The ChatGPT Observer #26

If you're short on time, here's your one-minute summary of this week's key AI news and advancements:

dvancements:

?? Top Story: OpenAI's "Project Strawberry" and next-gen AI model "Orion" promise groundbreaking capabilities, potentially succeeding GPT-4.

?? This Week's Spotlight: California passes AI Bill SB 1047, setting new standards for AI safety and accountability in the US.

?? Key Statistics & Insights: ChatGPT reaches over 200 million weekly users, while LLaMA models see 350 million downloads to date.

?? ChatGPT & OpenAI News: OpenAI enhances developer control over AI assistants' personality and functionality.

? Claude Sonnet News: Amazon partners with Anthropic to integrate Claude AI into Alexa devices, enhancing conversational abilities.

?? Humanoid Robots: Cornell researchers develop biohybrid robots controlled by fungal mycelia signals.

?? Corporate AI Developments: Apple reportedly investing in OpenAI to boost AI capabilities in iOS 18.

?? AI Innovations: Magic.dev introduces 100M token context windows for improved software development AI.

?? Applications of AI: Arizona State University adopts ChatGPT Edu for enhanced learning and research across campus.

? Responsible AI: OpenAI supports AI content labelling bill in California to ensure transparency in AI-generated content.

?? AI Learning Hub: Mashable lists various free ChatGPT courses available for skill enhancement in 2024.

?? Videos to Watch: Highlights include "15 Powerful Claude Artifacts Use Cases" and "UK's first 'Teacherless' classroom using artificial intelligence".

?? For more details, read the full newsletter!

?? Top Story: OpenAI's "Project Strawberry" and the Next-Gen AI Model "Orion" – A New Era in AI Development

OpenAI is gaining a lot of attention with its latest initiative, "Project Strawberry," and the development of a new AI model codenamed "Orion." Here’s a combined look at what we know so far:

1. Project Strawberry's Potential: OpenAI's "Project Strawberry" is an ambitious project focused on enhancing AI capabilities. While details are still emerging, the project is expected to introduce new features that could redefine AI interactions, such as more robust GPT models. This initiative is generating high expectations for significant advancements in AI functionalities, aiming to further embed AI into everyday life.
2. Demonstration to U.S. Officials: In a strategic move, OpenAI showcased "Project Strawberry" to U.S. government officials, highlighting its advanced AI capabilities. This presentation reflects OpenAI's commitment to innovation and transparency while preparing to navigate regulatory requirements. The project is being used to develop "Orion," a next-generation AI model that will likely set new standards in AI performance and safety.
3. Introducing Orion – The Successor of GPT-4: Orion is anticipated to be OpenAI's next leap forward in AI, potentially representing the long-awaited GPT-5 model. Built on advancements from "Project Strawberry," Orion is expected to offer improved reasoning capabilities, reduce errors (AI hallucinations), and enhance performance in high-stakes environments like healthcare and finance. This focus on precision and reliability is key to Orion's design, making it well-suited for applications that require dependable AI solutions.
4. Groundbreaking Capabilities: Orion will likely feature enhanced functionalities, such as the ability to solve complex math and programming problems, perform in-depth research, and generate high-quality synthetic data.
5. Future Applications and Industry Impact: As OpenAI prepares for the release of Orion, the AI community is buzzing with anticipation. This model is expected to transform how AI is applied across sectors, particularly those requiring high precision and reliability. Orion's potential to set new benchmarks for AI performance could shape the future landscape of artificial intelligence.

Overall, OpenAI’s developments with "Project Strawberry" and "Orion" signal a significant leap in AI capabilities, reflecting both a commitment to innovation and a careful approach to navigating the challenges of AI governance.

Read more: Digital Trends | The Information | Beebom | Geeky Gadgets | AIM | Newsweek

?? This Week's Spotlight: California's AI Bill SB 1047 and Its Global Impact

California is at the forefront of AI regulation with the passing of Senate Bill 1047, known as the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act. This landmark legislation reflects growing concerns about the rapid advancement of AI technologies and the need for robust safety measures to protect the public. Here's a closer look at the bill, why it matters, and how it compares to AI regulations around the world, with a special focus on Australia.

California's AI Bill SB 1047: Key Details

1. Purpose and Scope:

• SB 1047 targets developers of advanced AI systems, specifically those that cost over $100 million to train. The bill’s primary aim is to ensure that AI systems are developed with safety and accountability as top priorities.

2. Key Provisions:

• Safety Mechanisms: Developers must implement safety features, including the ability to deactivate AI models in cases where they pose significant risks.
• Risk Assessments: Regular risk assessments are required to identify and mitigate potential harms, such as critical or unintended impacts on individuals or society.
• Third-Party Audits: The bill mandates third-party audits of AI systems to provide an external check on compliance and safety standards.
• Whistleblower Protections: It includes protections for individuals who report safety violations or risks associated with AI models, encouraging transparency and accountability within the industry.

3. Legislative Status:

• Having passed the California State Assembly, the bill now awaits further approval from the State Senate and Governor Gavin Newsom. If enacted, it will set a significant precedent for AI governance in the United States.

4. Why It Matters:

• Balancing Innovation and Safety: SB 1047 aims to strike a balance between fostering AI innovation and ensuring public safety. This is crucial as AI systems increasingly influence critical areas such as healthcare, finance, and public safety.
• Setting Standards: As California is a major hub for technology companies, this bill could influence AI regulatory standards not just in the US but globally, setting a benchmark for other regions.
• Industry Impact: While some tech leaders, like Elon Musk, support the bill, others express concerns that stringent regulations could stifle innovation and competitiveness, highlighting the ongoing debate over how best to regulate AI.

Comparative Look: Global AI Regulations

1. European Union: AI Act

• The EU's AI Act is one of the most comprehensive regulatory frameworks, categorising AI systems based on their risk levels (unacceptable, high, limited, and minimal). It imposes stringent requirements for high-risk applications, such as mandatory transparency, risk management, and human oversight. The act is seen as a global standard-setter, pushing for ethical and accountable AI across Europe.

2. United States (Beyond California):

• While the US lacks a federal AI law, various states and agencies have issued guidelines to address privacy and ethics concerns. California's SB 1047 could inspire similar legislation across other states, pushing the country towards more cohesive AI governance.

3. China: Proactive and Strict Regulation

• China is known for its proactive stance on AI regulation, focusing on controlling AI applications that impact social management and security. The country imposes stringent rules on tech companies, including requirements for algorithmic transparency and ethical use of AI, aligning AI development with national interests.

4. United Kingdom: Sector-Based Guidelines

• The UK favours a flexible, sector-specific approach rather than a single overarching law, allowing for tailored regulation that fosters innovation while addressing ethical and safety concerns in specific industries.

5. Canada: Proposed AI and Data Act

• Canada's AI and Data Act is in the proposal stage, aiming to regulate high-impact AI systems through transparency, accountability, and public trust measures. It emphasises the importance of human oversight and ethical AI deployment.

Focus: Australia’s Approach to AI Regulation

1. Current Framework:

• Australia has released AI Ethics Principles as voluntary guidelines to steer the responsible development and use of AI. These principles focus on fairness, transparency, accountability, and the promotion of human-centric AI.

2. Regulatory Developments:

• Discussions are ongoing about establishing a more formal regulatory framework. The Australian government is considering approaches that balance the need for innovation with public safety and trust in AI systems.

3. Challenges and Opportunities:

• Australia's challenge lies in defining the right level of regulation that encourages AI growth while protecting citizens from potential harms. The focus is on creating adaptable laws that can evolve with technological advancements, learning from international examples like California and the EU.

4. Future Outlook:

• Australia is likely to adopt a mix of voluntary guidelines and sector-specific regulations, similar to the UK’s approach, to ensure flexibility and relevance in its AI governance.

Conclusion

California's SB 1047 represents a significant step in AI regulation, aiming to make advanced AI systems safer and more accountable. As AI continues to grow in influence and capability, balancing innovation with safety is becoming a global priority. While different regions adopt varied approaches, the shared goal remains clear: to harness AI’s potential responsibly and ethically. Australia's evolving regulatory landscape reflects this global trend, with a focus on fostering safe and trustworthy AI that aligns with public values and needs.

Read more: Scott Wiener Press Release | Business Insider | Reuters

?? Key Statistics:

• ChatGPT now has over 200 million weekly users, doubling from last November's figures.
• 92% of Fortune 500 companies are using OpenAI products.
• The use of OpenAI's API has also doubled since July.

Read more: Tom's Guide.

• 350 million downloads of LLaMA models to date, a 10x increase from last year.
• 20 million downloads of LLaMA models on Hugging Face in the last month.
• LLaMA's usage by token volume doubled from May to July 2024 after the release of LLaMA 3.1.
• Monthly LLaMA usage grew 10x from January to July 2024 among major cloud service providers.

Read more: Meta AI Blog.

?? ChatGPT & OpenAI News

• OpenAI Enhances Developer Control Over AI Assistants ??? OpenAI has announced new features allowing developers greater control over AI assistant behaviours. Developers can now customise their AI's personality, tone, and functionality, making it more aligned with user needs. This update is part of OpenAI’s broader strategy to offer more flexible and user-friendly AI tools. Read more: VentureBeat

? Claude Sonnet News

• Amazon and Anthropic Team Up to Bring Claude AI to Alexa Devices: Amazon has announced a partnership with Anthropic to integrate the Claude AI model into Alexa devices, aiming to make the virtual assistant more intuitive and conversational. The collaboration promises to enhance Alexa’s capabilities with advanced natural language understanding and improved responsiveness, offering users a more seamless experience. Read more: Mashable | WashingtonPost
• Anthropic's Artifacts Now Available for All Claude.ai Users: Artifacts are now accessible to all Claude.ai users on Free, Pro, and Team plans, enhancing collaboration by transforming conversations into visual and interactive outputs. Users can create and iterate on various content types like code snippets, diagrams, and dashboards. Team users can collaborate securely in Projects, fostering a creative and dynamic experience on Claude.ai. Read more: Anthropic.
• Personal Note: I'm blown away by the Artifact feature's ability to preview and interact with outputs and the remix functionality that lets you recreate and enhance published versions!

?? Humanoid Robots

• Biohybrid Robots Powered by Fungal Mycelia Signals ????: Cornell researchers have developed biohybrid robots controlled by electrical signals from fungal mycelia. These robots, including a soft spider-like bot and a wheeled bot, can react to environmental changes like light due to the unique properties of mycelia. This innovative approach integrates living systems with robotics, offering potential applications in environmental sensing and agriculture. Read more: Tech Xplore.

?? Corporate AI Developments

• Apple Reportedly Investing in OpenAI, Boosting AI Capabilities in iOS 18: Apple is reportedly planning a significant investment in OpenAI, integrating more advanced AI capabilities into iOS 18. This collaboration could bring enhanced ChatGPT features and AI-driven functionalities to Apple devices, aiming to provide more personalised and efficient user experiences. The move highlights Apple's commitment to staying competitive in the evolving AI landscape. Read more: 9to5Mac
• Google Gemini Update: New "Custom Gems" and Enhanced Image Generation: Google’s latest Gemini update introduces "Custom Gems," allowing users to create tailored AI assistants for specific tasks like coding or brainstorming. Additionally, the update features "Imagen 3," an upgraded image generation model that produces high-quality visuals across different styles, from realistic to artistic. These features are designed to enhance creativity and productivity for advanced users. Read more: Google Blog | Decrypt | Engadget.
• Sagard Names First Head of AI to Enhance Investment Strategy and Portfolio Operations: Sagard, part of Power Corporation, has hired Parinaz Sobhani as its first head of AI to integrate AI into investment strategies and portfolio companies. Sobhani will help Sagard-backed firms adopt AI for products, internal systems, and data-driven decisions, aiming to boost efficiency and returns. Her expertise will guide Sagard’s AI adoption, impact financing decisions, and shape the firm’s future strategy. Read more: The Logic.
• AI Drives New Gold Discovery Efforts in Northern Goldfields: Iceni Gold is using AI and machine learning to identify 40 new gold targets at its 14 Mile Well project in Northern Goldfields, WA. Partnering with SensOre, the company has applied AI techniques to geoscience data, creating AI-enhanced predictions for potential gold sites. Drilling for priority targets is planned for 2024, aiming for new gold discoveries. Read more: The West Australian.
• Google Gemini Now Accessible Directly from Chrome Address Bar ??: Google's Gemini AI is now integrated into Chrome's address bar, allowing users to interact with the AI directly while browsing. This feature aims to enhance the search experience by providing instant answers and contextual information without leaving the webpage. It marks a significant step in making AI more accessible and embedded into daily browsing activities. Read more: Search Engine Journal.

• Major Investments Propel AI-Powered Coding Platforms to New Heights ??

The AI-powered coding landscape is rapidly advancing with significant investments and innovations. Codeium, a leader in AI-driven development tools, has secured $150 million in Series C funding led by General Catalyst, raising its valuation to $1.25 billion. This funding will accelerate the enhancement of Codeium's AI coding platform, which focuses on providing smarter and more efficient coding solutions for developers.

Meanwhile, Andreessen Horowitz (a16z) has invested in Cursor, a fork of VS Code tailored for AI-assisted programming. Cursor integrates seamlessly with the latest LLMs and the VS Code plugin ecosystem, offering features like next action prediction, natural language edits, and codebase chat. Competing with tools like GitHub Copilot and Replit Ghostwriter, Cursor differentiates itself by focusing on a seamless developer-AI collaboration experience within the development workflow.

These investments highlight the growing interest in AI-enhanced coding tools that aim to improve efficiency, accuracy, and collaboration in software development.

Read more: Codeium - Business Wire, Cursor - a16z, Cursor.

?? AI Innovations

• Magic.dev Introduces 100M Token Context Windows for Enhanced Software Development AI: Magic.dev's new 100M token context window focuses on improving AI in software development by providing extensive context for code synthesis and documentation. This allows AI to better understand large codebases and libraries. The initiative is supported by partnerships with Google Cloud and NVIDIA to build advanced AI supercomputers, enhancing AI training and inference efficiency. Read more: Magic.dev.

• GameNGen: AI Model Simulates DOOM in Real-Time Using Stable Diffusion ??: GameNGen, developed by Google and Tel Aviv University, uses a modified Stable Diffusion model to simulate the classic game DOOM interactively in real time. This neural network functions as a limited game engine, opening new possibilities for real-time video game synthesis. Despite some visual artifacts and challenges in maintaining temporal coherence, GameNGen demonstrates the potential of AI in game development. Read more: GameNGen, Ars Technica.

• Cerebras Launches World's Fastest AI Inference Solution: Cerebras Systems introduces the world's fastest AI inference solution, Cerebras Inference, delivering speeds up to 1,800 tokens per second for Llama 3.1 8B and 450 tokens per second for Llama 3.1 70B—20 times faster than NVIDIA GPUs. With pricing starting at 10 cents per million tokens, it offers unmatched speed and affordability for real-time AI applications. Read more: Cerebras.

• Alibaba Unveils Qwen2-VL: A Leading Vision-Language AI Model Outperforming GPT-4o Alibaba's Qwen2-VL is a new vision-language AI model that surpasses GPT-4o in several benchmarks, especially in document comprehension and multilingual text-image understanding. It handles images of various resolutions, videos over 20 minutes long, and excels in complex tasks like college-level problem-solving and mathematical reasoning. Qwen2-VL supports multiple languages, including European languages, Japanese, Korean, Arabic, and Vietnamese, making it ideal for global applications. Read more: Qwen2-VL Blog

?? Applications of AI

• Arizona State University Adopts ChatGPT Edu for Enhanced Learning and Research ??: Arizona State University (ASU) has integrated ChatGPT Edu, a customised AI tool, into its academic environment. The AI assists in writing, simulates healthcare scenarios, and recruits research participants. With over 200 active projects, ASU is leading in AI-driven education, fostering innovation and personalised learning experiences across its campus. Read more: TechRadar | Observer.
• AI and Visual Culture: "BeMyAI" and ChatGPT-4 Explore Blindness in Art: The article discusses "BeMyAI," an initiative using ChatGPT-4 to describe visual art to the visually impaired. The project reflects on AI's role in visual culture and its potential to change perceptions of art and accessibility. This exploration raises questions about AI's interpretative abilities and the experience of art beyond sight. Read more: ARTnews.
• AI in Central Texas Schools: Transforming Classrooms and Reducing Teacher Burnout: AI is being integrated into Central Texas classrooms, notably at Alpha Schools, to ease the burden on teachers and enhance student learning. By automating administrative tasks and creating personalized learning paths, AI is helping students learn faster and with more engagement. Texas legislators are considering expanding AI's role in education to tackle ongoing challenges. Read more: Austin American-Statesman.
• AI-Powered Reports: Police Use AI to Draft Crime Reports in Oklahoma City: Oklahoma City police are using "Draft One," an AI tool, to create crime reports from body cam audio, saving time and improving accuracy. While officers appreciate the efficiency, there are concerns about the reliability and legal implications of AI-generated reports. Some cities are cautious, limiting use to minor incidents, while others adopt it more widely. Read more: PennLive.
• AI in Automotive Supplier Selection: Streamlining Processes ??: AI is being used in the automotive industry to help select suppliers, reducing costs and improving efficiency. Companies like Ford and Magna are applying AI to analyse data, manage risks, and streamline supplier choices, although human relationships and decision-making remain crucial. The technology helps with tasks like supply chain disruptions, but its role in strategic partnerships is still evolving. Read more: Detroit News.
• AI Detects Cancer and Viral Infections with Nanoscale Precision: A new AI tool, AINU, analyses high-resolution cellular images to identify cancer and viral infections at a nanoscale level. It detects subtle changes as small as 20 nanometers, potentially allowing for early diagnosis and better monitoring of diseases. This advancement could transform clinical diagnostics and treatment strategies by providing more precise insights. Read more: Medical Xpress.
• Yale Announces $150 Million Investment to Support Leadership in AI: Yale University has announced a $150 million investment to enhance its leadership in AI research and education. This initiative aims to develop new AI-driven technologies, strengthen AI ethics and policy research, and create interdisciplinary programs to equip future leaders with AI expertise. The funding will support faculty recruitment, research centres, and collaboration with other institutions. Read more: Yale News.

? Responsible AI

• California Passes Controversial AI Safety Bill Dividing Tech Giants: California's Assembly has passed SB 1047, an AI safety bill mandating companies spending $100M+ on AI training to implement safety measures. This bill has split Silicon Valley; OpenAI opposes it, while Elon Musk supports it. The decision now lies with Gov. Gavin Newsom, who has until September 30 to act. Read more: Business Insider.
• OpenAI Supports AI Content Labelling Bill in California: OpenAI backs California’s AB 3211 bill, requiring tech companies to watermark AI-generated content to ensure transparency, particularly during elections. This bill aims to differentiate between human-made and AI-generated content, mitigating misinformation risks. The initiative reflects a growing focus on AI ethics and transparency amidst global elections and AI proliferation. Read more: iTnews | TechCrunch.
• Tom Hanks Warns Fans About AI Deepfake Ads Misusing His Likeness: Tom Hanks has alerted fans about an AI-generated advertisement using his likeness without consent. He clarified via social media that he has no association with the ad. This highlights growing concerns over AI and deepfake technologies being used to mislead audiences and infringe on personal rights. Read more: The Hollywood Reporter.
• OpenAI and Anthropic Commit to Rigorous AI Model Testing Before Public Release : OpenAI and Anthropic have agreed to a new initiative to ensure AI models undergo extensive safety testing before being released to the public. This step is aimed at preventing potential misuse and ensuring safe, responsible deployment of advanced AI technologies. The agreement reflects a growing focus on AI safety and ethics in the industry. Read more: Mashable.

?? AI Learning Hub:

• Free ChatGPT Courses to Boost Your AI Skills in 2024 ?? Mashable lists various free ChatGPT courses available as of August 2024, ranging from beginner to advanced levels. These courses help users learn how to use ChatGPT effectively for tasks like content creation, coding, and more. They are hosted on different platforms, making it easy to enhance AI skills without any cost. Learn more: Mashable.

?? Videos to Watch

• 15 Powerful Claude Artifacts Use Cases

• UK's first 'Teacherless' classroom using artificial intelligence

• How Google, Microsoft & Amazon are Raiding AI Startups for Talent.

??? We'd Love to Hear from You!

Did you find this edition insightful? Like, share your thoughts, and ask us anything.

Stay curious, and until next time, enjoy exploring the vast possibilities AI offers.

???? Subscribe on LinkedIn The ChatGPT Observer Newsletter

As we explore the captivating world of AI technologies, it's important to note that the content of this article is a product of human effort, enhanced by AI tools and additional resources. We have made sure to properly attribute and link all referenced articles to their original sources.