FAIR discontinued a research project once AI bots developed a language that was unintelligible to humans, [15] inciting conversations about dystopian fear of artificial intelligence going out of control. [16] However, FAIR clarified that the research had been shut down because they had accomplished their initial goal to understand how languages are generated, rather than out of fear. [15]
FAIR was renamed Meta AI following the rebranding that changed Facebook, Inc. to Meta Platforms Inc. [17]
In 2022, Meta AI predicted the 3D shape of 600 million potential proteins in two weeks. [18]
In February 23, 2022, live event Inside the Lab: Building for the Metaverse with AI, the Meta AI team discussed the major advancements in research and development in artificial intelligence. [19] One such tool is the BuilderBot, which allows users to generate virtual worlds by using voice commands. Other tools include the No Language Left Behind, a system capable of automatic translation between written languages, and a Universal Speech Translator, a system capable of instantaneous speech-to-speech translation.
Meta AI's computer vision research aims to extract information about the environment from digital images and videos. [20] One example of computer vision technology developed by AI is panoptic segmentation, which recognizes objects in the foreground but also classifies the scenes in the background. [21] Meta AI seeks to improve Visual Question Answering technology, in which a machine answers human user questions about images using cycle-consistency, having the machine generate a question in addition to the answer to address linguistic variations in the questions. [22]
Artificial intelligence communication requires a machine to understand natural language and to generate language that is natural. Meta AI seeks to improve these technologies to improve safe communication regardless of what language the user might speak. [23] Thus, a central task involves the generalization of natural language processing (NLP) technology to other languages. As such, Meta AI actively works on unsupervised machine translation. [24] [25] Meta AI seeks to improve natural-language interfaces by developing aspects of chitchat dialogue such as repetition, specificity, response-relatedness and question-asking, [26] incorporating personality into image captioning, [27] and generating creativity-based language. [28]
In 2018, Meta AI launched the open-source PyText, a modeling framework focused on NLP systems. [29]
In February 2023, Meta AI launched LLaMA (Large Language Model Meta AI), a large language model ranging from 7B to 65B parameters. [30] Subsequently, Meta AI released LLaMA 2 in July 2023 [31] and LLaMA 3 in April 2024. [32]
Facebook and Instagram use Meta AI research in ranking & recommendations in their newsfeeds, ads, and search results. [33] Meta AI has also introduced ReAgent, a toolset that generates decisions and evaluates user feedback. [34]
Machine learning and AI depend on the development of novel algorithms, software, and hardware technologies. As such, Meta AI's systems research teams study computer languages, compilers, and hardware applications. [35]
Meta AI studies the mathematical and theoretical foundations of artificial intelligence. Meta AI has publications in learning theory, optimization, and signal processing. [36]
The MTIA v1 is Meta's first-generation AI training and inference accelerator, developed specifically for Meta's recommendation workloads. It was fabricated using TSMC's 7 nm process technology and operates at a frequency of 800 MHz. In terms of processing power, the accelerator provides 102.4 TOPS at INT8 precision and 51.2 TFLOPS at FP16 precision, while maintaining a thermal design power (TDP) of 25 W. [37] [38] [39]
The accelerator is structured around a grid of 64 processing elements (PEs), arranged in an 8x8 configuration, and it is furnished with on-chip and off-chip memory resources along with the necessary interconnects. Each PE houses two processor cores (one with a vector extension) and several fixed-function units optimized for tasks such as matrix multiplication, accumulation, data movement, and nonlinear function calculation. The processor cores utilize the RISC-V open instruction set architecture (ISA), with extensive customization to perform the required compute and control tasks.
The accelerator's memory subsystem uses LPDDR5 for off-chip DRAM resources and can be scaled up to 128 GB. Additionally, it possesses 128 MB of on-chip SRAM that is shared amongst all the PEs for faster access to frequently used data and instructions. The design encourages parallelism and data reuse, offering thread and data-level parallelism (TLP and DLP), instruction-level parallelism (ILP), and memory-level parallelism (MLP).
MTIA accelerators are mounted on compact dual M.2 boards, enabling easier integration into a server. The boards connect to the host CPU via PCIe Gen4 x8 links and have a power consumption as low as 35 W. The servers hosting these accelerators utilize the Yosemite V3 server specification from the Open Compute Project. Each server houses 12 accelerators, interconnected through a hierarchy of PCIe switches, allowing workloads to be distributed across multiple accelerators and executed concurrently.
MTIA v2 is Meta's second-generation AI training and inference accelerator, significantly enhancing performance and efficiency for AI workloads, particularly in recommendation and ranking models. Fabricated with TSMC's 5 nm technology, it operates at 1.35 GHz and provides 708 TOPS at INT8 precision (with sparsity) and 354 TFLOPS at FP16 precision, representing substantial improvements over MTIA v1. [40]
Key architectural enhancements include an 8x8 grid of processing elements (PEs), increased local PE storage (384 KB per PE), on-chip SRAM (256 MB), and off-chip LPDDR5 memory (128 GB). Memory bandwidth improvements are also significant, with local memory at 1 TB/s per PE, on-chip memory at 2.7 TB/s, and off-chip LPDDR5 at 204.8 GB/s.
MTIA v2 features an improved network on chip (NoC) architecture for low-latency coordination between PEs. The system supports up to 72 accelerators in a rack-based setup, using PCIe Gen5 links for enhanced bandwidth and scalability.
The software stack, fully integrated with PyTorch 2.0, includes the Triton-MTIA compiler backend for high-performance kernel optimization, improving developer productivity. Early results show a 3x performance improvement over MTIA v1, with a 6x increase in model serving throughput and a 1.5x improvement in performance per watt.
Feature | MTIA v1 | MTIA v2 |
---|---|---|
Process node | TSMC 7nm | TSMC 5nm |
Frequency | 800MHz | 1.35GHz |
Instances | 1.12B gates, 65M flops | 2.35B gates, 103M flops |
Die size | 19.34mm x 19.1mm, 373mm² | 25.6mm x 16.4mm, 421mm² |
Package | 43mm x 43mm | 50mm x 40mm |
Voltage | 0.67V logic, 0.75V memory | 0.85V |
TDP | 25W | 90W |
Host Connection | 8x PCIe Gen4 (16 GB/s) | 8x PCIe Gen5 (32 GB/s) |
GEMM TOPS | 102.4 TFLOPS/s (INT8) | 708 TFLOPS/s (INT8) (sparsity) |
51.2 TFLOPS/s (FP16/BF16) | 354 TFLOPS/s (INT8) | |
354 TFLOPS/s (FP16/BF16) (sparsity) | ||
177 TFLOPS/s (FP16/BF16) | ||
SIMD TOPS | Vector core: | Vector core: |
3.2 TFLOPS/s (INT8) | 11.06 TFLOPS/s (INT8) | |
1.6 TFLOPS/s (FP16/BF16) | 5.53 TFLOPS/s (FP16/BF16) | |
0.8 TFLOPS/s (FP32) | 2.76 TFLOPS/s (FP32) | |
SIMD: | SIMD: | |
3.2 TFLOPS/s (INT8/FP16/BF16) | 5.53 TFLOPS/s (INT8/FP16/BF16) | |
1.6 TFLOPS/s (FP32) | 2.76 TFLOPS/s (FP32) | |
Memory Capacity | Local memory: 128 KB per PE | Local memory: 384 KB per PE |
On-chip memory: 128 MB | On-chip memory: 256 MB | |
Off-chip LPDDR5: 64 GB | Off-chip LPDDR5: 128 GB | |
Memory Bandwidth | Local memory: 400 GB/s per PE | Local memory: 1 TB/s per PE |
On-chip memory: 800 GB/s | On-chip memory: 2.7 TB/s | |
Off-chip LPDDR5: 176 GB/s | Off-chip LPDDR5: 204.8 GB/s |
User controls
Meta AI offers options for users to customize their interaction with its features. Users are able to mute the AI chatbot on platforms like Facebook, Instagram, and WhatsApp [41] , temporarily halting notifications from the chatbot. Some platforms also offer the ability to hide certain AI elements from their interface. To locate the relevant settings, users can consult the platform's help documentation or settings menu.
Concerns
Since May 2024, the Meta AI chatbot has summarized news from various outlets without linking directly to original articles, including in Canada, where news links are banned on its platforms. This use of news content without compensation has raised ethical and legal concerns, especially as Meta continues to reduce news visibility on its platforms. [42]
Artificial intelligence (AI), in its broadest sense, is intelligence exhibited by machines, particularly computer systems. It is a field of research in computer science that develops and studies methods and software that enable machines to perceive their environment and uses learning and intelligence to take actions that maximize their chances of achieving defined goals. Such machines may be called AIs.
Vladimir Naumovich Vapnik is a computer scientist, researcher, and academic. He is one of the main developers of the Vapnik–Chervonenkis theory of statistical learning and the co-inventor of the support-vector machine method and support-vector clustering algorithms.
Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thus perform tasks without explicit instructions. Recently, artificial neural networks have been able to surpass many previous approaches in performance.
In artificial intelligence, symbolic artificial intelligence is the term for the collection of all methods in artificial intelligence research that are based on high-level symbolic (human-readable) representations of problems, logic and search. Symbolic AI used tools such as logic programming, production rules, semantic nets and frames, and it developed applications such as knowledge-based systems, symbolic mathematics, automated theorem provers, ontologies, the semantic web, and automated planning and scheduling systems. The Symbolic AI paradigm led to seminal ideas in search, symbolic programming languages, agents, multi-agent systems, the semantic web, and the strengths and limitations of formal knowledge and reasoning systems.
Léon Bottou is a researcher best known for his work in machine learning and data compression. His work presents stochastic gradient descent as a fundamental learning algorithm. He is also one of the main creators of the DjVu image compression technology, and the maintainer of DjVuLibre, the open source implementation of DjVu. He is the original developer of the Lush programming language.
Yann André LeCun is a French-American computer scientist working primarily in the fields of machine learning, computer vision, mobile robotics and computational neuroscience. He is the Silver Professor of the Courant Institute of Mathematical Sciences at New York University and Vice-President, Chief AI Scientist at Meta.
A cognitive computer is a computer that hardwires artificial intelligence and machine learning algorithms into an integrated circuit that closely reproduces the behavior of the human brain. It generally adopts a neuromorphic engineering approach. Synonyms include neuromorphic chip and cognitive chip.
An AI accelerator, deep learning processor, or neural processing unit (NPU) is a class of specialized hardware accelerator or computer system designed to accelerate artificial intelligence and machine learning applications, including artificial neural networks and machine vision. Typical applications include algorithms for robotics, Internet of Things, and other data-intensive or sensor-driven tasks. They are often manycore designs and generally focus on low-precision arithmetic, novel dataflow architectures or in-memory computing capability. As of 2024, a typical AI integrated circuit chip contains tens of billions of MOSFETs.
Explainable AI (XAI), often overlapping with Interpretable AI, or Explainable Machine Learning (XML), either refers to an artificial intelligence (AI) system over which it is possible for humans to retain intellectual oversight, or refers to the methods to achieve this. The main focus is usually on the reasoning behind the decisions or predictions made by the AI which are made more understandable and transparent. XAI counters the "black box" tendency of machine learning, where even the AI's designers cannot explain why it arrived at a specific decision.
Graphcore Limited is a British semiconductor company that develops accelerators for AI and machine learning. It has introduced a massively parallel Intelligence Processing Unit (IPU) that holds the complete machine learning model inside the processor.
The Open Neural Network Exchange (ONNX) [] is an open-source artificial intelligence ecosystem of technology companies and research organizations that establish open standards for representing machine learning algorithms and software tools to promote innovation and collaboration in the AI sector. ONNX is available on GitHub.
Mila - Quebec AI Institute is a research institute in Montreal, Quebec, focusing mainly on machine learning research. Approximately 1000 students and researchers and 100 faculty members, were part of Mila in 2022. Along with Alberta's Amii and Toronto's Vector Institute, Mila is part of the Pan-Canadian Artificial Intelligence Strategy.
Karen Hao is an American journalist and data scientist. Currently a contributing writer for The Atlantic and previously a foreign correspondent based in Hong Kong for The Wall Street Journal and senior artificial intelligence editor at the MIT Technology Review, she is best known for her coverage on AI research, technology ethics and the social impact of AI. Hao also co-produces the podcast In Machines We Trust and writes the newsletter The Algorithm.
Specialized computer hardware is often used to execute artificial intelligence (AI) programs faster, and with less energy, such as Lisp machines, neuromorphic engineering, event cameras, and physical neural networks. As of 2023, the market for AI hardware is dominated by GPUs.
BrainChip is an Australia-based technology company, founded in 2004 by Peter Van Der Made, that specializes in developing advanced artificial intelligence (AI) and machine learning (ML) hardware. The company's primary products are the MetaTF development environment, which allows the training and deployment of spiking neural networks (SNN), and the AKD1000 neuromorphic processor, a hardware implementation of their spiking neural network system. BrainChip's technology is based on a neuromorphic computing architecture, which attempts to mimic the way the human brain works. The company is a part of Intel Foundry Services and Arm AI partnership.
Generative artificial intelligence is artificial intelligence capable of generating text, images, videos, or other data using generative models, often in response to prompts. Generative AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics.
Llama is a family of autoregressive large language models released by Meta AI starting in February 2023. The latest version is Llama 3 released in April 2024.
Open-source artificial intelligence is the application of open-source practices to the development of artificial intelligence resources.
Michael Karl Gschwind is an American computer scientist who currently is a director and principal engineer at Meta Platforms in Menlo Park, California. He is recognized for his seminal contributions to the design and exploitation of general-purpose programmable accelerators, as an early advocate of sustainability in computer design and as a prolific inventor.