#137 - Salesforce Copilot, Chip Crunch, Meta Rival to ChatGPT, AI for paralysis patients
Sep 20, 2023
auto_awesome
Jessica Dai, a PhD student at Berkeley and co-founder of the techno-optimism publication Reboot, discusses exciting advancements in AI. She dives into Salesforce's new AI assistant, Einstein Copilot, and examines the challenges posed by AI chip shortages. The conversation also highlights Meta's plans for an open-source generative AI rival to ChatGPT and groundbreaking AI applications that help individuals with paralysis communicate. Dai emphasizes the balance between innovation and ethical considerations in the rapidly evolving landscape of AI.
The podcast discusses the trend of open sourcing AI models and its impact on business use, highlighting the release of GPT-3 and the potential rivaling with an even larger model, GPT-4.
Researchers at the University of California have developed an AI-powered device that allows paralyzed patients to communicate via a digital avatar, translating brain signals into speech and providing a means for expression.
Deep dives
Tech leaders call for government action in AI
Tech leaders, including Musk, Zuckerberg, and others, recently participated in a six-hour conversation with lawmakers in the Senate regarding AI. Topics discussed ranged from open-source models and potential uses of AI to job concerns and the idea of a new regulator for AI. The meeting aimed to inform lawmakers about AI and its implications, although opinions on its effectiveness varied. The session may have also allowed industry players to advocate for their preferred vision of AI.
US court rules AI-generated artwork cannot be copyrighted
A US district judge has ruled that AI-generated artwork cannot be copyrighted, establishing legal precedent for protections of AI-generated works. The ruling follows a case involving Stephen Thaler, a computer scientist who attempted to copyright AI-generated art in 2012. This decision highlights the complexity of copyright laws in relation to AI-generated content and raises questions about ownership and legal protection for such works.
AI models for business use and Indian developers on hugging face leaderboard
The podcast episode discusses the recent trend of open sourcing AI models by companies like Meta (formerly Facebook) and the impact on business use. They have released models like GPT-3 and recently announced the possibility of rivaling GPT-4 with an even larger model. While there are debates on whether this can be considered true open source and concerns about the accessibility of hosting and serving these models, the move is said to benefit academia and research. In a related development, Indian developers have made their mark on the hugging face leaderboard with the fine-tuned language model Gen Z 70B, showcasing the increasing proliferation of open source models and the potential for further advancements.
Restoring communication for paralyzed patients with AI and US legislative developments on AI laws
Researchers at the University of California have developed a device that enables paralyzed patients unable to speak to communicate via a digital avatar. This AI-powered system translates brain signals into words and voice, providing a means for expression. In the realm of AI legislation, two US senators have proposed a framework for AI laws, which includes licensing and auditing of AI systems, the creation of a federal oversight office for AI, and addressing liability for privacy and civil rights violations. Meanwhile, the US Copyright Office has launched a study on AI and copyright law, inviting public feedback to inform regulatory endeavors.
Check out our sponsor, the SuperDataScience podcast. You can listen to SDS across all major podcasting platforms (e.g., Spotify, Apple Podcasts, Google Podcasts) plus there’s a video version on YouTube.