About
- AGI is a new leap in automation, producing flexible control systems performing work on command, under changeable objectives and situations.
- Its continual learning capability maintains a causal world model that is simultaneously used for real-time planning and prediction.
- Verification thanks to static guarantees about its operational envelope, which are available both for human certification/audit and introspective decision-making by the system.
- Deployment of AGI proceeds in phases of process modeling, assistance/recommendation, intelligent control, optimization, and flexible re-tasking; each phase capitalizing on the previous.
- Safe operation thanks to interactive goal alignment, verifiability, and obedience.
Presenter
1644572220-b80ee53164a29851
Bas Steunebrink
Co-founder, Director of Artificial General Intelligence at NNAISENSE
From an early age, Bas has asked how intelligence allows one to perform competently despite inevitably insufficient resources. To better understand natural bounded rationality, his research initially focused on artificial emotions before moving to silicon-friendly approaches to general intelligence as an IDSIA postdoc, where he received several best paper awards and a grant from the Future of Life Institute. At NNAISENSE, Bas heads up the effort to develop general-purpose AI.