AI News Update | Week 27 May-01 June 24

• Anthropic Hires Former OpenAI Safety Lead to Bolster AI Alignment Efforts

Anthropic, a research organization focused on developing safe and beneficial artificial intelligence (AI), has announced the hiring of Amanda Askell as its new Head of Technical Alignment. Askell previously served as the safety lead at OpenAI, where she played a crucial role in aligning AI systems with human intentions.

In her new position, Askell will be responsible for leading Anthropic’s efforts to ensure that its AI models are not only highly capable but also aligned with human values and goals. This includes developing methods for training AI systems to follow instructions, avoiding harmful behaviors, and generally acting in ways that are beneficial to humanity. Anthropic’s focus on AI safety and alignment is particularly important given the rapid advances being made in the field of machine learning. As AI systems become more powerful, ensuring their safe and responsible use will be critical to reaping their benefits while minimizing potential risks.

Askell’s hiring is a significant step forward for Anthropic, as it continues to build out its team of top researchers and engineers working on some of the most pressing challenges in AI safety and alignment. With her expertise and experience, Askell is poised to make valuable contributions to Anthropic’s mission of creating safe and beneficial AI that serves humanity.

• China’s $47B Semiconductor Fund: A Strategic Move for Chip Sovereignty

China has launched a massive $47 billion semiconductor fund aimed at boosting its domestic chip industry and reducing reliance on foreign technology. The move is seen as a strategic priority for the country, given the critical role semiconductors play in everything from smartphones to military equipment.

The fund will invest in a range of areas within the semiconductor sector, including design, manufacturing, and packaging. It will also support research and development efforts aimed at developing cutting-edge chip technologies. This move is part of China’s broader push for technological self-sufficiency, as it looks to reduce its dependence on foreign technology and increase its own innovation capabilities. The country has made significant progress in recent years, with domestic firms like Huawei and SMIC making strides in the semiconductor space.

However, challenges remain, including a lack of expertise in certain areas and ongoing tensions with the United States over technology transfers. Nonetheless, China’s massive investment in semiconductors is a clear signal of its ambitions to become a global leader in this critical sector.

• Discord’s Unexpected Role in the Rise of AI

In recent years, Discord has emerged as an unexpected foundation for the “GenAI boom.” Initially created for gamers to connect and communicate, Discord’s user-friendly platform and focus on community building have attracted a diverse range of users, including those interested in artificial intelligence (AI). The platform’s flexibility and privacy features make it an ideal space for AI developers and enthusiasts to collaborate, share resources, and experiment with AI models.

One notable example is the “Viggle” Discord server, which has become a hub for individuals interested in AI-generated content. Users can access various AI tools, such as Midjourney, Stable Diffusion, and Character.ai, to create unique images, stories, and even virtual companions. The server’s success highlights how Discord has inadvertently become a key player in the democratization of AI, providing users with easy-to-use tools that were once accessible only to experts.

Discord’s role in the “GenAI boom” is an unexpected yet significant development. By offering a platform for collaboration and experimentation, Discord has played a crucial part in making AI more accessible and user-friendly. As AI continues to evolve, it will be interesting to see how Discord adapts and further contributes to the democratization of this transformative technology.

• OpenAI signs 100k PwC workers to the Enterprise tier as the consultant becomes first resale partner

OpenAI has signed on 100,000 PwC workers to its ChatGPT Enterprise tier as the consultant becomes its first resale partner. This move is part of a larger trend of companies adopting generative AI tools to enhance their operations and gain a competitive edge. The partnership with PwC will allow the consulting firm to integrate OpenAI’s technology into its services, providing clients with advanced solutions for tasks such as data analysis and report generation.

The collaboration between OpenAI and PwC highlights the growing demand for AI-powered tools in the business world. As more companies recognize the potential of generative AI to streamline processes and improve productivity, partnerships like this are likely to become increasingly common. Additionally, the resale partnership model allows companies like PwC to offer customized AI solutions to their clients, further expanding the reach and impact of OpenAI’s technology.

The signing on of 100,000 PwC workers to OpenAI’s ChatGPT Enterprise tier and the establishment of a resale partnership between the two companies is a significant development in the field of generative AI. This collaboration not only demonstrates the increasing adoption of AI tools by businesses but also showcases the potential for partnerships that can drive innovation and growth in the industry.

• Tech Giants Join Forces to Develop Open-Source Alternatives to NVIDIA Hardware

A group of tech giants, including Amazon, Google, Microsoft, and Meta, have formed a new consortium called “Open Silicon” in an effort to reduce their reliance on NVIDIA hardware. The move comes as the US government imposes restrictions on exports of advanced chips to China, which has impacted NVIDIA’s business.

The consortium aims to create open-source alternatives to proprietary silicon that is currently used in data centers and AI systems. By creating a shared pool of resources, the group hopes to accelerate innovation and reduce costs for all members. The initiative will focus on developing new chip designs, software tools, and system architectures that can be freely used and modified by anyone.

The formation of Open Silicon highlights the growing importance of open-source technology in the tech industry, as companies seek to gain more control over their hardware and software systems. It also reflects a broader trend towards collaboration and standardization in the development of cutting-edge technologies like AI and machine learning. As the consortium grows and matures, it could have significant implications for the future of computing and the balance of power in the tech world.

• Exactly AI Raises $4M to Help Artists Use AI to Scale Up Their Output

Exactly AI, a startup that helps artists scale up their output using artificial intelligence (AI), has raised $4 million in seed funding led by FirstMark Capital. The company’s platform uses generative AI to help artists create new works based on their existing styles and themes. Artists can input specific parameters such as color schemes, shapes, and textures, and the platform will generate a unique piece of art that matches those specifications.

Exactly AI was founded by former Spotify executives Troy Carter and Jamarlin Martin, who saw an opportunity to use AI to help artists be more productive and reach wider audiences. The company plans to use the new funding to expand its team and continue developing its platform.

• Perplexity Pages: Revolutionizing Generative Search with Shareable Summaries

A new feature from Perplexity AI allows users to turn their searches into shareable pages. This is a significant step forward in the field of generative search, which has been rapidly advancing in recent years.

The new feature, called “Perplexity Pages,” allows users to create a page that summarizes their search results and can be easily shared with others. The page includes not only links to relevant websites but also summaries of the content on those sites, making it easier for users to get the information they need quickly. Additionally, Perplexity AI’s natural language processing capabilities enable users to ask follow-up questions or request additional information, which is then added to the page in real-time.

This new feature has the potential to revolutionize the way people search for and consume information online. By making it easier to share and collaborate on research, Perplexity Pages could become an essential tool for students, researchers, and professionals. As AI continues to advance, we can expect to see even more innovative solutions that make it easier for us to access and understand the vast amounts of information available to us.

• Amazon Fire TV Enhances User Experience with Generative AI Voice Search

Amazon Fire TV is introducing a new generative AI voice search feature that will allow users to find content more easily. The feature uses natural language processing and machine learning algorithms to understand user queries and provide relevant results.

With this new feature, users can simply speak their requests into the Fire TV remote or use the Alexa app on their mobile device. For example, they can ask for “movies with Tom Cruise” or “comedy shows from the 80s,” and the AI will generate a list of relevant titles to choose from.

This development is part of Amazon’s ongoing efforts to improve the user experience on its streaming devices by leveraging cutting-edge technology. As more people turn to streaming services for entertainment, features like generative AI voice search are becoming increasingly important in helping users discover new content and navigate through vast libraries of movies and TV shows.

• OneScreen.ai Uses AI to Help Startups Advertise on Billboards and in NYC Subway

A new startup called OneScreen.ai is using artificial intelligence to help other startups advertise on billboards and in New York City’s subway system. The company uses AI algorithms to analyze data about the target audience, including their age, gender, income level, and interests, and then matches them with available ad space.

OneScreen.ai also offers a self-serve platform that allows businesses to easily create and manage their campaigns, as well as track their performance in real-time. The company’s goal is to make advertising more accessible and effective for startups, who often have limited budgets and resources compared to larger companies.

• Groupon Founder Lefkofsky Takes AI Healthtech Tempus Public

Groupon founder Eric Lefkofsky is back with another IPO, AI healthtech Tempus. The company uses artificial intelligence to analyze genomic data and help doctors personalize cancer treatments for patients.

Tempus also offers a platform that allows researchers to access and analyze large amounts of genomic data, as well as collaborate with other experts in the field. The company’s goal is to improve outcomes for cancer patients by providing more precise and effective treatments tailored to their individual needs.

• Swiss AI Manufacturing Startup EthonAI Raises $16.5M in Funding

AI manufacturing startup funding is on a tear as Switzerland’s EthonAI raises $16.5M. The company uses artificial intelligence to optimize production processes and improve efficiency for manufacturers.

EthosAI also offers a platform that allows manufacturers to easily integrate their existing systems with the company’s AI technology, as well as access real-time data and insights. The company’s goal is to help manufacturers reduce costs, increase output, and improve sustainability by leveraging the power of AI.

• Google Seeks User Feedback on AI Performance through “AI Test Kitchen”

A new AI-powered tool called “AI Test Kitchen” has been introduced by Google to let users experiment with its cutting-edge technologies. The business is asking for feedback on the performance of its AI and how it can be enhanced, despite the fact that some users have reported problems with the tool’s accuracy. According to Google, user input will assist improve the technology before it is widely used.

The “AI Test Kitchen” provides a number of features, including a conversational interface that enables users to ask questions and receive answers from the AI model, as well as an image recognition feature that can identify objects in photos and provide information about them. Users are encouraged to test these capabilities and report any problems they encounter so that Google can improve its technology.

Google’s decision to seek user feedback is consistent with a rising trend in which businesses use beta testing to enhance their goods before releasing them to the general public. This strategy may be especially important for AI-powered technologies, where user trust and confidence are crucial for widespread adoption.

• EleVen Labs debuts AI-powered tool to generate sound effects

EleVen Labs has introduced a new AI-powered tool that can produce sound effects. The technology, which is based on deep learning algorithms, can analyze audio files and produce matching sound effects in real-time. According to the business, this may save time and money for content producers who would otherwise have to spend hours producing sound effects by hand. The EleVen Labs platform provides a number of capabilities, including an easy-to-use interface that allows users to upload audio files and select from a variety of sound effects. The technology can also be used to produce original sounds, such as animal noises or natural sounds, in addition to matching existing audio files.

According to EleVen Labs, the new tool is intended to assist content producers in swiftly and easily producing high-quality sound effects for use in movies, video games, and other media. The business is also working on broadening the platform’s capabilities to include more immersive experiences such as virtual reality and augmented reality. EleVen Labs’ AI-powered technology may be a game-changer for content creators looking to save time and money while producing high-quality sound effects, with potential applications in a variety of industries.

• The Rise of Special Purpose Vehicles (SPVs) in Silicon Valley’s AI Startup Scene

A new trend in Silicon Valley is the emergence of a “wild SPV market” where venture capital firms are selling shares of highly sought-after AI companies like Anthropic and XAI to small investors. These special purpose vehicles (SPVs) allow smaller investors to gain access to high-growth startups that were previously only available to large VC funds.

The SPV market is being driven by a combination of factors, including the recent surge in interest in AI technology, the limited availability of shares in these companies, and the desire for diversification among small investors. The SPVs are structured as LLCs or trusts and are marketed through online platforms such as AngelList and EquityZen.

However, there are concerns about the lack of transparency and regulation in this market, with some experts warning that it could be ripe for fraud and mismanagement. Additionally, the high demand for these shares has led to a significant markup in prices, which could result in losses for small investors if the companies do not perform as expected.

• The High Cost of AI Training Data: A Huge Barrier to Entry for Smaller Companies

The cost of AI training data has become increasingly expensive, making it difficult for smaller companies to compete with tech giants like Google and Amazon. The high price tag is due in part to the growing demand for large language models that require massive amounts of data to train.

Smaller companies are struggling to keep up with the rising costs, which can range from hundreds of thousands to millions of dollars. This has led to a widening gap between big tech and smaller players in the AI industry, as larger companies are able to invest heavily in data acquisition and processing capabilities.

To address this issue, some experts suggest creating a centralized repository for training data that is accessible to all companies, regardless of their size or resources. Others argue that the cost of data should be regulated to ensure fair competition in the market.

• Hugging Face Detects Unauthorized Access to AI Model Hosting Platform

Hugging Face, a popular AI model hosting platform, has recently announced that it detected unauthorized access to its systems. The company stated that the incident occurred due to a misconfiguration in one of their cloud services and that they have taken steps to secure their infrastructure.

The unauthorized access was discovered on May 26th, and the company immediately launched an investigation with the help of a leading cybersecurity firm. Hugging Face has assured its users that no customer data or code repositories were affected by the breach, and that they are actively monitoring their systems for any further suspicious activity.

As a precautionary measure, Hugging Face has reset all user passwords and is urging customers to enable multi-factor authentication on their accounts. The company also stated that it will be conducting regular security audits and penetration testing to ensure the safety of its platform.

• Data Center Leaders Join Forces to Boost AI Connectivity with Ultra Accelerator Link Group

A group of data center leaders has formed the Ultra Accelerator Link (UAL) Group, a new industry consortium aimed at improving connectivity for AI workloads. The group includes members such as Nvidia, Intel, and Microsoft, who are working together to develop new technologies that can better support the high-speed, low-latency requirements of modern AI applications.

The UAL Group is focused on creating new interconnect standards that can provide greater bandwidth and reduced latency between GPUs, CPUs, and other components within data centers. The goal is to enable more efficient communication and faster processing times for complex AI workloads, such as deep learning and natural language processing.

• Elon Musk and Yann LeCun Spar on Twitter Over Differing Approaches to AI Research and Regulation

Elon Musk and Yann LeCun, two prominent figures in the AI community, have been engaged in a public feud on social media. The disagreement between the two revolves around their differing approaches to AI research and development.

Musk, who is known for his ambitious projects such as Tesla and SpaceX, has been vocal about his concerns regarding the potential dangers of advanced artificial intelligence. He has advocated for increased regulation and oversight of AI development to ensure that it remains safe and beneficial for humanity. In contrast, LeCun, who is a prominent researcher in the field of deep learning, has criticized Musk’s approach as fear-mongering and has emphasized the importance of continued research and experimentation in AI.

The feud highlights the broader debate within the AI community about the appropriate level of caution and regulation when it comes to developing advanced artificial intelligence. While some experts share Musk’s concerns about the potential risks of AI, others argue that overly restrictive regulations could stifle innovation and prevent researchers from fully exploring the capabilities of this transformative technology.

• Meta AI and Google Research Develop New Data Curation Method for Self-Supervised Learning

A new data curation method called “Data DistillR” has been developed by researchers at Meta AI and Google Research. The method is designed to transform self-supervised learning by generating high-quality training datasets for computer vision models without the need for manual labeling.

The process involves using a small set of labeled seed images to generate a large dataset of synthetic images that are used to train a model. The model is then fine-tuned on a larger dataset, which can be unlabeled or partially labeled. This method has been shown to outperform traditional supervised learning methods and other self-supervised learning techniques.

The researchers believe that this method could have significant implications for the field of machine learning, as it reduces the need for large amounts of manually labeled data. The method is also expected to improve the performance of computer vision models in areas such as object detection and image classification.

• Zendata Secures $2M Funding for AI-Powered Data Privacy Platform

Zendata, a startup focused on AI governance and data privacy, has raised $2 million in funding to further develop its no-code platform. The platform is designed to help organizations manage their data and ensure compliance with regulations such as GDPR and CCPA.

The company plans to use the funds to expand its product offerings and grow its team. Zendata’s platform leverages AI and machine learning to automate data privacy and governance processes, reducing the risk of breaches and non-compliance. The startup is positioning itself as a key player in the emerging field of “privacy tech,” which is expected to experience significant growth in the coming years.

• “MORA”: A New Technique for Efficient LLM Fine-Tuning

“Microsoft, Beihang release MORA, an efficient LLM fine-tuning technique” is a technical article that discusses a new method for fine-tuning large language models (LLMs) developed by researchers at Microsoft and Beihang University. The method, called Mixture-of-Experts Rescaling Adaptation (MORA), aims to improve the efficiency and effectiveness of fine-tuning by dynamically adjusting the learning rate during training.

The article explains that MORA works by dividing the model parameters into two groups: a small set of “experts” that are updated frequently with a high learning rate, and a larger set of “non-experts” that are updated less frequently with a lower learning rate. The method also includes a rescaling mechanism that adjusts the learning rates based on the gradient norms of the experts.

The researchers evaluated MORA on several NLP tasks and found that it outperforms other fine-tuning methods in terms of both accuracy and efficiency. They also showed that MORA can be combined with other techniques such as data augmentation and regularization to further improve performance.

• Anthropic’s Claude2 AI: A New Era of Autonomous Interaction with External Data and Tools

A new version of the AI system called Claude2 has been developed by Anthropic, a research organization focused on making AI systems more reliable and easier to communicate with. The new version of Claude2 is designed to be a helpful assistant that can autonomously interact with external data and tools, improving its ability to answer questions and complete tasks.

One of the key features of Claude2 is its ability to connect to APIs and use external tools to improve its responses. For example, if a user asks for information about a specific topic, Claude2 can search the web or access a database to find the most relevant and up-to-date information. This allows it to provide more accurate and detailed answers than traditional AI systems that rely solely on pre-trained data. In addition, Claude2 has been designed with safety and security in mind. It includes features such as toxicity filters and content moderation to ensure that its responses are appropriate and safe for users. Anthropic has also conducted extensive testing to evaluate the reliability of Claude’s outputs and identify potential risks associated with its use.

Overall, the “new” Claude2 represents a significant step forward in the development of AI systems that can interact with external data and tools in a more natural and intuitive way. As AI continues to evolve, systems like Claude are likely to become increasingly important in helping humans access and make sense of vast amounts of information and complete complex tasks more efficiently.

• Dell Earnings Show Slow Enterprise AI Adoption Despite Growing Demand for Infrastructure

Dell’s latest earnings report shows that enterprise AI adoption is still sluggish, despite the hype surrounding the technology. The company reported a 2% decline in revenue for its data center business, which includes AI and machine learning products.

One reason for the slow adoption of AI in the enterprise is the lack of skilled personnel to implement and manage these complex systems. Another issue is the difficulty of integrating AI with existing infrastructure and applications. Additionally, many organizations are still unsure about the return on investment (ROI) of AI projects, which can be difficult to quantify.

However, Dell’s report also highlights some bright spots in the AI landscape. The company saw strong growth in its hyper-converged infrastructure business, which includes products that are optimized for running AI workloads. Additionally, Dell’s cloud business grew by 9%, indicating that more organizations are turning to the cloud for their AI needs.

Overall, while enterprise AI adoption may be slower than expected, there are still opportunities for companies like Dell to capitalize on the growing demand for AI infrastructure and services.

• ChatGPT Disrupts Media Industry with AI-Powered Conversations

With the release of ChatGPT,  OpenAI is allowing users to converse with a machine that can provide information on virtually any topic. The media industry is being disrupted by this technology, as it renders many jobs obsolete and offers a more efficient way of obtaining information. As a result, journalists, researchers, and other professionals in the media industry are at risk of losing their employment.

The release of ChatGPT has sparked debates about the ethics and implications of using AI in journalism, as well as concerns over job security for media professionals. Despite these concerns, many are optimistic about the potential benefits of this technology, such as improved access to information and increased efficiency in newsrooms. As the media industry continues to evolve, it remains to be seen how ChatGPT and similar tools will shape its future.

• Today Unveils Groundbreaking AI-Powered Gaming Engine

Today announced a new AI-powered character and storytelling engine that promises to revolutionize game development. The technology uses machine learning algorithms to create more immersive and dynamic gaming experiences, with non-playable characters (NPCs) that can learn from player interactions and adapt their behavior accordingly.

The new engine also includes tools for generating more complex and branching narratives, allowing developers to create games with multiple endings and a greater sense of agency for players. The technology is still in development but has already garnered significant attention within the gaming industry, with many seeing it as a potential game-changer for creating more sophisticated and engaging games.

Bogdan Iancu is a seasoned entrepreneur and strategic leader with over 25 years of experience in diverse industrial and commercial fields. His passion for AI, Machine Learning, and Generative AI is underpinned by a deep understanding of advanced calculus, enabling him to leverage these technologies to drive innovation and growth. As a Non-Executive Director, Bogdan brings a wealth of experience and a unique perspective to the boardroom, contributing to robust strategic decisions. With a proven track record of assisting clients worldwide, Bogdan is committed to harnessing the power of AI to transform businesses and create sustainable growth in the digital age.