• AI KATANA
  • Posts
  • The Next Token of Progress: 4 Unlocks on the Generative AI Horizon

The Next Token of Progress: 4 Unlocks on the Generative AI Horizon

Also: Balancing the yin and yang of AI for the benefit of all

Author

AI KATANA
June 25, 2023

Welcome!

Today we spotlight a comprehensive analysis by Andreessen Horowitz on the future of Large Language Models (LLMs), focusing on the four major evolutionary paths these models are anticipated to take: steering, memory, "arms and legs", and multimodality. With each development aiming to enhance AI utility, performance, and integration, we're truly on the verge of a transformative era for this technology. In other news, Amazon makes its move with the launch of a $100 million Generative AI Innovation Center, a playground for exploration and development in the generative AI space. Bain Capital Ventures weighs in on the role of generative AI in financial services, whilst we're also exploring how to balance the 'yin and yang' of AI for societal benefit.

Sliced just for you:

  • 🔓 The Next Token of Progress: 4 Unlocks on the Generative AI Horizon

  • 🛝 Amazon's generative AI playground is open

  • 💲 The State of Generative AI in Financial Services

  • ☯️ Balancing the yin and yang of AI for the benefit of all

Large language models (LLMs) have revolutionized the tech industry by demonstrating exceptional natural language understanding and ushering in the generative era of AI. Their evolution, though not without criticism, is marked by four upcoming innovations: steering, memory, "arms and legs", and multimodality. Steering aims to improve control over LLM outputs, enhancing user trust and output personalization, particularly vital in industries like healthcare and enterprise. Memory upgrades enhance context understanding, expanding the amount of text that models can process and improving their ability to maintain context. The "arms and legs" evolution aims to allow LLMs to perform tasks based on their outputs, transforming them from passive responders to active participants. Finally, multimodality aims to enable LLMs to handle various data types, like images and sound, increasing their utility. However, for a comprehensive understanding of these innovations and their implications, the full article is recommended.

Amazon Web Services (AWS) is enhancing its AI initiatives through the launch of a $100 million Generative AI Innovation Center. The center is a part of Amazon's dual strategy: to diversify interest in generative AI beyond the current fascination with ChatGPT, and to increase AWS's share in the cloud computing market for running generative AI services. The center, acting as an AI sandbox and a tutoring service, aims to link AWS AI strategists, data scientists, and engineers with customers. AWS encourages its customers to initially clean their data, brainstorm, and build prototypes, with the ultimate goal of developing and deploying their own custom generative AI products and services. AWS anticipates that the use cases will concentrate on enhancing customer experiences, optimizing business operations, and accelerating creative production. Sri Elaprolu, the head of the new center, underscores the importance of cloud computing for training on internet-scale data, an endeavor that can't be effectively achieved in a data center or a server farm.

Bain Capital Ventures, in their comprehensive report, explores the impact of generative AI in the financial services industry. They argue that traditional and generative AI will coexist within financial services organizations, with generative AI filling gaps left unfulfilled by traditional AI. For instance, generative AI could enhance customer experience, employee satisfaction, and shareholder earnings, areas often neglected or inadequately addressed by traditional AI. Despite the significant investments in technology, including traditional AI, the financial services sector still has room for improvement, particularly in customer satisfaction and service cost reduction. The report also suggests that generative AI could be crucial for financial services organizations to reach their full potential.

The article "Balancing the yin and yang of AI for the benefit of all" highlights the importance of understanding that AI is not neutral but shaped by the values and beliefs of those who develop it. As AI becomes more widespread, it becomes crucial to ensure that the benefits are universal and that potential risks don't undermine our rights and freedoms. It raises concerns about the impact of AI on democratic processes, asserting that unchecked AI could disrupt the social contract between governments and citizens. Governments, facing the need to digitalize and optimize public action, are increasingly utilizing AI to improve various aspects of governance. However, as the technology presents degrees of autonomy, key questions about its development, data processing, and auditability need to be addressed to ensure citizen trust.

🛠️ AI tools update

Midjourney, an AI-powered image synthesis model, recently unveiled its 5.2 version, which includes a novel "zoom out" feature. This feature keeps a central synthesized image intact while building a larger scene around it, much like zooming out with a camera lens. This is similar to the "outpainting" technique used by OpenAI's DALL-E 2, except that Midjourney's feature only works on images generated within its own service. Users can experiment with this on the Midjourney Discord server by generating a v5.2 image and then using the special "Zoom" buttons to zoom out.

💵 Venture Capital updates

The UK government has announced a £21 million ($26.75 million) fund aimed at accelerating the deployment of promising AI tools in National Health Service (NHS) hospitals across the country. This decision comes amidst the rise of powerful large language models that have the potential to revolutionize healthcare. Despite previous controversies around AI and data privacy within the NHS, recent advances in AI have spurred the UK to position itself at the forefront of the AI industry, offering support for AI development in various sectors including healthcare.

Korean telecom company KT has revealed plans to invest KRW7 trillion ($5.4 billion) to develop a large-scale artificial intelligence (AI) system and expand its AI-based services by 2025. The goal is to achieve this by broadening the company's AI-related offerings in logistics, robotics, healthcare, and other sectors. Specifically, KRW4 trillion will be used to acquire large-scale AI technologies, KRW2 trillion for advancing AI infrastructure and associated cloud services, and KRW1 trillion to explore new models for AI-related services in robotics, education, and healthcare sectors.

🫡 Meme of the day

⭐️ Generative AI image of the day