Browsing: AI

META Unveils CM3leon A Versatile and Efficient AI Generative Model

Interest in generative AI models has surged, driven by advancements in natural language processing and image generation.Interest in generative AI models has surged, driven by advancements in naturallanguage processing and image generation. META, a prominent player in the AI researchdomain, has introduced CM3leon, a cutting-edge multimodal model. Multimodal meansthe AI is capable of both text-to-image and image-to-text generation.CM3leon’s unique approach combines a recipe derived from text-only language models.Meta’s model will employ large-scale retrieval-augmented pre-training and multitasksupervised fine-tuning stages. Better Performance in Image GenerationDespite being trained with five times fewer computational resources than previoustransformer-based methods, CM3leon achieves state-of-the-art performance in text-to-image generation. Notably, it exhibits the versatility of autoregressive models whilemaintaining low training costs and efficient inference. This tokenization-based model goes beyond conventional text-to-image approaches. Itcan generate complex sequences of text and images conditioned on arbitrary content.Unlike other specialized image generation models, CM3leon’s large-scale multitaskinstruction tuning significantly enhances performance across various vision-languagetasks, such as image caption generation and visual question answering. Ethical Image Data SourcingMeta announced that it takes an ethical approach to image data sourcing, using onlylicensed images from Shutterstock to avoid issues related to ownership and attribution.This socially responsible methodology sets CM3leon apart from its competitors.In a comparison with widely-used benchmarks, CM3leon achieves an impressive FIDscore of 4.88, outperforming Google’s Parti model and setting a new standard for text-to-image generation. A Frechet Inception Distance (FID) score of 0.0 indicates a perfectscore.CM3leon exhibits an ability to generate intricate compositional objects, evident inexamples like a potted cactus donning sunglasses and a hat.…

China Tightens Rules for Generative AI Platforms

Chinese authorities are taking steps to regulate the artificial intelligence (AI) industry, aiming to ensure consumer safety. Chinese authorities are taking steps to regulate the artificial intelligence (AI) industry,aiming to ensure consumer safety. The Cyberspace Administration of China (CAC) hasproposed additional requirements for firms launching generative AI systems, includingthe need for licensing. If approved, the new licensing regime could be introduced by theend of the month, with the CAC acting as the regulatory watchdog.These measures build upon the Chinese government’s earlier plan, which allowed a 10-day window for AI product registration with relevant authorities. The new proposal aimsto ensure that AI products align with the government’s objectives. Data Sources Required to be Cited According to the draft rules, all AI-generated content should not subvert state power,advocate for the overthrow of the socialist system, incite the division of the country, orundermine national unity.To combat misinformation, the draft rules also emphasise the importance of clearlabelling for AI-generated content. Developers are urged to take precautions to ensurecompliance with existing laws against discrimination and copyright regulations whenusing generative AI platforms. The legislation also includes a mandatory securityreview: developers are required to submit relevant information to authorities, includingdata used for training AI models.China has demonstrated a strong commitment to policing AI, as evidenced by therestrictions placed on OpenAI’s ChatGPT due to violations of laws and regulations. InMay, Chinese authorities arrested an individual for using ChatGPT to spread fake news,potentially resulting in a ten-year prison sentence.…

OpenAI’s GPT-4 Faces Criticism

OpenAI, the creator of ChatGPT, is facing significant backlash from industry experts who are disappointed with the latest GPT-4 model.OpenAI, the creator of ChatGPT, is facing significant backlash from industry experts whoare disappointed with the latest GPT-4 model.Once praised for its speed, the recent version of the AI model has been widely criticisedfor its perceived weakness, laziness, and lack of intelligence. Despite OpenAI’s initialclaims, users have expressed their frustration with the model, deeming it subparcompared to its predecessors.Many users have taken to Twitter and online forums to voice their concerns about GPT-4’s underwhelming performance. They report instances of incorrect responses, anapparent loss of track of input data, and difficulties in following instructions. Even thesimple task of including brackets in software code has proven to be a challenge for theAI model. Essentially, GPT-4 appears to be merely regurgitating recently generated…

Hollywood Unites in AI Strike

Leaders of Hollywood’s actors’ union, the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA)Leaders of Hollywood’s actors’ union, the Screen Actors Guild-American Federation ofTelevision and Radio Artists (SAG-AFTRA), joined forces with screenwriters in anunprecedented joint strike. The result is a widespread production shutdown across theentertainment industry.The strike was triggered after negotiations for a new contract with major studios andstreaming services reached a stalemate. Producers defended their stance, noting thatthey compensate actors and writers fairly; also adding a landmark move to protect theindustry from artificial intelligence.The First in More than 60 YearsThis marks the first time in over six decades that two major Hollywood unions havesimultaneously gone on strike. Recent protests harken back to the 1960 protests led byRonald Reagan, who later became the US president. The decision to strike was met withfervor, as actors’ union president Fran Drescher, known for her role in “The Nanny,”passionately criticized industry executives for prioritizing Wall Street and greed overrecognizing the invaluable contributions of the workforce.The strike prompted picketing screenwriters outside Netflix’s Hollywood offices,chanting “Pay Your Actors!” in solidarity. Actors will join the picket lines alongsidewriters outside studio headquarters in New York and Los Angeles starting on Friday.Notably, the strike’s announcement disrupted the premiere schedule of ChristopherNolan’s film “Oppenheimer,” as cast members including Cillian Murphy, Emily Blunt, andMatt Damon left the event in support of the strike.75th Emmy Awards AffectedThis industry-wide strike, the first involving film and television actors since 1980, casts ashadow over the upcoming 75th Emmy Awards. Union rules prevent actors fromparticipating in interviews, promotions, or even attending the ceremony. The strikeguidelines also prohibit actors from making personal appearances, promoting theirwork through podcasts or premieres, and engaging in any production-related activities,including auditions, rehearsals, and voiceovers.While international shoots may technically continue, the impact of the strike on US-based writers and performers is expected to have a ripple effect throughout theindustry. Disney CEO Bob Iger expressed concern over the strike’s potential “verydamaging effect” on the entire entertainment landscape. Iger highlighted the existingdisruptions caused by streaming platforms and artificial intelligence.Differing ViewsThe contract negotiations between SAG-AFTRA and the Alliance of Motion Picture andTelevision Producers, representing industry giants such as Disney, Netflix, and Amazon,proved contentious. The union accused studios of undercutting modest-but-essentialincome streams, including long-term residuals. The guild also chastised companies forneglecting inflationary adjustments and failing to address the challenges posed by AI.The Alliance of Motion Picture and Television Producers defended their position, statingthat they presented a generous deal that included significant increases in minimum pay,higher pension and health contributions, and a groundbreaking AI proposal to safeguardactors’ digital likenesses. They expressed regret that the union had chosen the path ofstrike, which they believe will lead to financial hardships for numerous industryprofessionals.SAG-AFTRA, representing more than 160,000 screen actors, broadcast journalists,announcers, hosts, and stunt performers, received solidarity from Broadway actors,emphasizing their shared concerns.Actors Joining in SolidarityIt is important to note that the striking actors, including the famous faces that dominatethe industry, are not seeking financial gains for themselves, as their individual contractsnegotiated by agents often exceed the minimums being contested in the strike. Instead,the unions’ primary focus is on fairer profit sharing, addressing the challenges of thestreaming era, and protecting their creative work from unauthorized AI usage.As Hollywood braces for this significant industry disruption, the outcomes of the strikewill have far-reaching implications for the future of the entertainment landscape and therelationships between unions, studios, and streaming platforms.

Google Unveils NotebookLM: An AI-Driven Note-Taking Assistant

Google has recently introduced NotebookLM, an AI-powered note-taking app designedto enhance learning efficiency.Google has recently introduced NotebookLM, an AI-powered note-taking app designedto enhance learning efficiency. Initially known as Project Tailwind, this experimentaloffering from Google Labs aims to reimagine note-taking software by incorporating apowerful language model at its core. NotebookLM is now being rolled out gradually to aselect group of users in the United States, with continuous improvements planned forthe future. Users from other countries can sign up for the waitlist.In the face of information overload, extracting meaningful insights has become adaunting task. Google acknowledges this challenge and seeks to facilitate the synthesisof facts and ideas from multiple sources.NotebookLM addresses this issue by leveraging language models and your existingcontent to provide valuable insights promptly. Acting as a virtual research assistant, itcan summarise information, untangle complex concepts, and foster creativeconnections, all based on the sources you choose.What sets NotebookLM apart from conventional AI chatbots is its ability to “ground” thelanguage model in your personal notes and selected sources.By doing so, NotebookLM becomes well-versed in the information that matters most toyou. Currently, users can ground the AI in specific Google Docs, and additional formatswill be supported in the future.Summary, Questioning, and Idea GenerationOnce you’ve selected your Google Docs, NotebookLM offers three primary functions:1. Summary Generation: Upon adding a Google Doc to NotebookLM, it automaticallygenerates a summary along with key topics and questions to enhance yourcomprehension of the material.2. Questioning Capability: Users can delve deeper into their uploaded documents byasking questions. For instance, a medical student could upload a scientific article onneuroscience and request NotebookLM to create a glossary of key terms related todopamine.Similarly, an author working on a biography could upload research notes and ask for asummary of all interactions between two historical figures.3. Idea Generation: NotebookLM is built with the ability to assist users in generatingfresh ideas. For instance, a content creator can upload ideas for new videos and requestthe AI to generate a script on a specific topic. Likewise, an entrepreneur preparing asales pitch can inquire about potential investor questions.Fact-checking Still a MustWhile NotebookLM’s source-grounding minimises the risk of model “hallucinations,” it isessential to fact-check the AI’s responses against the source material. Google ensureseasy fact-checking by providing citations with each response, featuring the mostrelevant quotes from your selected sources.Ultimately, the development team has two primary goals: actively involving users in theproduct’s development and ensuring responsible deployment of the technology.Feedback from users will be instrumental in making NotebookLM a genuinely valuabletool. Google also upholds a strict set of safety criteria aligned with its AI Principles andimplements necessary safeguards before expanding the user base and introducing newfunctionalities.Importantly, NotebookLM only has access to the source material you upload, and yourfiles and interactions with the AI remain private and inaccessible to other users. Googledeclared that the data collected is not used to train new AI models.As NotebookLM is in its initial evolution phase, Google is committed to actively seekinguser feedback to ensure its usefulness and effectiveness.

China’s Baichuan Intelligence Unveils Open Source Language Model to Compete with OpenAI

Baichuan Intelligence, a startup founded by Wang Xiaochuan, the founder of Sogou, hasintroduced its next-generation large language model Baichuan-13B.Baichuan Intelligence, a startup founded by Wang Xiaochuan, the founder of Sogou, hasintroduced its next-generation large language model Baichuan-13B. Wang, a computer scienceprodigy from Tsinghua University, aims to establish China’s version of OpenAI. Baichuan isconsidered one of China’s most promising developers in the field of large language models(LLMs). The model, based on the Transformer architecture like OpenAI’s GPT, has 13 billionparameters and is trained on Chinese and English data. Baichuan-13B is open source andoptimised for commercial applications.Training Data Comparable to GPT 3.5Baichuan-13B is trained on 1.4 trillion tokens, surpassing Meta’s LLaMa, which uses 1 trilliontokens in its 13 billion-parameter model. Wang has expressed his intention to release a large-scale model comparable to OpenAI’s GPT-3.5 by the end of this year. Within a short period,Baichuan has made significant progress, expanding its team to 50 people by the end of Apriland launching its first LLM, Baichuan-7B, in June.​Baichuan-13B is now available for free to approved academics and developers who wish to useit for commercial purposes. Notably, the model offers variations that can run on consumer-grade hardware, addressing the constraints posed by U.S. AI chip sanctions on China.Baichuan-7B is an open-source, large-scale pre-training language model, meticulously crafted bythe visionary minds at Baichuan Intelligent Technology. Rooted in the architecture of theTransformer model, this model harnesses a staggering 7 billion parameters and has beennourished with the exposure to a staggering 1.2 trillion tokens. With its unwavering versatility,Baichuan-7B gracefully accommodates both the Chinese and English languages.High Performance Scores Across the BoardDuly celebrated as a front-runner amongst models of similar scale, Baichuan-7B has emergedvictorious in renowned Chinese and English benchmarks, including the esteemed C-EVAL andMMLU assessments, etching its name at the pinnacle of linguistic excellence.This model consistently surpasses its counterparts of similar parameter magnitude, reigningsupreme as the pre-eminent native pre-trained model in the realm of Chinese languagecomprehension. In the AGIEval assessment, Baichuan-7B outshines other open-sourcecontenders, including LLaMA-7B, Falcon-7B, Bloom-7B, and ChatGLM-6B, by an astonishingmargin, securing an impressive score of 34.4 points.Baichuan-7B conquers the C-EVAL examination with a commanding score of 42.8 points,outshining ChatGLM-6B’s 38.9 points. In the Gaokao evaluation, the model reigns supreme withan exceptional score of 36.2 points, firmly establishing its dominance among pre-trainedmodels of comparable parameter scale.AGIEval, a celebrated benchmark initiative by Microsoft Research Institute, represents anexhaustive endeavour to assess the cognitive and problem-solving capacities of fundamentalmodels. C-Eval, a collaborative creation by Shanghai Jiao Tong University, Tsinghua University,and the University of Edinburgh, serves as a comprehensive examination evaluating the prowessof Chinese language models, encompassing 52 diverse subjects across various industries.The Gaokao benchmark, crafted by the esteemed research team at Fudan University, leveragesthe Chinese college entrance examination questions as a dataset, offering a rigorousexamination of large models’ aptitude in Chinese language comprehension and logicalreasoning.Baichuan-7B’s mastery extends effortlessly into the realm of English. In the esteemed MMLUassessment, baichuan-7B astounds with an extraordinary score of 42.5 points, effortlesslysurpassing the English open-source pre-trained model, LLaMA-7B, and the Chinese open-sourcemodel, ChatGLM-6B, by significant margins.A key determinant of success in large-scale model training resides within the training corpusitself. Baichuan Intelligent Technology diligently constructs a high-quality pre-training corpus,drawing from rich Chinese learning data and seamlessly integrating high-quality English data.This data amalgamation encompasses a vast array of Chinese and English internet data, open-source Chinese and English data, alongside a substantial corpus of meticulously curatedknowledge.