The Biden administration has initiated a discussion on the debate surrounding the openness of the most powerful artificial intelligence systems. This move comes as part of a broader executive order signed by President Joe Biden in October to address the rapidly evolving technology landscape.
The White House announced that it is soliciting public feedback on the advantages and disadvantages of making key components of AI systems publicly available for use and modification. This inquiry is a crucial aspect of the executive order and aims to gather diverse perspectives on the issue.
There is a divide among tech companies regarding the openness of AI models, with some highlighting the risks associated with widely accessible components, while others emphasize the importance of open science for researchers and startups. Companies like Meta Platforms and IBM have been vocal proponents of an open approach to AI.
The executive order specifically mentions 'dual-use foundation models with widely available weights' and underscores the need for further examination of these models. The weights in AI models are numerical values that significantly impact their performance.
The potential benefits of publicly sharing these weights include fostering innovation, but there are also significant security risks, such as the potential compromise of the model's safeguards. Commerce Secretary Gina Raimondo has been tasked with consulting experts and presenting recommendations on how to balance these benefits and risks by July.
The National Telecommunications and Information Administration under the Commerce Department has also opened a 30-day comment period to gather input for a report to the president. The goal is to find solutions that support both innovation and safety in the development and deployment of AI technologies.
While some companies like Meta Platforms have embraced an open approach to AI, others like Google have leaned towards a more closed model. However, Google recently introduced a new group of open models called Gemma, which are derived from its closed technology used in the Gemini chatbot app. Google emphasized the importance of prioritizing safety in releasing open models and called for a nuanced and collaborative approach to assessing the risks and benefits of AI openness.
