Meta is struggling to stay ahead in AI after years of investing in research and academia. With the rise of practical AI products like ChatGPT, the company is attempting to shift its focus to commercial application, but this task is challenged by a significant departure of top AI employees and internal restructuring.

1. Meta’s Shift in AI Strategy:

  • The company is diverting resources to develop practical AI products and features.
  • A new AI group has been formed, reporting directly to the Chief Product Officer.
  • This group is working on generative AI models to be used across Meta’s products.
  • Meta has launched an AI language model called LLaMA and an AI model called Voicebox.

2. Departure of AI Talent and Leadership Changes:

  • Significant number of Meta’s AI employees have left the company, including those who co-authored AI research papers.
  • The company’s AI strategy is now more controlled by CEO Mark Zuckerberg and top executives.

3. Challenges and Potential Impact:

  • Meta’s heavy focus on original research in the past has somewhat hindered the company’s commercialization efforts.
  • If Meta fails to quickly capitalize on its AI technology, it risks falling behind competitors.
  • However, successful commercialization could enhance user engagement and create a more robust metaverse.

4. Meta’s Past and Future in AI:

  • Meta began investing in AI in 2013 and hired Yann LeCun, a renowned AI scientist, to lead a new research division.
  • A number of generative AI products are currently under development, including AI agents for Messenger and WhatsApp, AI stickers, and a photo generation feature for Instagram.
  • The company also hopes to integrate AI technology into its metaverse.

5. Challenges in AI Product Release:

  • Meta’s risk tolerance has been impacted by years of scrutiny over its user-privacy practices.
  • Concerns about public reputation have affected the development and release of large language models.
  • Past AI releases, like BlenderBot 3 and Galactica, faced criticism for producing incorrect or offensive content.
  • These concerns pose additional hurdles in Meta’s shift towards commercial AI applications.

Source (WSJ)

PS: I run a ML-powered news aggregator that summarizes with an AI the best tech news from 40+ media (TheVerge, TechCrunch…). If you liked this analysis, you’ll love the content you’ll receive from this tool!