• "Towards AGI"
  • Posts
  • Towards AGI’s New Podcast Series Hosted by AI Agents Fred and Zara

Towards AGI’s New Podcast Series Hosted by AI Agents Fred and Zara

Towards AGI has always strived to make complex topics more accessible.

Author

Shen Pandi
September 27, 2024

In partnership with

A Thought Leadership platform to help the world navigate towards Artificial General Intelligence We are committed to navigate the path towards Artificial General Intelligence (AGI) by building a community of innovators, thinkers, and AI enthusiasts.

Whether you're passionate about machine learning, neural networks, or the ethics surrounding GenAI, our platform offers cutting-edge insights, resources, and collaborations on everything AI.

What to expect from Towards AGI: Know Your Inference (KYI): Ensuring your AI-generated insights are accurate, ethical, and fit for purpose Open vs Closed AI: Get expert analysis to help you navigate the open-source vs closed-source AI GenAI Maturity Assessment: Evaluate and audit your AI capabilities Expert Insights & Articles: Stay informed with deep dives into the latest AI advancements But that’s not all!

We are training specialised AI Analyst Agents for CxOs to interact and seek insights and answers for their most pressing questions. No more waiting for a Gartner Analyst appointment for weeks, You’ll be just a prompt away from getting the insights you need to make critical business decisions. Watch out this space!!

Visit us at https://www.towardsagi.ai to be part of the future of AI. Let’s build the next wave of AI innovations together!

Towards AGI’s New Podcast Series Hosted by AI Agents Fred and Zara

Towards AGI is excited to announce the launch of its latest initiative—a podcast series that will bring the world of Artificial General Intelligence (AGI) to listeners everywhere. As part of its ongoing mission to explore and innovate within the AI space, Towards AGI has developed podcasts based on its widely read newsletters, now available in an engaging audio format.

Hosted by AI Agents Fred and Zara

At the heart of this new series are Fred and Zara, AI agents designed to bridge the gap between human and AI-generated content. Their role in the podcasts is to present the latest developments, discussions, and insights around AGI with clarity, efficiency, and a unique AI-driven perspective. By combining human expertise with AI curation, Towards AGI aims to create a dynamic and thought-provoking listening experience.

Why a Podcast?

Towards AGI has always strived to make complex topics more accessible. With the growing interest in AGI and the increasing importance of AI in both research and industry, the podcast series offers a new way for listeners to stay informed. Whether commuting, working, or relaxing, listeners can now engage with deep AI insights, industry trends, and ethical debates in an easily digestible format.

What Makes This Podcast Series Special?

  • AI-Driven Content: Fred and Zara represent the future of AI-driven narratives, offering listeners a unique blend of machine precision and human creativity in every episode.

  • AGI Deep Dives: From the core principles of AGI to its potential applications in the real world, each episode will explore key developments and provide actionable insights.

  • Ethical Considerations: The podcast will dive into the ethics of AGI, discussing responsible AI development and its societal impacts.

  • Industry Insights: With contributions from AI experts and professionals, the podcast will provide a holistic view of AGI's role in the future of technology.

What to Expect in Upcoming Episodes

Listeners can look forward to episodes that cover a variety of topics, including:

  • Breakthrough innovations in AI research.

  • Real-world applications of AGI across industries like healthcare, finance, and education.

  • Discussions around responsible AI and the ethics of AGI.

  • Interviews with leading AI thinkers and organizations.

How to Listen?

The first episode is released today on our social media platforms, and listeners can subscribe via their preferred podcast platforms. Towards AGI will also provide updates and behind-the-scenes content through its newsletters and social media channels.

This new podcast series is yet another step in Towards AGI’s mission to explore the future of Artificial General Intelligence. By merging the best of human insight and AI-driven storytelling, Towards AGI is creating a space where listeners can deepen their understanding of one of the most important technological developments of our time.

Gen AI Is the Future: FLAME University VC Stresses AI Skills for Student Success

“Generative AI is here to stay—it’s the new internet. We are actively incorporating emerging technologies like prompt engineering and data analytics into our curriculum,” said Professor Dishan Kamdar, Vice Chancellor of FLAME University. With AI becoming an essential skill for professionals, Professor Kamdar emphasized that the university updates its curriculum every four years to ensure students remain aligned with the fast pace of technological change.

In an exclusive interview with indianexpress.com, Kamdar highlighted the importance of experiential learning, multidisciplinary education, the need for AI upskilling, and the growing significance of financial and digital literacy in today’s world.

As 2024 approaches, there is a heightened emphasis on financial and digital literacy, pushing academic institutions to adopt a more iterative approach. “Financial literacy isn’t just vital for 2024—it’s a lifelong necessity. If people understood the impact of compounding and applied that knowledge to their savings and investments, we would be in a much stronger financial position. Managing money is a crucial life skill, and we equip students with this knowledge,” Kamdar explained.

Similarly, he acknowledged that digital literacy has become essential. He stressed the importance of familiarizing students with tools like Excel, R, and Python, which have become standard in the workplace. “We prioritize the holistic development of our students, ensuring they acquire both functional expertise and essential life skills,” he added.

AI in Indian Universities Despite the excitement surrounding AI, there remains a skills gap in India, with many academic institutions yet to fully incorporate AI into their curricula. When asked about AI’s role in Indian universities, Kamdar expressed his belief in using technology to enrich the learning experience and ensure that the curriculum remains “rigorous, relevant, and future-focused.”

Kamdar emphasized that Indian universities must focus on their students to stay ahead in the AI era. “If you want your students to succeed, you must be willing to evolve.”

Experiential Learning at FLAME University FLAME University is one of the institutions that champions experiential learning. Explaining FLAME’s approach, Kamdar said that their philosophy is rooted in “learning by doing.” The university’s holistic approach incorporates hands-on projects, internships, field visits, simulations, and case studies across various disciplines.

“At the end of their first year, students participate in the Development Activities Program (DAP), where they spend 30 days working with NGOs, government agencies, and other organizations to better understand socio-economic challenges and engage with people from diverse backgrounds,” he said.

HP Report Uncovers Generative AI’s Role in Crafting Advanced Malware

In its latest threat intelligence report, HP revealed that cybercriminals are increasingly using generative AI to create sophisticated malware. The report highlighted a specific malicious campaign targeting French-speaking users, where AI was used to develop the malware. 

First detected in June, the AI’s involvement became apparent through the presence of code comments—something AI-generated code typically includes. The campaign employed HTML smuggling to distribute a password-protected ZIP archive, which researchers unlocked via brute force. Upon analyzing the code, they discovered that the attackers had commented on every line of code, an uncommon practice for human-written code. Additionally, the structure of the code, along with the use of native language for function names and variables, further suggested that AI was responsible for writing the malware.

Security researchers have raised concerns that cybercriminals are now using generative AI to craft phishing emails and malware, allowing low-level threat actors to customize their attacks based on specific regions and platforms. AI is also being leveraged to accelerate the development of more advanced threats.

In 2023, reports surfaced indicating that threat actors had been using OpenAI’s ChatGPT to write code and launch cyberattacks. While OpenAI took steps to prevent malicious uses of its chatbot, it appears that cybercriminals have found ways to bypass the security measures in generative AI models, enabling them to exploit the technology for broader malicious campaigns.

Ai2 Launches Molmo: Open-Source AI Set to Compete with Industry Titans

AI enthusiasts, rejoice: there’s a new multimodal large learning model available to explore. Seattle-based non-profit AI research organization, the Allen Institute for AI (Ai2), has just launched Molmo—a family of multimodal AI models designed to compete with the capabilities of proprietary vision-based systems from major tech companies like OpenAI and Anthropic.

Multimodal refers to the ability to process different types of data, such as text, images, audio, video, and sensory information. Molmo has demonstrated exceptional skill in interpreting visual data, ranging from everyday objects to complex charts and messy whiteboards.

In a video demonstration, Ai2 showcased how Molmo can create AI agents capable of performing personalized tasks, like ordering food or converting handwritten notes into formatted code.

“This model advances AI development by enabling interaction with the world through pointing out elements,” said Matt Deitke, a researcher at Ai2. He noted that Molmo's success comes from a meticulously curated dataset that trains the AI to understand images through text. Molmo was trained on approximately 1 million images—a relatively small amount compared to the billions used by competitors—resulting in reduced computational demands and fewer errors in responses, according to the research paper.

Ani Kembhavi, Ai2’s senior director of research, explained the reasoning behind this approach: “We focused on using extremely high-quality data at a scale 1,000 times smaller.” This strategy, Kembhavi added, led to models that are as effective as leading proprietary systems but with fewer inaccuracies and faster training times.

The Molmo family includes multiple models of different sizes. MolmoE-1B is a mixture of expert models with 1 billion active parameters (7 billion total). Molmo-7B-O is the most open model with 7 billion parameters, while Molmo-7B-D serves as a demonstration model. At the high end, Molmo-72B is the most advanced model in the series.

Initial tests indicate that even the smaller 7-billion-parameter models perform on par with much larger proprietary systems. This efficiency makes Molmo more accessible to a wider range of developers and researchers, potentially accelerating innovation in the AI field. The team employed novel data collection techniques, such as using speech-based image descriptions from human annotators, which resulted in richer, more detailed captions. They also integrated 2D pointing data to improve tasks like counting and object identification.

Molmo's release is being rolled out in phases. Ai2 has already provided a demo, inference code, a research paper on arXiv, and select model weights. Over the next two months, additional components will be released, including a more comprehensive technical report, the datasets used for training, additional model weights and checkpoints, and the training and evaluation code. By making Molmo’s code, data, and model weights publicly available, Ai2 aims to promote open AI research and innovation—standing in contrast to the closed systems of many leading AI models and potentially accelerating progress in the field.

Meta Launches Llama 3.2: Open Source AI with Vision, Now Mobile-Ready

It’s been a great week for open-source AI. On Wednesday, Meta unveiled an update to its cutting-edge large language model, Llama 3.2, which now includes vision capabilities in addition to text processing. 

Notably, some versions are compact enough to run on smartphones without sacrificing quality, allowing users to have private, on-device AI interactions, apps, and customizations without sending their data to external servers.

Introduced during Meta Connect, Llama 3.2 comes in four versions, each offering different capabilities. The most powerful models, with 11 billion and 90 billion parameters, can process both text and images. These models excel at tasks like chart analysis, image captioning, and identifying objects in images based on natural language descriptions.

Llama 3.2's launch comes in the same week as Allen Institute’s Molmo, which claimed to be the best open-source multimodal vision LLM in synthetic benchmarks, performing on par with GPT-4o, Claude 3.5 Sonnet, and Reka Core in tests. Additionally, Meta introduced two smaller models, with 1 billion and 3 billion parameters, which prioritize speed and efficiency for simpler, repetitive tasks. These lightweight models are adept at multilingual text processing and can integrate well with programming tools. Despite their small size, they feature an impressive 128K token context window, comparable to more powerful models like GPT-4o, making them ideal for summarization, instruction-following, and rewriting tasks on devices.

Meta's engineers achieved this by using structured pruning to remove unnecessary data from the larger models and knowledge distillation to transfer knowledge from the larger models to the smaller ones. As a result, these smaller models outperform rivals in their class, surpassing models like Google's Gemma 2 2.6B and Microsoft's Phi-2 2.7B in various benchmarks. 

Meta is also prioritizing on-device AI by partnering with hardware leaders like Qualcomm, MediaTek, and Arm to ensure Llama 3.2 works seamlessly with mobile processors. Cloud providers such as AWS, Google Cloud, and Microsoft Azure are also offering immediate access to the models on their platforms.

Llama 3.2’s vision capabilities are the result of architectural adjustments. Meta’s engineers added adapter weights to the existing language model, connecting pre-trained image encoders to the text-processing core. This ensures the model's ability to process images does not compromise its text-processing performance, with users expecting equal or better results compared to Llama 3.1.

Llama 3.2 is open-source, according to Meta’s standards, with the models available for download on Llama.com and Hugging Face, as well as through its partner network. Those wanting to run it on the cloud can use Google Collab Notebooks or platforms like Groq, generating nearly 5,000 tokens in under 3 seconds.

Sam Altman Denies Executive Exits Linked to OpenAI Restructuring

Sam Altman, CEO of OpenAI, addressed recent rumors about the departure of three senior executives, denying any link between their exits and an upcoming company restructuring. Speaking at the Italian Tech Week conference in Turin, Altman clarified that the resignations were not connected to organizational changes the board had been discussing for months, according to Reuters.

The announcement that Mira Murati, OpenAI's long-serving Chief Technology Officer, was leaving the company on Wednesday came unexpectedly. Soon after, senior research executives Barret Zoph and Bob McGrew also announced their departures. On the same day, reports surfaced that OpenAI was planning to restructure into a for-profit benefit corporation, shifting away from its non-profit board control to attract more investors.

Altman dismissed the rumors, asserting that much of what had been reported was inaccurate. He explained that the board had been independently considering the restructuring for nearly a year as part of OpenAI's long-term strategy. He also emphasized that the leadership departures were unrelated to these discussions. Altman further clarified that Bob, Barret, and Mira each made independent, amicable decisions to leave. He acknowledged that while the departures were sudden, they were coordinated to ensure a smooth transition in leadership.

Altman expressed his appreciation for the contributions of the departing executives and noted that the organizational shift would allow him to spend more time focusing on the technical aspects of operations, a role he had less time for due to broader responsibilities. He suggested that the restructuring would simplify the company's structure and enhance its ability to grow and innovate.

Altman has already outlined the future leadership structure. Mark will assume the role of Senior Vice President of Research, working alongside Jakub, the Chief Scientist. “Mark has deep technical expertise," Altman noted, adding that he has also developed strong leadership and management skills in recent years.

Other key leadership appointments include Josh Achiam as the new Head of Mission Alignment and Matt Knight as Chief Information Security Officer. Altman acknowledged that the leadership changes are happening earlier than anticipated but assured that OpenAI is ready for the transition.

OpenAI's Ambitious Plan to Distribute AI Like Electricity Across the Globe

Late last year, OpenAI CEO Sam Altman began promoting a bold plan aimed at securing the computing power his company needs to develop more advanced artificial intelligence. In meetings with investors from the United Arab Emirates, computer chip manufacturers in Asia, and officials in Washington, he proposed a multitrillion-dollar initiative to build new computer chip factories and data centers worldwide, including in West Asia. While some participants and regulators were skeptical, discussions have continued, extending into Europe and Canada.

According to nine sources familiar with the company’s talks, OpenAI’s vision for the future of global technology involves constructing numerous data centers that would provide a vast reservoir of computing power for advancing AI. When news leaked that Altman, 39, was seeking trillions of dollars in investment—about a quarter of the U.S. annual economic output—he faced criticism. Additionally, U.S. officials expressed concerns about an American company building critical technology in West Asia, as such projects would require approval under export control regulations.

In response, Altman has scaled back his ambitions to hundreds of billions of dollars and adopted a new approach: courting U.S. government officials by prioritizing the construction of data centers in the United States first. 

Despite the initial skepticism, Altman’s campaign demonstrated how, in just a few years, he has become one of the world’s most influential tech leaders, securing meetings with Middle Eastern financiers, Asian manufacturing giants, and key U.S. regulators in a matter of weeks. This initiative also reflects the tech industry’s drive to fast-track the development of a technology that many believe could be as revolutionary as the Industrial Revolution.

Unlock the future of problem solving with Generative AI!

If you're a professional looking to elevate your strategic insights, enhance decision-making, and redefine problem-solving with cutting-edge technologies, the Consulting in the age of Gen AI course is your gateway. Perfect for those ready to integrate Generative AI into your work and stay ahead of the curve.

In a world where AI is rapidly transforming industries, businesses need professionals and consultants who can navigate this evolving landscape. This learning experience arms you with the essential skills to leverage Generative AI for improving problem-solving, decision-making, or advising clients.

Join us and gain firsthand experience in how state-of-the-art technology can elevate your problem solving skills using GenAI to new heights. This isn’t just learning; it’s your competitive edge in an AI-driven world.

The Daily Newsletter for Intellectually Curious Readers

If you're frustrated by one-sided reporting, our 5-minute newsletter is the missing piece. We sift through 100+ sources to bring you comprehensive, unbiased news—free from political agendas. Stay informed with factual coverage on the topics that matter.

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 on Generative AI. Each edition brings you the most compelling news and insights from the forefront of Generative 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 the leading organizations behind some of the most advanced AI models in open-source, closed-sourced AI.

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!

Keep reading