AI has made remarkable strides over recent years by training on massive amounts of human-generated data and fine-tuning with expert human examples and preferences.
But in key domains such as mathematics, coding and science, the knowledge extracted from
human data is rapidly approaching a limit. The majority of high-quality data sources – those that can actually improve a strong agent’s performance – have either already been, or soon will be consumed.
To progress significantly further, a new source of data is required. This data must be generated in a way that continually improves as the agent becomes stronger; any static procedure for synthetically generating data will quickly become outstripped. This can be achieved by allowing agents to learn continually from their own experience, i.e., data that is generated by the agent interacting with its environment.
AI is at the cusp of a new period in which experience will become the dominant medium of improvement and ultimately dwarf the scale of human data used in today’s systems. This promises to usher in an unprecedented level of ability.
Read the preprint of a chapter that will appear in the book, Designing an Intelligence, published by MIT Press.
Be First to Comment