The Singularity Just Got A Lot Closer

New tool allows programmers to build artificial intelligence into almost any software application.

SDK for Machine Learning

A new technology enables almost any application to learn like a human. The Topic-Mapper software development kit (SDK) by ai-one inc. reads and understands unstructured data without any human intervention. It allows developers to build artificial intelligence into almost any software program. This is a major step towards what Ray Kurzweil calls the technological singularity – where superhuman intelligence will transform history.

Unlike other machine learning approaches, ai-one’s technology extracts the inherent meaning of data without the need for any external references. A team of researchers spent more than eight years and $6.5 million building what they call “biologically inspired intelligence“ that works like a brain. It learns patterns by reading data at the bit-level. “It has no preconceived notions about anything,” explains founder Walt Diggelmann, “so it works in any language and with any data set. It simply learns what you feed it. The more it reads, the more it learns, the better it gets at recognizing patterns and answering questions.”

Technical Advances

Lightweight Ontologies (LWO)

The technology incorporates two major technical advances: First, it automatically creates what ai-one describes as a “lightweight ontology” (LWO), The system determines the relationships between data elements as they are fed into the system. The primary benefit of LWO is that it is completely objective — it makes associations without editorial (human) bias. LWOs are also very adaptive, automatically recalculating when ingesting new data. Unlike traditional ontologies, LWOs require no maintenance.

Dynamic Topologies

Second, ai-one’s technology generates “dynamic topologies” that transforms the data structure to find the best answer to any question. The benefits of dynamic topologies include incremental learning – the system gets smarter as it is exposed to more questions. Moreover, it is able to deal with ambiguity and unknown situations. The result is that the system can answer the questions that a person wouldn’t normally know to ask.

The SDK opens the door for many new, disruptive software applications. For example, it can replace search algorithms with more accurate “answer engines” that deliver the most precise answer to any question.

“We offer a core programming technology,” says Tom Marsh, President and COO of ai-one. “The possibilities are almost endless. Our business model is to license the SDK to software developers to build end-user applications. Our goal is to get Topic-Mapper to as many well-qualified programmers as possible and let the creativity of the market take over.”

Adoption has been quick.

The first version of the SDK was released in February 2011. In less than three months, more than 20 consulting partners signed up to use the technology to build commercial applications – mostly in Europe. Swissport matches passenger manifests against the US Department of Homeland Security’s No-Fly List. The core technology is used by Swiss law enforcement CSI labs to match shoeprints and other evidence from multiple crime scenes. Most recently, ai-ibiomics announced it will use ai-one’s SDK to read genome sequences to provide personalized medical services in Germany.

A logical next step is for the technology is to enable eCommerce, social media and other online applications to provide end-users with the most relevant, most accurate information for any given situation.

ai-one will be showcasing the technology at booth #107 during the SemTech 2011 conference on June 7-8. Developers can request a 30-day evaluation copy online at http://www.ai-one.com/solutions/semtech2011.

Tags: , , , , , , ,

Comments are closed.