Last Week in AI #181 - Google Chatbots, Cerebras vs Nvidia, AI Doom, ElevenLabs Controversy
31 snips
Sep 15, 2024 Dive into the latest in AI as Google reveals its Gemini 1.5 models and customizable chatbots, while Cerebras Systems lays down the challenge to Nvidia with breakthrough AI inference services. Explore the ethical dilemmas surrounding AI biases and transparency, alongside urgent discussions on regulation like California's SB1047. The controversy over voice cloning and the legal pushback faced by ElevenLabs adds another layer to the ongoing dialogue about synthetic media. Plus, find out how China is navigating around US export restrictions on AI chips.
AI Snips
Chapters
Transcript
Episode notes
Startups: Focus on One Use Case
- Startups should focus on a clearly defined use case and excel at it.
- Plod's NotePin demonstrates this by focusing on note-taking, unlike other AI pins trying to do everything.
Cerebras Challenges Nvidia
- Cerebras Systems challenges Nvidia with a faster, cheaper AI inference service.
- Their wafer-scale chip with integrated memory achieves impressive bandwidth, potentially outperforming Grok.
Cerebras Chip Design: Pros and Cons
- Cerebras's wafer-sized chip design offers advantages in memory bandwidth.
- However, limitations in wafer size restrict memory capacity compared to stacking multiple GPUs like Nvidia's H100.