Machine content generation
View Patent ↗Computerized systems and methods are disclosed to generate a document from one or more first and second text prompts, generating one or more context-sensitive text suggestions using a transformer with an encoder on the text prompts and a decoder that produces a text expansion to provide the context-sensitive text suggestions based on the one or more first and second text prompts by applying generative artificial intelligence with token biased weights for zero-shot, one-shot or some-shot generation of the artificial intelligence context-sensitive text suggestions from the one or more first and second text prompts.
1. A method to generate a document having a predetermined outline, comprising:
providing a user interface to receive a first text to plan a document structure from the first text, wherein the document structure comprises the predetermined outline;
searching one or more sources to locate one or more second texts matching the first text;
providing the first text and the one or more second texts to automatically expand into one or more artificial intelligence context-sensitive texts using a transformer with a decoder that produces a text expansion to provide the context-sensitive text based on the first text and the one or more second texts by applying generative artificial intelligence with normalization and tokenization with zero-shot, one-shot or some-shot generation of the one or more artificial intelligence context-sensitive texts from the first text and the one or more second texts, wherein the transformer receives a stream of tokens with an attention mask at one or more self-attention layers, and a causal mask is used for the text tokens;
generating the document with the predetermined outline by combining the generated artificial intelligence context-sensitive text based on the predetermined outline and the first text and the one or more second texts, wherein the first text or the one or more second texts comprises a chapter or section overview; and
wherein the generated document comprises the chapter or section outline expanded with the generated artificial intelligence context-sensitive text.
2. The method of claim 1 , wherein the generated document comprises a fiction work, a non-fiction work, a computer readable code, a machine specification, a patent application, a blog, a legal document, a response to a query, a response to a search, a marketing content, or a mechanical description.
3. The method of claim 1 , wherein the document structure comprises one or more figures, wherein each of the one or more figures comprises a brief description of the drawing, a figure description overview, and an artificial-intelligence-generated detailed description of the figure from component texts corresponding to items in the figure.
4. The method of claim 1 , further comprising combining a title and a machine generated background text with the first text and providing the combined title, background, and the first text to a learning machine to synthesize artificial intelligence context-sensitive text.
5. The method of claim 1 , further comprising: extracting one or more reference signs from a figure and annotating each the one or more reference signs with a part name; and
forming an artificial intelligence-generated reference text suggestion from the one or more reference signs.
6. The method of claim 1 , wherein the transformer comprises a generative pre-trained transformer (GPT) model or a BERT (Bidirectional Encoder Representations from Transformers) model.
7. The method of claim 1 , further comprising:
determining when component, module, code, data structure, or image perform a similar task and showing the determined component, module, code, data structure, or image to a user and breaking-down the one or more second texts into one or more alternate components with different component text and show the one or more alternate components as an artificial-intelligence-generated design satisfying the first text or the one or more second texts.
8. The method of claim 1 , further comprising:
generating a part list by detecting noun phrases (NPs) in the generated document and corresponding numbers for the NPs.
9. The method of claim 1 , wherein the generated document comprises a response to a query.
10. The method of claim 1 , wherein the generated document comprises a patent application.
11. The method of claim 10 , further comprising from one or more sets of input text, generating artificial intelligence context-sensitive text for at least one of: a title, a background, a summary, a description, additional claims and an abstract.
12. The method of claim 10 , further comprising:
detecting at least one of: an error, an antecedent basis error, a claim support error, and a functional language.
13. The method of claim 10 , further comprising:
generating one or more figures or drawings from a user specification.
14. The method of claim 1 , further comprising:
responding to the first text by searching for one or more responsive documents to supplement the first text and applying the generative artificial intelligence to provide a search result.
15. The method of claim 1 , further comprising: generating a multimedia file, a blog, a book, a story, a legal document, a business document, a plan, or a chatbot response.
16. The method of claim 1 , further comprising:
receiving a document as a the first text and applying generative artificial intelligence to generate a response to a query based on the document.
17. The method of claim 1 , further comprising iteratively searching the first text or the one or more second texts.
18. A system to generate a document having a predetermined outline, the system comprising:
a processor; and
a data storage storing computer readable code, that when executed by the processor, cause the system to perform operations, the operations comprising:
providing a user interface to receive a first text to plan a document structure from the first text, wherein the document structure comprises the predetermined outline;
searching one or more sources to locate one or more second texts matching the first text;
providing the first text and the one or more second texts to automatically expand into one or more artificial intelligence context-sensitive texts using a transformer with a decoder that produces a text expansion to provide the context-sensitive text based on the first text and the one or more second texts by applying generative artificial intelligence with normalization and tokenization with zero-shot, one-shot or some-shot generation of the one or more artificial intelligence context-sensitive texts from the first text and the one or more second texts, wherein the transformer receives a stream of tokens with an attention mask at one or more self-attention layers, and a causal mask is used for the text tokens;
generating the document with the predetermined outline by combining the generated artificial intelligence context-sensitive text based on the predetermined outline and the first text and the one or more second texts, wherein the first text or the one or more second texts comprises a chapter or section overview; and
wherein the generated document comprises the chapter or section outline expanded with the generated artificial intelligence context-sensitive text.
19. The system of claim 18 , wherein the operations further comprises iteratively searching the first text or the one or more second texts.