Provenance for repurposing of artist-specific assets using non-fungible tokens (NFTs)
The technology disclosed relates to a provenance system for tracing repurposing of targeted artificial intelligence (AI) systems. A training logic is configured to train an AI system on a training dataset that satisfies a target artist configuration by requiring that at least some training samples in the training dataset are sourced from a target artist, and to generate a trained version of the AI system (trained AI system). The target artist configuration characterizes a work of the target artist. The trained AI system is configured to construct an output that satisfies that target artist configuration. A deployment logic is configured to make available the trained AI system via a blockchain network. A provenance logic is configured to provide access to the trained AI system, and to validate that new outputs generated by repurposing the trained AI system satisfy the target artist configuration.
1. A provenance system for tracing repurposing of targeted artificial intelligence systems, comprising:
memory storing a training dataset that satisfies a target artist configuration by requiring that at least some training samples in the training dataset are sourced from a target artist, wherein the target artist configuration characterizes a work of the target artist;
training logic, having access to the memory, and configured to train an artificial intelligence system on the training dataset, and to generate a trained version of the artificial intelligence system, wherein the trained version of the artificial intelligence system is configured to construct an output that satisfies that target artist configuration;
deployment logic configured to make available the trained version of the artificial intelligence system via a blockchain network; and
provenance logic configured to provide access to the trained version of the artificial intelligence system, and to validate that new outputs generated by repurposing the trained version of the artificial intelligence system satisfy the target artist configuration;
wherein the repurposing of the trained version of the artificial intelligence system further comprises processing a new input through the trained version of the artificial intelligence system, and generating a new output that applies the target artist configuration to the new input, and
wherein the new input is a new image, and the new output is a reconstructed version of the new image overlayed with the target artist configuration.
2. The provenance system of claim 1 , wherein the provenance logic is further configured to assign a parent non-fungible token (NFT) to the trained version of the artificial intelligence system.
3. The provenance system of claim 2 , wherein the parent NFT is configured to be traced back from offspring NFTs that are assigned to downstream results of the repurposing.
4. The provenance system of claim 3 , wherein the downstream results of the repurposing are digital assets generated using the trained version of the artificial intelligence system.
5. The provenance system of claim 3 , wherein the downstream results of the repurposing are digital assets generated using further trained versions of the trained version of the artificial intelligence system.
6. A computer-implemented method of tracing repurposing of targeted artificial intelligence systems, including:
storing a training dataset that satisfies a target artist configuration by requiring that at least some training samples in the training dataset are sourced from a target artist, wherein the target artist configuration characterizes a work of the target artist;
training an artificial intelligence system on the training dataset, and to generate a trained version of the artificial intelligence system, wherein the trained version of the artificial intelligence system is configured to construct an output that satisfies that target artist configuration;
deploying the trained version of the artificial intelligence system via a blockchain network;
providing access to the trained version of the artificial intelligence system; and
validating that new outputs generated by repurposing the trained version of the artificial intelligence system satisfy the target artist configuration;
wherein the repurposing of the trained version of the artificial intelligence system further including processing a new input through the further trained version of the artificial intelligence system, and generating a new output that applies a combination of the target artist configuration and the further target configuration to the new input.
7. The computer-implemented method of claim 6 , wherein the repurposing of the trained version of the artificial intelligence system further comprises further training the trained version of the artificial intelligence system on a further training dataset that satisfies a further target configuration, and generating a further trained version of the artificial intelligence system, wherein the further trained version of the artificial intelligence system is configured to construct an output that satisfies a combination of the target artist configuration and the further target configuration.
8. The computer-implemented method of claim 6 , wherein a further training dataset requires that at least some training samples in the further training dataset are sourced from a further target artist, wherein a further target configuration characterizes a work of the further target artist.
9. The computer-implemented method of claim 8 , wherein the further target configuration conditions generation of the new output on one or more user-supplied parameters.
10. The computer-implemented method of claim 9 , wherein the one or more user-supplied parameters include a random seed.
11. The computer-implemented method of claim 10 , wherein the random seed is based on a Mandelbrot set.
12. A non-transitory computer-readable storage medium impressed with instructions to provide provenance of tracing repurposing of targeted artificial intelligence systems, the instructions, when executed on a processor, perform operations comprising:
storing a training dataset that satisfies a target artist configuration by requiring that at least some training samples in the training dataset are sourced from a target artist, wherein the target artist configuration characterizes a work of the target artist;
training an artificial intelligence system on the training dataset, and to generate a trained version of the artificial intelligence system, wherein the trained version of the artificial intelligence system is configured to construct an output that satisfies that target artist configuration;
deploying the trained version of the artificial intelligence system via a blockchain network;
providing access to the trained version of the artificial intelligence system; and
validating that new outputs generated by a repurposed trained version of the artificial intelligence system satisfy the target artist configuration;
wherein the repurposing of the trained version of the artificial intelligence system further including processing a new input through the trained version of the artificial intelligence system, and generating a new output that applies the target artist configuration to the new input, and
wherein the new input is a new image, and the new output is a reconstructed version of the new image overlayed with the target artist configuration.
13. The non-transitory computer-readable storage medium of claim 12 , further including assigning a parent non-fungible token (NFT) to the trained version of the artificial intelligence system, wherein the parent NFT is configured to be traced back from offspring NFTs that are assigned to downstream results of the repurposing.
14. The non-transitory computer-readable storage medium of claim 12 , wherein the repurposing of the trained version of the artificial intelligence system further comprises:
further training the trained version of the artificial intelligence system on a further training dataset that satisfies a further target configuration; and
generating a further trained version of the artificial intelligence system.
15. The non-transitory computer-readable storage medium of claim 14 , wherein:
the further trained version of the artificial intelligence system is configured to construct an output that satisfies a combination of the target artist configuration and the further target configuration;
the repurposing of the trained version of the artificial intelligence system further including processing a new input through the further trained version of the artificial intelligence system, and generating a new output that applies the combination of the target artist configuration and the further target configuration to the new input;
the further training dataset requires that at least some training samples in the further training dataset are sourced from a further target artist;
the further target configuration characterizes a work of the further target artist; and
a further target configuration conditions generation of the new output on one or more user-supplied parameters.
16. The non-transitory computer-readable storage medium of claim 15 , wherein the user-supplied parameters comprises one or more of:
a speech type;
an art type;
a current state of the blockchain network, wherein the current state of the blockchain network includes block hashes and surrounding blockchain nodes; or any combination thereof.