Discussions around the leading voices of the AI zeitgeist have not often included the old hands of computing like IBM, Intel, Sony Group, or Dell.
But on Tuesday, the four corporations—along with the younger Meta, a host of top universities, as well as a collection of tech startups and foundations—announced an “AI Alliance” in an apparent attempt to challenge the perceived dominance of OpenAI, Microsoft, Google, and recently Amazon.
“To some degree, but unfortunately, to a large degree, the last year of conversation and dialogue around AI has been focused on a very small number of institutions,” Darío Gil, a senior vice president at IBM and head of the corporation’s research lab, told Fortune. “The reality is that this field is much, much larger than that.”
When asked who he was referring to when he said a “very small number of institutions,” Gil declined to specify: “You know who.”
‘Open’ AI
The formation of the AI Alliance continues a longstanding debate among developers about the values of the “open” and “closed” development of artificial intelligence.
Despite its name, OpenAI, the creator of ChatGPT, has kept its models, or mammoth AI algorithms, under lock and key. Developers can only access them only with permission from OpenAI, which counts Microsoft as its biggest backer. Google, another AI frontrunner, as well as Amazon, which recently unveiled its answer to ChatGPT and invested in buzzy AI startup Anthropic, have also not open-sourced, or let researchers fully download, their models. All tech giants have cited the reasons of competition and safety for why they’ve locked up their technology.
This tight-fistedness has led to consternation in the research community and among competing businesses. (In fact, competitors watched with glee as OpenAI’s corporate leadership fell into disarray in November.) “There’s been a lot of debate about: Should the future of AI be closed and proprietary? Or what is the role of open source, open science, and open innovation in the field?” Gil, the IBM executive, said.
The AI Alliance falls into the latter camp. The group of over 50 has coalesced around a number of broad objectives, including the creation of common frameworks for evaluating the strength of AI algorithms, devotion of capital to AI research funds, and collaboration on open-source models.
In addition to the corporate giants, other participants include the chip manufacturers AMD and Cerebras, AI startups like Hugging Face and Stability AI, and Ivy League universities like Yale, Cornell, and Dartmouth.
As an example, Gil pointed to IBM’s work with NASA on a recently open-sourced AI model trained on geospatial data, which he says can help track deforestation or predict crop yields. He also said IBM has committed approximately $100 million to universities to support AI research projects over the next five years, and that the computing titan has worked with Meta to build out an open-source toolkit for AI development.
As for governance, Gil said that the alliance is still working out the details. The focus so far has been on building out a coalition and hashing out the organization’s objectives. Next steps include the formation of “technical working groups” for the more than 50 participants as well as the design of a governance structure that may lead to an external nonprofit.
Ben Weiss
Source link
