The article also highlights the need for AI to be accountable, developed and used according to specific community needs. Padilla applauds policy and regulation attempts to guide AI development and implementation, citing the Distributed AI Research Institute's principles as a good roadmap. He also emphasizes the need for AI adoption to be sustainable and to have an affirmative impact on organizations. The article concludes by noting the confusion around the terms 'open' and 'open source' in the context of AI, often used more for marketing than as technical descriptors.
Key takeaways:
- Thomas Padilla, Archive.org’s deputy director, discussed the definition of open artificial intelligence (AI) at the NISO Plus conference, stating that open AI should be reusable, transparent, accountable, and sustainable.
- Meta's Llama 2, despite declaring itself as open source, has license terms that restrict its use by companies with over 700 million active monthly users, leading to debate over its open source status.
- OpenAI's ChatGPT and the open language model OLMo were also discussed, with Padilla emphasizing the importance of transparency and accountability in AI development.
- Padilla also highlighted the need for AI to be developed sustainably, with an awareness of interdependence, threats, and opportunities, and to have an affirmative impact on the evolution of roles in organizations.