Summary

  • Former OpenAI CTO, Mira Murati, has launched AI research and product company, Thinking Machines Lab.
  • The startup’s focus is on “multimodal” capabilities, and human-AI collaboration, rather than solely agentic systems.
  • It will aim to “advance AI by making it broadly useful and understandable through solid foundations, open science and practical applications” says Murati.
  • The company’s website states that it wants to build more “flexible, adaptable, and personalised” AI systems.
  • It plans to achieve this by focusing on three salient areas, developing strong foundations to build more capable AI systems, fostering a culture of open science to help the field understand and improve these systems, and lastly, adapt and customize AI systems to work for the needs of its users.
  • The company’s small team of engineers and scientists has already created popular open-source projects, including PyTorch, the OpenAI Gym Python library, the Fairseq sequence modeling toolkit, and Meta AI’s Segment Anything.

By Taryn Plumb

Original Article