• AI KATANA
  • Posts
  • New Study Reveals Data Management Is a Top Challenge in the AI Revolution

New Study Reveals Data Management Is a Top Challenge in the AI Revolution

Also: How two former Google employees in Japan are building AI inspired by fish and bees

Author

AI KATANA
August 20, 2023

Welcome!

Today in AI, a pivotal study by S&P Global Market Intelligence and WEKA highlighted data management as a primary technical challenge in large-scale AI deployment, as the rise of generative AI in 2023 amplifies infrastructure and sustainability concerns. Meanwhile, Sakana, founded by ex-Google employees, champions a nature-inspired approach, using collective behaviors reminiscent of fish and bees for AI development. On the data frontier, the convergence of unstructured data with advanced AI models like GPT-3.5 is pushing the boundaries of applications, though not without ethical challenges. In the arts, musician Grimes envisions AI as a co-creator in her music, while on the tech front, Adobe introduces its AI-augmented tool, Adobe Express, set to disrupt content creation. Lastly, in the venture capital scene, Index Ventures strikes a strategic partnership with Oracle, ensuring GPU access for startups, and Singapore-based HPC-AI Tech secures a robust 22 Million USD in Series A funding.

Sliced:

  • 0️⃣ New Study Reveals Data Management Is a Top Challenge in the AI Revolution

  • 🎏 How two former Google employees in Japan are building AI inspired by fish and bees

  • 🤖 Unstructured Data and Large AI Models: A Powerful Combo

  • 🎧 Grimes Unveils Bold Vision For The Future of AI With Her Music

A recent global study by S&P Global Market Intelligence, in partnership with WEKA, has indicated a surge in AI adoption by enterprises and research institutions aiming for new value propositions. However, the implementation of AI on a large scale faces hurdles due to data infrastructure and sustainability challenges, particularly with the rise of generative AI in 2023. The findings, based on a survey of over 1,500 AI professionals from medium to large organizations across APAC, EMEA, and North America, underscore that data management is the most significant technical obstacle to AI deployment. While 69% of participants are harnessing AI for new revenue streams, only 28% have achieved broad implementation yielding substantial business value. Key concerns include AI's impact on energy consumption and carbon footprint, with aging data systems affecting AI's sustainability performance. The report emphasizes the necessity for modern data infrastructure to accommodate evolving AI workloads across diverse platforms, from edge to cloud, to ensure efficient and sustainable AI integration at scale.

Sakana, a start-up founded by former Google employees David Ha and Llion Jones, is pioneering a unique approach to AI development by drawing inspiration from nature, particularly the collective behavior observed in schools of fish and bee colonies. Instead of developing massive AI systems like many contemporary giants in the field, Sakana aims to design several smaller AI models that function collectively, much like a "swarm." The duo believes that this methodology could potentially lead to AI that is not only more cost-effective but also more adaptable and efficient. With significant credentials in AI research, Ha and Jones hope that by mimicking natural algorithms, they can capture the adaptability and resilience observed in nature. Although still in its nascent stages, Sakana's approach represents a departure from the current trend of building larger and more complex AI models.

In the realm of artificial intelligence, the synergy of unstructured data and large AI models is reshaping the future of AI applications. Unstructured data, including text, images, and videos, has burgeoned in the digital age, offering a vast trove of insights that traditional methods struggle to harness. In contrast, large AI models, built on transformer architectures like GPT-3.5, excel in comprehending and processing this data, revolutionizing areas like natural language processing, image analysis, recommendation systems, and even healthcare. While this combination promises transformative advancements across sectors, challenges around computational demands, transparency, and ethical concerns arise. It is vital that as we harness this powerful duo, we prioritize responsible AI practices that ensure fairness, accountability, and respect for privacy.

In a recent discussion, musician Grimes, also known as Claire Boucher, revealed her innovative approach towards incorporating artificial intelligence (AI) into her music. Rejecting the idea of AI as just a tool, Grimes embraces it as a co-creator, pushing artistic boundaries and merging it with human expression. Central to this vision is her decision to open-source her musical persona, allowing fans to connect and reinterpret her essence. Additionally, she delves into the realm of transhumanism, seeing AI as integral to future human identity. Challenging her peers, Grimes encourages the artistic community to view AI not as competition, but as a partner in innovation. Separately, in a lighter vein, she humorously discussed a potential "cage match" between tech magnates Elon Musk and Mark Zuckerberg, highlighting the complexities of witnessing such an event involving a loved one.

🛠️ AI tools updates

Adobe has unveiled its AI-enhanced Adobe Express, powered by the generative AI model, Firefly. The tool, which underwent beta testing, provides users with a comprehensive design solution, facilitating the creation of diverse content ranging from social media graphics to videos and PDFs. Comparable to platforms like Canva, Adobe Express comes with features such as video and page templates, text-to-image generation, and capabilities to edit both images and videos, including background removal. Govind Balakrishnan, Adobe's SVP, emphasizes the AI-centric nature of the tool, highlighting its transformative potential in content creation. Furthermore, content generated using Express is deemed commercially safe, as Firefly is trained on Adobe's proprietary stock image library. While a free version offers limited features, a $10 monthly subscription provides extensive benefits, including access to Adobe's vast stock content and advanced design features. Currently available on desktop, a mobile version is anticipated soon.

💵 Venture Capital updates

Amid a growing demand for graphics processing units (GPUs) in the AI sector, startups are struggling to secure the necessary computing power, largely due to big corporations and nations hoarding the supply. The crisis has seen founders resort to various strategies to acquire GPUs. In this context, global investment firm, Index Ventures, formed a strategic partnership with Oracle to provide its portfolio companies with coveted Nvidia H100 and A100 chips, alleviating some of their compute power challenges. This move allows Index Ventures' early-stage AI ventures to bypass wait times which can stretch from three months to a year. The deal doesn't involve Index Ventures purchasing the chips outright; rather, they make pre-commitments and foot the cloud bills, while Oracle oversees the cluster management. This innovative arrangement also offers Oracle early access to potential future tech giants. While the partnership isn't exclusive, it illustrates the lengths to which firms are willing to go to ensure their startups remain competitive.

Singapore-based HPC-AI Tech, a company specialized in large AI model training, has successfully secured 22 Million USD in its Series A funding. This achievement follows two prior fundraising rounds since the company's inception 18 months ago. The funding will be primarily used for team expansion and further business development to advance HPC-AI Tech's ambitious growth plans. The firm is recognized for its Colossal-AI deep learning system, which allows for cost-effective implementation of large AI models, addressing industry challenges and earning notable recognition in the AI community. HPC-AI Tech's partnerships with major companies and institutions worldwide reinforce its objective of becoming a dominant player in the AI sector, comparable to OpenAI's global stature. The founder, Prof. Yang You, emphasizes the company's dedication to pushing the boundaries of AI application across various industries, with a vision to lead in Southeast Asia's AI realm.

🫡 Meme of the day

⭐️ Generative AI image of the day