Today we’re introducing Perspectives, a space dedicated to thinking in public about the ideas that can advance the safe and robust deployment of AI systems and agents.
These systems have far-reaching implications for most of society, yet they risk being overly shaped by the relatively small community building them. Given that, we believe it’s important to build a “big tent” for AI – to increase the diversity of backgrounds, disciplines, and schools of thought participating in debates about how AI systems are built and deployed.
Perspectives aims to provide a forum for that expanded debate. This space will serve as a curated conversation about topics including AI research, technical safety, policy, ethics, best practices, and other areas directly relevant to the real-world consequences of AI. The impacts of AI are cross-cutting and they call for a multidisciplinary understanding.
We hope to offer a holistic, multidisciplinary picture of how AI is shaping the world and, more than that, to expand the space of ideas that will define our future with this technology.
Shaping the future of AI together
As an organization, we’re developing AI agents with general intelligence that can be safely deployed in the real world. This provides us with the opportunity to share deep insights about how these systems work and what can be done to ensure their safety and robustness. At the same time, we recognize that perspectives from many disciplines are vital to enhancing our shared understanding of how to design and build these systems. We intend to facilitate that conversation.
In particular, we believe we can do that by:
- Developing our own ideas about how to create broadly beneficial AI systems — both to share what we’ve learned and to invite feedback from others
- Highlighting and hosting external content that contributes new insights
- Providing a venue for content that embraces nuance and exploration
- Serving as an index of useful resources that make it easier for others to join the conversation
As a company, we believe in working with the garage door open (see our earlier open source works, like Avalon), and would like to continue that effort as we develop our own approaches to AI safety, policy, and engineering best practices.
We’re excited to share more of our thinking, highlight the expertise and perspectives of others, and showcase data – and research-driven insights — all to foster collaboration and help experts in related fields.
If you would like us to feature your work, or you want to respond to something we have shared, please contact us here.
To stay up to date with Perspectives from Imbue, join our mailing list here.