MC.18: Jim Fan's Robotics Predictions, LinkedIn's Productivity Framework, and Trellus' AI Cold Call Coach

Updates & tools

Jim Fan: The biggest thing in 2024, other than LLMs, is Robotics. Period.
Jim thinks we are ~3 years away from the ChatGPT moment for physical AI agents. 2023 was an early preview of the foundation for robots: Multimodal LLMs, progress on robust hardware, and simulators to create real-world data. TWITTER

LinkedIn: Developer Productivity and Happiness Framework
This is a repository documenting the system at Linkedin. It describes the goals, signals and metrics they use to change their infrastructure and workflows according to these goals. To sum it up, they try to minimize context-switching and provide the fastest feedback loop. MORE

Trellus: AI Cold Call Coach
They use real-time analytics to provide sales reps with a personalized AI sales coach during their call. It's interesting to think about other non-obvious ways where AI could observe our actions or listen to our conversations and provide insights and guidance. MORE

CrewAI: A Framework for Orchestrating Autonomous AI Agents
CrewAI is designed to enable AI agents to assume roles, share goals, and operate as a cohesive unit. My feelings about agents are fuzzy; it's already hard enough to get one prompt right, so creating a bunch of agents feels even less predictable. However, the value, if it works, is immense. MORE

Taskade: Super-powered checklists with AI agents
For an Emacs user who likes minimal stuff like me, this tool feels overwhelming. What's interesting is that they try to integrate AI everywhere in a team workspace. It's worth a look. MORE

Artflow: Consistent Characters with Controllable Shot Types
You can create a character, maintain their appearance across scenes, and even change their poses. Having more control over AI-generated images is a crucial next step. Unreal influencers will flood social apps, and I'm not keen to see that. MORE

Scientific American: New Class of Antibiotics Discovered Using AI 
It’s easy to forget other sides of AI besides LLMs and image generation but other use cases are blossoming. Here a deep-learning algorithm helped identify new compounds that are effective against antibiotic-resistant infections in mice, opening the door to AI-guided drug discovery. MORE

TOKYO: Nikon, Sony and Canon fight AI fakes with new camera tech
I think it's smart and necessary. The public will need a way to check that the content they see is real and not generated, especially for journalism. I’m thinking about how to navigate a world with generative AI. Should we try to have human-only content platforms? Does it really matter (AI content could be better at some point than some human-produced content)? How would it work? MORE