"Artificial Intelligence" is an over-used term, commonly used for pattern recognition abilities acquired via Deep Learning.
Our company is named after the concept of Additional Intelligence: (1) Added information and (2) AI as a tool that can autonomously learn, reason and communicate - but not autonomously execute. A tool for the augmentation of human intelligence and knowledge, instead of a black-box module in a software system that "does things", such as recognize faces and notifies the Police when it thinks it saw a wanted individual, or recognize the road and its traffic and operate the controls of a car.
Our system lets R&D staff in STEM fields "brainstorm" via uniquely presented cross-correlated data from the relevant patent- and scientific literature, for example. Or it gives medical specialists or attorneys instant access to relevant data such as novel diagnosis- and treatment methods or jurisprudence.
The "Additional Intelligence" our system provides goes far beyond mere keyword search. Syllogic™ - our AI system - self-learns synonyms in your field and thus provides excellent search capabilities but where it really shines is its ability to absorb vast amounts of topical data, extract all knowledge-concepts and classify them into their proper categories.
Syllogic learns autonomously but also is able to communicate in English about what it learned. It has an actual, human-level understanding about the data it has been provided with, which means you'd be able to ask it: "List all desalination methods" and the system will instantly present a list with more methods than Wikipedia provides. With links to the scientific literature and patents, of course. But Syllogic's true strength lies in its ability to let you instantaneously "fly" through the entire cross-correlated relevant data-space. You're developing a room-temperature superconductor? Instead of spending ages searching for what your colleagues have been doing lately, Syllogic lets you "fly" through the distilled data and take turns, follow leads, dig into the details.
Syllogic keeps itself up-to-date by constantly downloading and absorbing the data from new patents, new scientific literature and other sources.
Since usage scenarios vary greatly and since the software has not reached the "shrinkwrap" stage yet, our company will develop modules to suit your requirements.
The system does not run on our servers or in the cloud. You will have full control and privacy over your data and how you use the system. Neither does it "phone home". Just like our own company, our customers operate in fields where valuable IP is closely guarded. The typical hardware cost of a high-powered system is approx. $50,000, with two RAID6 arrays (50 to 250 TB storage typical), UPS and tape backup providing the safety of redundancy and 1 TB RAM enabling instantaneous navigation through the Knowledge Hyperscape. Terabytes of learning data is efficiently processed by massively multi-cored processors.
Our R&D machine with 1 PB of storage in 3 x RAID6 and 2 TB RAM
Frank de Groot, founder of Additional Intelligence LLC, has been writing software professionally since 1981. In 2005, Microsoft Research Cambridge published a paper titled Baysian Pattern Ranking for Move Prediction in the Game of Go where they detail how they implemented a basic version of Mr. de Groot's previously published "Deep Learning" pattern system that learned millions of patterns from Go game records, and they specifically mention that their goal was to implement Mr. de Groot's Go pattern learning system but that they failed to achieve the same level of performance. On March 20, 2009, the recruitment dept. of Google Engineering asked Mr. de Groot whether he was interested in a position at Google. DeepMind's AlphaGo, also from Cambridge and now acquired by Google, utilizes an embodiment of Mr. de Groot's pattern system to prune and direct its search tree.
Mr. de Groot's Go software was written in the period 2002 to 2006 and can be downloaded here (325 MB, for MS Windows).