US 12,755,182 B2Granted

Partial quantum mirror mode for artificial intelligence (AI) models

AssigneeBank of America Corporation

InventorsManu Kurian, George Albero

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems, methods, and apparatus are provided for remediating an AI hallucination and determining and limiting excessive branching. An AI query may be received at multiple processors including a quantum processor or at a quantum processor having multiple threads, and an AI search may be executed at multiple processors or on multiple quantum threads. A continuous hashing algorithm may hash the AI search data and partially mirrored AI search data and compare the hashes. When the hashes are not identical, the partially mirrored AI search data may be deleted. The AI search may be terminated and reinitiated at the last point the hashes are identical. The AI search data may be partially mirrored at the point that the search is resumed. The results of partial mirroring may be fed back to update the AI model.

Claims20

Dependent claims

A method for monitoring performance of an artificial intelligence (AI) system by partial mirroring of data in a quantum computing environment, the method comprising:

performing, by a processor using a first AI engine, a first AI operation in response to a first search request to generate a first data stream comprising first data segments;
performing, at a quantum computing system comprising a quantum processor and using a second AI engine, a second AI operation in response to a second search request to generate a second data stream that partially mirrors the first data stream and comprises second data segments that together correspond to less than all of the first data stream;
wherein:
the quantum processor processes data as a plurality of qubits;
the first AI engine and the second AI engine use a same AI model;
and
each of the first data segments and the second data segments is associated with a respective time stamp;
and
determining whether one or more of the second data segments are being partially mirrored or are diverging from one or more of the corresponding first data segments in the first data stream by:
hashing a respective one of the first data segments that includes a time stamp to obtain a first hash value;
hashing a respective one of the second data segments that corresponds to the time stamp to obtain a second hash value;
and
comparing the first and second hash values to determine whether the first and second hash values are matched or mismatched.

10 more elements

One or more non-transitory computer-readable media storing computer-executable instructions, which, when executed on a processor on a computer system, perform a method for monitoring performance of an artificial intelligence (AI) system by partial mirroring of data in a quantum computing environment, the method comprising:

performing, at a classical computer comprising a first processor, an AI operation using a first AI engine in response to a first search request to generate a first data stream comprising first data segments;
performing, at a quantum computing system comprising a quantum processor, the AI operation using a second AI engine in response to a second search request to generate a second data stream that partially mirrors the first data stream and comprises second data segments that together correspond to less than all of the first data stream;
wherein:
the quantum processor processes data as a plurality of qubits;
the first AI engine and the second AI engine use a same AI model and are configured to perform the AI operation;
and
each of the first data segments and the second data segments is associated with a respective time stamp;
hashing a respective one of the first data segments that includes a time stamp to obtain a second hash value;
hashing a respective one of the second data segments that corresponds to the time stamp to obtain a first hash value;
and
comparing the first and second hash values to determine whether the first and second hash values are matched or mismatched;
and
when the first and second hash values match, continuing to partially mirror on the quantum computing system the one or more of the first data segments in the first data stream, perform hashing on the one or more first data segments in the first data stream and one or more of the second data segments in the second data stream to monitor whether results of the AI operation that is performed at both the classical computer and the quantum computing system are matching or diverging.

10 more elements

Dependencies are inferred from the claim text.