Search

The revelation comes just months after Microsoft's competitor Apple unveiled its own AI strategy named Apple Intelligence. [ABSTRACT FROM AUTHOR]

The exciting AI-driven image enhancement feature will only be available on Copilot+ PCs. [ABSTRACT FROM AUTHOR]

Adobe's AI-based video generator be used to extend existing videos or create completely new ones. [ABSTRACT FROM AUTHOR]

The Raspberry Pi AI Camera works with all Raspberry Pi models and is now available for purchase. [ABSTRACT FROM AUTHOR]

OpenAI is preparing to launch its new flagship AI model, Orion, which is expected to be significantly more powerful than the GPT-4. Microsoft, a major investor in OpenAI, is reportedly ready to introduce Orion on Azure as early as November. Unlike previous models, Orion will not initially be released through the AI assistant Chat GPT, but will first be provided to select companies for product development. The exact name of the new model, whether GPT-5 or something else, remains uncertain, as both OpenAI and Microsoft have not confirmed details. [Extracted from the article]

Anyone can now sign up for Midjourney and create AI-generated images without paying, but there's a catch. [ABSTRACT FROM AUTHOR]

OpenAI, the developer of ChatGPT, is working on a new kind of reasoning AI model called "Strawberry." This model aims to generate responses based on instructions and also has the potential to plan ahead and conduct independent research on the internet. The details of how Strawberry works and its completion timeline are currently unknown. OpenAI continues to research new AI opportunities, but no specific information about Strawberry has been provided. [Extracted from the article]

The UN Secretary-General's Advisory Body on Artificial Intelligence has released a report titled "Governing AI for Humanity," which outlines a plan to manage the risks associated with AI and promote its potential worldwide. The report emphasizes the importance of including underrepresented groups in discussions about AI and proposes seven recommendations to improve current AI governance. It calls for international cooperation to establish a globally inclusive and distributed architecture for AI governance and urges governments and stakeholders to work together to protect human rights in the development and use of AI. [Extracted from the article]

French AI startup Mistral has released its first multimodal model, the Pixtral 12B, which can handle both text and images. The model uses 12 billion parameters and is based on Mistral's Nemo 12B text model. It can answer questions about images via URLs or images encoded with base64. The image data used to develop the Pixtral 12B is unclear. The model can be downloaded and used under an Apache 2.0 license without restrictions. [Extracted from the article]

Salesforce, a cloud-based customer relationship management (CRM) software company, is considering implementing a new pricing model that would charge customers for each AI chat. The CEO, Marc Benioff, envisions introducing 1 billion AI agents into the user environment by 2026 through the Agentforce AI platform, with an estimated cost of around $2 per chat. This pricing model aims to reflect the value of Salesforce's AI services and provide customers with more flexibility. [Extracted from the article]

Google's Gemini AI model, which allows users to generate images of people, is back after being temporarily shut down due to controversy. The previous version of the feature was criticized for producing historically inaccurate images, particularly related to skin color. The updated Gemini 1.5 model aims to address these issues and also cannot generate images of famous individuals. Google has emphasized its commitment to improving AI security and responsibility to ensure ethical use of the technology. [Extracted from the article]

A new study from Arize AI reveals that 56% of the largest companies in the United States consider AI to be a "risk factor," a significant increase from previous years. This sentiment is shared by 90% of media and entertainment companies and 86% of software and technology firms. The risks associated with AI include increased competition, ethical concerns, and potential impacts on human rights, employment, and privacy. Additionally, when discussing generative AI, only 33% of companies viewed it as an opportunity, while two-thirds considered it risky. [Extracted from the article]

Meta and Universal Music Group have signed a new agreement that allows Meta's users to share songs from Universal's music library on Meta-owned platforms without violating copyright. The deal also addresses the issue of unauthorized AI-generated content, which refers to music that is scraped by AI systems without the creator's consent. Universal Music Group's chief digital officer and executive vice president, Michael Nash, expressed the company's commitment to protecting the rights of artists and songwriters. [Extracted from the article]

Nvidia, a technology company, reportedly collected large amounts of data from YouTube without permission to train its artificial intelligence (AI) models. The data was used to train various AI models, including one for automated driving, a human-like AI avatar, and a tool for building 3D worlds. This data collection raises ethical and legal concerns, as YouTube's terms of use prohibit the use of its data without permission. Nvidia is already facing legal action for alleged copyright violations. [Extracted from the article]

OpenAI has developed a tool to watermark text generated by its AI assistant, ChatGPT, but has chosen not to use it due to internal disagreements. The company conducted a survey which revealed that four times as many people prefer tools that can identify AI content rather than watermarking. Additionally, 30% of ChatGPT users stated that they would use the service less if it contained watermarked content. The quality of the content generated by ChatGPT is not expected to be affected by the implementation of watermarking. [Extracted from the article]

AI startup Suno has confirmed that its AI model was trained on copyrighted material from major music companies Universal Music Group, Warner Music Group, and Sony Music Group. Suno argues that this training should be classified as fair use under US law because it creates a new end product. However, the music companies disagree, stating that the use of copyrighted material without consent is not fair use and constitutes unfair competition. [Extracted from the article]

Apple has released a new collection of AI functions called Apple Intelligence, which will enhance Siri and generate text and images. These features are currently available in the developer betas for IOS 18.1 and iPad OS 18.1, with a public release expected later this year. Interestingly, Apple has reportedly used Google's chips, rather than Nvidia's, to train two of its central AI models. This decision is notable because Nvidia is known for producing highly sought-after AI processors. Apple has not yet commented on this development. [Extracted from the article]

Google's AI unit, Deepmind, has developed two new AI systems, AlphaProof and AlphaGeometry 2, that can solve complex mathematical problems in a manner similar to human intelligence. These systems have achieved a significant milestone in AI development by successfully solving four out of six questions in the International Mathematical Olympiad (IMO). One question was solved in just a few minutes, while others took up to three days. AlphaProof even managed to solve the competition's most difficult problem, which only five out of 600 human participants were able to solve. This advancement in AI technology demonstrates the potential for AI to reason and solve abstract mathematical problems. [Extracted from the article]

Regulatory authorities in the EU, UK, and US have signed a joint statement to ensure effective competition in the AI sector. They acknowledge the rapid development of generative AI (genAI) and the potential for new ways to compete, innovate, and catalyze opportunities. The parties pledge to work together under their respective laws to maintain a competitive AI market and ensure fair treatment for consumers and businesses. This includes preventing exclusionary tactics and closely scrutinizing investments and collaborations between established AI companies and emerging players in the market. [Extracted from the article]

Major technology companies, including Apple and Nvidia, have reportedly used data from YouTube videos to train their artificial intelligence (AI) tools. The data collection, created by EleutherAI, includes transcripts from various news and entertainment channels, such as Khan Academy, BBC, and PewDiePie. However, harvesting material from YouTube without permission is against the platform's rules. EleutherAI has not yet commented on the review conducted by Proof News. [Extracted from the article]

OpenAI has developed an evaluation scale to assess how closely AI models can approach human levels of intelligence. The scale consists of five levels, with level one representing basic intelligence and level five representing artificial general intelligence (AGI), where an AI system can perform the work of entire organizations. Currently, today's AI models are judged to be at level one. OpenAI's evaluation scale is considered preliminary and may be adjusted in the future. [Extracted from the article]