Imagination Engines, Inc. (IEI) is a rapidly expanding company based upon a new and patented artificial intelligence paradigm called the “Creativity Machine,” wherein artificial neural networks autonomously engage in brainstorming sessions with one another to invent new ideas and plans of action. People such as Dennis Bushnell, NASA Langley’s chief scientist, have called this idea of “dreaming neural networks” monitored and controlled by other more “alert nets,” the best bet for those working in artificial intelligence (AI) to create human to transhuman level intelligence in machines.
Imagination Engines CEO Stephen Thaler earned a PhD in physics and has advanced degrees in fields such as artificial intelligence and chemistry. Employed by the now defunct aerospace manufacturer and defense contractor McDonnell Douglas for 15 years, he worked on diverse problems that included nuclear and laser interactions with materials and systems, stealth technology, and the use of high-energy lasers to produce diamond coatings.
The idea for the current venture stems from Thaler’s first experiments with neural networks in 1975 where he subjected an artificial neural network to progressively higher levels of heat-like computational disturbances to its interconnections. A natural neural network can be defined as any network of neurons or nuclei that are connected by synapses and function together to perform some function in the body. Its artificial counterpart is computer architecture that Thaler says “can be broadly summarized as collections of ‘on-off’ switches that automatically connect themselves so as to build intricate computer programs, without the need for human intervention.” What Thaler discovered was that at precisely tuned levels of such disturbances, these neural nets produced potential ideas that transcend their direct experience. It occurred to him that by allowing another neural net to monitor the former and micromanage the disturbances therein, an invention machine, vastly superior to genetic algorithms, was formed that could function within any field of human endeavor.
The value proposition is twofold: first, machine intelligence working every day, 24 hours a day, and a million times faster than humans, could crank out discoveries heretofore unknown to mankind, and second, tasks traditionally requiring very expensive and slow human contemplation, could now be carried out with at most a tenth of the outlay of time and effort. Thaler says that the TAM could include “everything imaginable” and perhaps be at the level of gross national products (GNPs). I don’t buy this, needless to say, and recommend that Thaler uses my Clarify Your Story framework to figure out how to pursue this business opportunity.
The company says that a great number of innovations have already occurred over the past 15 years of IEI operation that include everything from the invention of advanced toothbrush designs to an ultraintelligent control system for a constellation of military communications satellites. IEI Products include PatternMaster, a neural network trainer that visualizes problems; Agenda, a tool that allows users to isolate records in a database according to complex criteria; tailored robot brains that can learn; robotic simulation environments that can be used to build, train, and test battlefield robots; and advanced machine vision systems that are currently used for intelligent auto headlight control.
With regard to market dynamics, in 1994, Thaler experienced in microcosm at McDonnell Douglas, what is happening today with the economic downturn. McDonnell Douglas management eschewed the notion of lavish scientific budgets to theorize and experiment. They wanted pragmatic results and “no science fair experiments.” In direct response to this new and frugal attitude, Thaler aimed to prove the merits of conventional artificial neural networks in finding optimal process and manufacturing sweet spots within the aerospace industry. He believes that such ideational pioneering can be done on a shoestring using IEI’s core patents of brainstorming artificial neural nets.
Currently, the top market segments targeted are defense, homeland security, and alternative energy. Commercial customers have included Gillette, Bekaert, Magna, GE, Boeing, Veridian, GD, Anheuser-Busch, and and BAH with contracts in the $10,000 to $100,000 range. Among the company’s government customers have been the Air Force Research Laboratory, NASA, TSA, DHS, and NIST, with contracts in the $100,000 to $1 million range.
Over the past four years the company has grossed approximately $1 million a year, with most of that originating from government contracts and collaborative projects with the automotive industry. Government projects have generated a 10% ROI, while the automotive projects have generated 50% ROI. The most lucrative projects have involved corporate customers and partners.
The company’s primary investor has been the federal government, through Phase I and II Small Business Innovation Research (SBIR) awards and contracts, creating more business ideas and products based upon the IEI patent suite. The problem now is that the technology is more expansive than before and the company’s focus even more diffuse. The solution to being so potentially big (ill-defined?) was to form a subsidiary LLC called Imagitron, which now serves as a platform company that incubates and accelerates new business ventures based upon both the accomplishments and the intellectual property of its parent, IEI. This new company is chartered to carve out viable corporations that team with developers and investors to produce more targeted products and services for which definitive business plans may be written. IEI has already created such companies devoted to renewable energy, military robotics, and homeland security. In Michigan, the state Senate is proposing to set aside $18 million in support of one of these platform companies and set up “Centers for Creative Artificial Intelligence Excellence” to create jobs in that state.
Thaler says that there are no plans for an exit at present, and various strategies are being considered.
The business strategy looks convoluted to me, even though it seems that the company has some neat technologies. The results are self-evident. The company only makes $1 million a year after being in existence for 15 years. It needs focus, it needs strategy, and it needs to stop trying to be everything to everyone and become something meaningful to a group of customers who are willing to pay significant money for their offering. While I really like AI technologies, this is the kind of spray and pray that has earned the field its poor reputation in the business world.
Forbes Column 2009: A Time For AI
This segment is a part in the series : The 1M1M Deal Radar 2010