An alliance of tech heavyweights pushes for AI to have an open source future

More than 50 organizations have joined together to get the industry on board with open AI (no, not that one).

What happened: Meta and IBM are leading what is called the AI Alliance, a group advocating for an open source approach to developing AI, saying it is the faster way to innovate and identify societal risks.

  • The alliance’s projects will include creating benchmarks and tools for responsible AI, as well as advancing hardware development, academic research and models targeted to things like climate change.
  • Other companies involved include Intel, Oracle, Dell, AMD, Sony, and Stability AI. Also on board are a number of schools, research institutes, and non-profits, such as the Linux Foundation, UC Berkeley, and CERN.

Catch-up: Research-led, open source approaches have driven most AI development over the decades, but as it has become more commercialized, the likes of OpenAI, Google, Amazon and Microsoft (none of which are part of the new alliance) have made closed systems the market leaders.

  • Open source advocates say a more transparent approach that encourages sharing knowledge is the best way to both democratize advancements and ensure AI is developed responsibly.
  • Opponents, however, have claimed open access to top-tier models, like Meta’s Llama, can make it easier for people to create AI-powered spam, bots and scams.

Why it matters: The motives here, shockingly enough, may not be totally benevolent. Meta and IBM are falling behind OpenAI, Microsoft and Google in the AI race. Shaping an environment that makes their approaches the norm could help them make up some ground.