4 ms·
Here is a business model for copy right law firms: Use source-aware training, use the same datasets as used in LLM training + copyrighted content. Now the LLM
by DoctorOetker 2y ago
Here is a business model for copy right law firms:
Use source-aware training, use the same datasets as used in LLM training + copyrighted content. Now the LLM can respond not just what it thinks is most likely but also what source document(s) provided specific content. Then you can consult commercially available LLM's and detect copy right infringements, and identify copyright holders. Extract perpetrators and victims at scale. To ensure indefinite exploitation only sue commercially succesful LLM providers, so there is a constant new flux of growing small LLM providers taking up the freed niche of large LLM providers being sued empty.
- chrismorgan 2y ago> Use source-aware training My understanding (as one uninvolved in the industry) is that this is more or less a completely unsolved problem.
- DoctorOetker 2y agoIt's just training the source association together with the training set: https://github.com/mukhal/intrinsic-source-citation https://github.com/mukhal/intrinsic-source-citation The only 2 reasons big LLM providers refuse to do it is 1) to prevent a long slew of content creators filing class action suit. 2) to keep regulators in the dark of how feasible and actionable it would be, once regulators are aware they can perform the source-aware training themselves