AI is eating the world (and Google)
summary
Claims and demos focus on concrete outcomes: voice cloning via 11 Labs works in 10–15 minutes from public samples, face swaps run on a free framework called Deep phase lab, and a cheap upscaler (Bigjpeg) rescues a 2,200px image for a printed booth backdrop. Model scale examples include MOM (220 TB web crawl, ~3 billion pages) versus BERT (~56 million Wikipedia articles), and recruiting models reproduce bias (Amazon resume case). Tools for localization and voice (Murph), image generation (DALL·E, Stable Diffusion, Runway Gen 2) and PDF Q&A are shown to work, but LLMs have cutoff datastamps, heavy compute and carbon costs.
AI-written summary · about
chapters (13)
- 0:00Digital Marketing Europe 2023
- 2:20Keynote by Bastian Grimm
- 3:44AI explosion daily
- 8:55AI basics explained
- 12:02LLM theory and challenges
- 22:15AI tools overview
- 24:41AI text generation
- 26:46AI image generation
- 33:22AI audio/video generation
- 38:58AI productivity hacks
- 49:07Advanced ChatGPT uses
- 52:42AI future and LLMs
- 57:11Q&A
related talks
40mBeyond the Hype: Realistic Expectations of AI
18mAI will fundamentally change education
47mSponsored Session: Beyond the Buzz: Navigating AI's Myths and Realities
39mAI is everywhere!
18mSmall AI Teams with Huge Impact
19mAI: The current state of play
44mThe age of applied AI
13mAI as a Strategic Asset: Building Competitive Advantage