The only source of duly guaranteed
validations and valuations of your patents
Introducing: the AI-accelerated KX Patent Analyzer,
backed by powerful Patent Defense insurance
|
The only source of duly guaranteed validations and valuations of your patents Introducing: the AI-accelerated KX Patent Analyzer, backed by powerful Patent Defense insurance |
|
THE GROWING CONFLICT BETWEEN AI DESIGN TOOLS AND KSR'S "PREDICTABLE"
Every year, AI design tools become more powerful in terms of being to invent drugs, circuits, mechanical devices, computer programs - completely autonomously. If all the human/inventor/"individual" is doing is to press the "RUN" button, is that invention done by a "Whoever" to satisfy 35 USC 101? If one skilled of a PHOSITA - one skilled in the art - is to use such tools, are any inventions output from the program "obvious" and not patentable as the "predictable" outcomes of the PHOSITA under 35 USC 103?
What follows is a list of papers from PATNEWS and other publications on this issue, followed by a database of papers on AI design tools. Rules based on these development are being built into the Analyzer.
THE USPTO DOES NOT UNDERSTAND 'ARTIFICIAL INTELLIGENCE'
In October 2020, the USPTO released a report, Inventing AI: tracing the diffusion of artificial intelligence with U.S. patents, written by the Office of the Chief Economist of the USPTO. It is a boring statistical analysis of the number of AI patents issued over the years, with the usual pretty-but-uninformative landscape graphs. What is not discussed in the report, because the Chief Economist doesn't understand patent economics, are the following troubling concerns with AI patents:For example, the report cites U.S. Patent 8,930,178, Processing text with domain-specific spreading activation methods, without mentioning the patent is easily worked-around by running the natural language analysis offshore (rendering the Beauregard claims, the patent's only claims, worthless) - indeed, the patent teaches this work-around with the statement: "While the invention would typically be hosted by a server connected to the Internet, ..." -- which includes remote servers outside the U.S.
- The low quality of most AI patents (most cite no non-patent prior art)
- The lack of enforceability of most AI patents (do the AI method offshore)
- Do AI design patents render obvious inventions created by the AI? (this Web page)
- Why doesn't the USPTO use any of the 30 years of AI patents to reduce patent pendency from two years to two weeks?
- Why doesn't the USPTO use natural language AI to measure the incoherency and contradictions of 101 caselaw?
- Can a lawyer who has never programmed AI be competent to write an AI patent?
- As there is no definition of 'intelligence' in the science community, are all patent phrases using 'intelligence' (the NIST defintion is circular), failing of enablement?
ARTICLES ON THE GROWING CONFLICT BETWEEN AI DESIGN TOOLS AND KSR'S "PREDICTABLE"
- PATNEWS: Are outputs of the "Artificial Chemist" KSR-predictably obvious - 12 Jun 2020
- PATNEWS: Is first fully-AI-predicted drug KSR-predictably obvious? - 24 Feb 2020
- PATNEWS: Economist letter: the AI threat to drug development and patentability - 12 Jul 2018
- PATNEWS: Accelerated discovery/inventing - resulting inventions unpatentably obvious results of PHOSITA using (AI) tools? - 21 Jun 2017
- PATNEWS: Are supercomputer-designed drugs KSR obvious and thus unpatentable? - 05 May 2017
- SCRITed: Computers as inventors - legal and policy implications of AI on patent law - Dec 2016
- PATNEWS: How KSR-unpatentable are structured-based drug discoveries? - 16 Sep 2016
- PATNEWS: Are all future battery patents KSR-predictably obvious? - 03 Dec 2015
- PATNEWS: Drug discovery tools, machine learning and KSR: the unpatentability risk worsens - 02 Dec 2015
- PATNEWS: 3D crystal structure automatically determinable - and now unpatentably obvious? - 13 Nov 2015
- PATNEWS: Will drug discovery tools (e.g., IBM's KnIT) make 103 attacks easier? - 22 Oct 2015
- PATNEWS: In-silico drug design - all results obviously non-patentable? - 24 Nov 2014
- THESIS (GWU): Flashes of genius, toiled experimentation, and now artificial creation: a case for inventive process disclosures - argues AI discovered inventions are not patentable - 26 Aug 2011
- BU.J.SCI.TCH.L: Thinking about thinking machines: implications of machine inventors for patent law - 2002
PATENTED AI DESIGN TOOLS: ARE THE COMPANIES SELLING SUCH TOOLS MAKING IT HARDER FOR THE CLIENTS TO OBTAIN PATENTS WHEN USING THE TOOLS?
ISSUED PATENTS
- TCAD design template for fast prototyping of 2D and 3D CMOS image sensors - U.S. 10,528,684
- Method for producing and screening mass coded combinatorial libraries for drug discovery and target validation - U.S. 6,207,861
PATENT APPLICATIONS
- System and methods for machine learning for drug design and discovery - U.S. 2019/304568
- Single-molecule platform for drug discovery: methods and apparatuses for drug discovery, including discovery of anticancer and antiviral agents - U.S. 2018/195104
- Rapid identification of pharmacological targets and anti-targets for drug discovery and repurposing - U.S. 2017/147743
- Drug discovery methods for Aurora kinase inhibitors - U.S. 2011/269732
PAPERS ON AI DESIGN TOOLS: THEIR TECHNOLOGY AND INVENTIVE POWER
- RetroComposer: discovering novel reactions by composing templates for retrosynthesis prediction
- Univ. Texas, Decemebr 2021
- note: this is interesting, because they used a database of 50,000 reaction descriptions taken from U.S. patents to predict new reactions, making it hard to argue that these AI tools don't impact patent law- An in-depth summary of recent AI applications in drug design
- Univ. Minnesota, October 2021- ChemiRise: a data-driven retrosynthesis engines, that can propose complete retrosynthesis routes for organic compounds, trained with a patent database
- Accutar Biotechnology, August 2021- Unique AI method for generating proteins will speed up drug development using generative deep learning which uses a large amount of data from well-studied proteins; it studies this data and attempts to create new proteins based on it.
- EurekaAlert, 30 March 2021- An accessible machine-learning tool has been developed that can accelerate the optimization of a wide range of synthetic reactions -- and reveals how cognitive bias might have undermined optimization by humans
- Nature, 03 February 2021- Bayesian reaction optimization as a tool for chemical synthesis
- Nature, 03 February 2021- Deep learning model for finding new superconductors
- Univ. of Tokyo, 14 January 2021- Machine learning accelerates discovery for use in industrial processes
- EurekaAlert, 11 January 2021- Computational planning of the synthesis of complex natural products
- Nature, 13 October 2020- Scientists at MIT have developed an AI design program for robots, called RoboGrammar, which given a list of available components and description of the terrain to be navigated, generates an optimized structure and control program for a robot fulfilling those constraints
- EurekaAlert, 30 November 2020- Google's DeepMind AI program makes a gigantic leap in solving protein folding problems, that is, determining a protein's 3D shape from its amino acid sequence. It will enable quicker and more advanced automated drug discovery.
- Nature, 30 November 2020- A group at the University of Glasgow has developed software that turns academic papers into chemputer-executable programs that researchers can edit in plain English without learning to code.
- CNBC, 24 Oct 2020- We develop MolDesigner, a human-in-the-loop webuser-interface (UI), to assist drug developers leverage DL predictions to design more effective drugs. A developer can draw a drug molecule in the interface. In the backend, more than 17 state-of-the-art DL models generate predictions on important indices that are crucial for a drug's efficacy. Based on these predictions, drug developers can edit the drug molecule and reiterate until satisfaction. MolDesigner can make predictions in real-time with a latency of less than a second. - arXiv, 05 Oct 2020
- Artificial intelligence is throwing battery development into overdrive. Improving batteries has always been hampered by slow experimentation and discovery processes. Machine learning is speeding it up by orders of magnitude. - WIRED, 12 Oct 2020
- A group of Russian scientists at the Skolkovo Institute of Science and Technology used machine learning methods to predict superhard materials based on their crystal structure - EurelaAlert, 10 Sep 2020
- AI researchers at CMU and Univ. Calgary have created AI algorithm that will find the right material with desired properties for any project, discovering new materials automatically among the immense phase and compound spaces of materials - Physical Review Materials, 08 Sep 2020
- An AI drug design tool, Synthia, quickly designs 11 potential manfacturing processes for generic COVID-19 drugs (umifenovir and favipiravir), that use cheap, readily available starting materials - Science, 07 Aug 2020
- A mobile robot autonomously operates analytical instruments in a wet chemistry laboratory, performing a photocatalyst optimization task much faster than a human would be able to. - Nature, 09 Jul 2020
- Robotic 'scientists' will speed up the discovery of new drugs and chemicals - BBC, 06 Jul 2020
- Multifunctional meta-optic systems: inversely designed with AI - Georgia Tech tech report, 02 Jul 2020
- Artificial Chemist: an autonomous quantum dot synthesis bot - Advanced Materials, 04 Jun 2020
- Halicin, a powerful antibiotic discovered using machine learning for the first time at MIT - The Guardian, 20 Feb 2020
- A generalized method toward drug-target interaction prediction via low-rank matrix projection - 03 May 2018
- Directory of computer-aided drug design tools - Swiss Institute of Bioinformatics, 05 May 2018
- AI promises life-changing Alzheimer's drug breakthrough - Newsweek, 29 Jan 2018
- Recent advances in accelerated discovery through machine learning and statistic inference - Ann. Rev. Phys. Chem, 16 Jun 2017
- Supercomputers at the Univ. of Texas assist in search for new, better cancer drugs - TACC News, 01 May 2017
- Structure-based automated discovery of opioid analgesics with reduced side effects - Nature, 17 Aug 2016
- Machine learning, quantum mechanics and chemical compound space - Univ. Basel, 12 May 2016
- High througput pKa prediction using semi-empirical methods - Univ. Copenhagen, 03 Dec 2015
- Software predicts a slew of fiendish crystal structures - chemists succeed at forecasting how complex moleclues will assemble in 3D - Nature, 03 Nov 2015
- Cancer in silico drug discovery: a systems biology tool for identifying candidate drugs to target specific molecular tumor subtypes - Mol. Cancer. Thera., Dec 2014
- Development of a novel class of B-RafV600E-selective inhibitors through virtual screening and hierarchical hit optimization - Organic & Biomolecular Chemistry, 13 Jul 2012
- In Silico pharmacology for drug discovery: methods for virtual ligand screening and profiling - Br. J. Pharmacology, Sep 2007
The Shadow Patent Office, Mall Plaza Pance, Calle 18 #350, Avenida Cañagordas, Cali, Colombia
(c) 2021-2022 - The Shadow Patent Office (part of KukaXoco Colombia S.A.)