Welcome to the "Towards AGI" Newsletter

In our quest to explore the dynamic and rapidly evolving field of Artificial Intelligence, this newsletter is your go-to source for the latest developments, breakthroughs, and discussions. Each edition brings you the most compelling news and insights from the forefront of General AI (GenAI), featuring cutting-edge research, transformative technologies, and the pioneering work of industry leaders.

Highlights from GenAI, OpenAI and ClosedAI: Dive into the latest projects and innovations from OpenAI, the organization behind some of the most advanced AI models like GPT and DALL-E. Equally, we'll give you an inside look at the initiatives and strategies from ClosedAI, a hypothetical entity committed to pushing the boundaries of closed-source AI development.

Stay Informed and Engaged: Whether you're a researcher, developer, entrepreneur, or enthusiast, "Towards AGI" aims to keep you informed and inspired. From technical deep-dives to ethical debates, our newsletter addresses the multifaceted aspects of AI development and its implications on society and industry.

Join us on this exciting journey as we navigate the complex landscape of artificial intelligence, moving steadily towards the realization of AGI. Stay tuned for exclusive interviews, expert opinions, and much more!

TheGen.AI News

India Leads in Generative AI Adoption Across Asia-Pacific, Deloitte Report Finds

India leads the Asia-Pacific region in adopting Generative Artificial Intelligence (GenAI), with 93% of students and 83% of employees using the technology, according to Deloitte India report. However, only half of these employees believe their managers are aware of their use of this technology. The report, titled "Generative AI in the Asia Pacific: Young Employees Lead as Employers Play Catch-Up," surveyed 11,900 individuals across 13 countries and predicted a 182% increase in daily GenAI usage over the next five years.

The study also highlighted that GenAI could impact 17% of working hours in the region, totaling approximately 1.1 billion hours annually. Users in the Asia-Pacific save an average of 6.3 hours weekly, while Indian users save about 7.85 hours, aiding in skills development. Furthermore, 86% of Indian users and 83% in the region believe GenAI boosts their productivity and enhances ideation.

Despite these advantages, the report indicates a gap in business implementation, with 75% of users feeling their companies lag in embracing GenAI. Anjani Kumar, a partner at Deloitte India, emphasized the need to overcome implementation challenges and deepen understanding of GenAI to maximize its benefits. Kumar urged leaders to leverage these technologies not just for efficiency but also to innovate business models and processes.

GoDaddy Integrates Generative AI for Enhanced Domain Name Selection

GoDaddy is now utilizing Generative AI to assist users in selecting appropriate domain names for their websites. By entering a business description or keywords, users can generate distinctive domain names that reflect their brand's identity and objectives, according to a company announcement.

This innovation follows the introduction of GoDaddy Studio with AI, which enables users to create professional-grade content for businesses or personal brands. Furthermore, GoDaddy’s Prompt Library offers entrepreneurs tools to understand how AI can optimize their business operations, ultimately saving time and promoting business growth.

“AI is transformative for small businesses, enhancing efficiency by automating tasks such as content creation, product descriptions, social media management, and now domain name searches. At GoDaddy, we are committed to equipping small businesses with the latest tools and technologies to help them thrive in a competitive market,” stated Apurva Palnitkar, Senior Director of Marketing at GoDaddy India.

Google I/O 2024: Generative AI to Organize Search Results for Enhanced Discovery

At the Google I/O 2024 event, Google revealed plans to expand the use of generative AI across its search results pages for specific queries. This development builds on the existing AI Overview feature, which provides a concise summary of search topics and became widely available on Tuesday following tests in Google’s AI Labs.

Elizabeth Harmon Reid, the head of Google Search, indicated in a press briefing before the announcement that generative AI would not only enhance the AI Overview but would also be integrated throughout the search experience. Google intends to deploy this AI-driven format particularly when users seek inspiration, such as planning an anniversary trip to Dallas. Upcoming implementations will include search results for dining, recipes, and eventually extend to movies, books, hotels, shopping, and more.

Reid demonstrated how the AI could tailor search pages, such as displaying a list of "anniversary-worthy restaurants" complete with AI-generated review summaries and additional thematic content like rooftop dining spots in Dallas. The search results will also include links to further traditional results at the bottom of the page. The positioning of ads within this new format, however, remains unspecified.

During a pre-conference press briefing, when asked about the future of traditional Google Search in the era of its Gemini project, CEO Sundar Pichai emphasized Google’s commitment to evolving with user needs rather than directly addressing the survival of the conventional search model. Pichai expressed enthusiasm about expanding the scope of user assistance through these AI enhancements, seeing them as opportunities for growth and user engagement enhancement.

TheOpen.AI News

Open Source Initiative Explores Definition of Open Source AI

The Open Source Initiative (OSI), a nonprofit that manages the Open Source Definition for software licenses, is now turning to the public to help define what constitutes Open Source AI. This initiative comes after two years of deliberation on how to classify AI systems under open source terms, given the uniqueness of machine learning models and their usage which doesn't completely align with traditional software definitions.

The OSI is concerned that the terminology used in current OSI-approved licenses, such as referring to machine learning models as "programs," may not be fully applicable to the AI context. According to Stefano Maffulli, the executive director of OSI, AI challenges the conventional understanding of open source, prompting a reassessment of these principles in the AI realm.

To tackle this, OSI plans to host a series of global workshops at various international conferences across the US, Europe, Africa, Asia, the Pacific, and Latin America through September. The goal is to gather broad feedback on the draft of its Open Source AI Definition, currently in version 0.0.8.

However, Bruce Perens, the original author of the Open Source Definition, expressed skepticism about creating a separate open source category for AI. Perens, who parted ways with the OSI four years ago, believes the existing definition should suffice for all software, including AI. He criticized the move as potentially diluting the open source brand and highlighted concerns that the outputs of AI, such as those from large language models, often amount to plagiarism due to their training on copyrighted materials without appropriate permissions. Perens anticipates that such issues will ultimately be resolved in court, much like the Napster case.

Hugging Face Unveils LeRobot, Aiming to Democratize Robotics Technology

Hugging Face has introduced LeRobot, a new machine learning model tailored for practical robotics applications. LeRobot serves as a comprehensive platform, featuring a broad library for data sharing, visualization, and the training of sophisticated models, specifically designed around the PyTorch framework.

The main objective of LeRobot is to democratize access to robotics by providing readily accessible models, datasets, and tools. This initiative aims to make it easier for anyone interested in robotics to engage by sharing datasets and leveraging pretrained models.

Remi Cadene, formerly a staff scientist at Tesla, Inc., and currently with Hugging Face, described on his X account that LeRobot aspires to be for robotics what the Transformers library has been for NLP.

LeRobot enhances the ease of starting projects by providing pretrained models and integrating smoothly with physics simulators. It has been tested in environments like AlohaTransferCube, where it was benchmarked against models from the original ACT repository over 500 episodes to gauge its effectiveness.

Further evaluations were carried out in the PushT environment against models trained with Diffusion Policy code. These tests also spanned 500 episodes and included success metrics to thoroughly assess LeRobot's practical capabilities.

LeRobot is designed to work with a variety of robotic hardware, ranging from simple educational robotic arms to advanced humanoid robots used in research, making it a versatile and scalable AI system for controlling different types of robots.

As an open-source project on GitHub, LeRobot is part of Hugging Face's effort to decentralize innovation by encouraging a global community of developers, researchers, and hobbyists to contribute and benefit from the latest developments in AI and robotics.

The launch of LeRobot has generated significant excitement within the AI and Robotics community, with community members on X heralding the onset of a robotics boom.

Soket AI Labs Launches Pragna-1B, India’s First Open-Source Multilingual AI Model

Soket AI Labs, an Indian AI research company, announced on Wednesday the launch of Pragna-1B, India's first fully open-source multilingual foundation model, developed in collaboration with Google Cloud to promote the adoption of Generative AI. Pragna-1B supports multiple Indian languages including Hindi, English, Bengali, and Gujarati, and is tailored specifically for the Indian market to provide clear and transparent AI integration for businesses.

Abhishek Upperwal, the founder of Soket AI Labs, emphasized the benefits of their partnership with Google Cloud in the pre-training phase of Pragna-1B, noting that it enabled the model to be both cost-effective and efficient. Although Pragna-1B operates with fewer parameters, it still matches the performance of similar models in language processing tasks due to its high efficiency and innovative design.

Bikram Singh Bedi, the Vice President and Country MD of Google Cloud India, also expressed enthusiasm about the collaboration, stating that the launch of Pragna-1B on Google Cloud represents a significant advancement in technology for Indian languages. He highlighted that the model enhances scalability and efficiency for organizations, further aiming to broaden AI innovation across India.

Additionally, Soket AI Labs plans to make its AI Developer Platform available on the Google Cloud Marketplace and will include the Pragna series models in the Google Vertex AI model registry. This move is intended to offer developers a robust and efficient platform for model fine-tuning.

This partnership underscores the commitment to developing large-scale models and compiling high-quality datasets for Indian languages, which will contribute to driving AI innovation within the country.

TheClosed.AI News

U.S. Stock Futures Hold Steady as Reddit Shares Surge on OpenAI Deal

U.S. stock futures remained stable on Friday, showing little movement after a recent session where the Dow briefly hit a record high intraday. This pause follows a decline in stock prices on Thursday, despite optimism earlier in the week driven by lower-than-expected inflation data, which had fueled hopes of potential interest rate cuts by the Federal Reserve later this year.

In other news, Reddit's stock prices soared by 11.5% after hours following the announcement of a new partnership with OpenAI. This deal allows OpenAI to utilize Reddit’s content for its ChatGPT service, while Reddit will integrate advanced AI features for its users and moderators. This development is part of OpenAI's broader strategy to collaborate with media companies to enhance its AI models with trusted data sources, although it faces legal challenges from some publishers concerning copyright issues.

Meanwhile, Boeing is preparing for a critical vote where shareholders will decide if the outgoing CEO, David Calhoun, should continue to serve on the board. This vote comes amid ongoing leadership changes and safety concerns, highlighted by a recent incident involving a Boeing 737 Max jet, which has negatively impacted the company's stock price.

From an international perspective, China reported a significant rise in industrial output in April, exceeding expectations and suggesting continued growth in its manufacturing sector. However, consumer spending remains subdued, as indicated by weaker retail sales figures for the same period.

In the commodities market, oil prices edged higher on Friday, on track for weekly gains driven by signs of increased demand and easing inflation in the U.S. Both U.S. crude and Brent crude saw price increases, supported by a reduction in U.S. oil inventories and anticipations of higher demand during the upcoming summer travel season.

Sundar Pichai Responds to Allegations of OpenAI's Unauthorized Use of YouTube Content

OpenAI may have used YouTube videos to train its AI model, Sora, without obtaining the necessary permissions. Google's CEO, Sundar Pichai, has stated that he will take action if these allegations prove to be true. OpenAI's CTO, Mira Murati, has expressed uncertainty about whether Sora's training involved YouTube content, noting that the model generally uses public and licensed data. However, a New York Times article claims that OpenAI has used transcripts from over a million hours of YouTube videos for training purposes.

When CNBC inquired about potential breaches of Google's policies, Pichai indicated that it is OpenAI's responsibility to ensure compliance with terms of service. The scrutiny around OpenAI's data practices has intensified, leading to legal challenges, including a lawsuit from the New York Times accusing the AI firm of copyright infringement for using its content without authorization.

Furthermore, the Authors Guild has filed a lawsuit claiming that OpenAI's language models heavily depend on copyrighted content without fair compensation or acknowledgment to the original creators. These legal issues highlight the ongoing concerns about OpenAI's approach to data sourcing, particularly its use of copyrighted materials without permission, sparking debates over intellectual property rights within the AI industry.

The Authors Guild's lawsuit emphasizes the extent of copyrighted content allegedly used by OpenAI, including millions of articles, books, and other works, without proper compensation or attribution. The outcomes of these lawsuits could significantly impact how AI developers handle data usage and intellectual property rights moving forward.

OpenAI vs. Google: Altman Sparks Debate Over Event Presentation Styles

OpenAI CEO Sam Altman recently stirred up some controversy with his comments on Google's event styling amidst intense rivalry in the AI industry. Both tech giants held their major annual events focused on AI innovations, where OpenAI unveiled its advanced GPT-4o model capable of processing diverse data types in real-time. The next day, Google revealed significant updates to its AI systems, including a new agent named "Project Astra" and improvements in Google Search.

Rather than critiquing Google's technological advancements, Altman took to X (formerly Twitter) to comment on the visual differences between the events of the two companies. He posted a comparison highlighting OpenAI's minimalistic setup in a stylish, dimly lit office against Google's elaborate, brightly illuminated stage featuring artificial windows and cloud decorations in vibrant Google colors.

While Google's event lasted nearly two hours with various speakers and even included a performance by musician Marc Rebillet, OpenAI kept its presentation brief, under half an hour, and hosted it within its office settings, without Altman present. Google engineer Zachary Nado responded to Altman's post, pointing out the close timing of OpenAI's announcements with those of Google, suggesting a strategic rivalry.

Adding to the competitive atmosphere, Google launched a demo video demonstrating its AI capabilities just 30 minutes before OpenAI's event commenced, highlighting the ongoing competition between the two companies.

Join the movement - If you wish to contribute thought leadership, then please fill the form below. One of our team members will be back in touch with you shortly.

Form link

Keep reading