Gamalon, Inc. was originally funded by DARPA (Defense Advanced Research Projects Administration) to collaborate with Stanford and MIT to create AI that can be relied upon to follow orders. By combining large deep learning models with state-of-the-art control, we enable AI for mission-critical applications where fact and reliability are on the line.
For example, our technology is being used by Product Genius, a product line that brings the power of the most advanced AI to small and medium-sized businesses so that they can compete with online marketing and sales against the tech giants. We help stores provide their visitors with great information so that shoppers can make good buying decisions more easily. This information needs to be truthful and on-brand. In scientific A/B tests, Product Genius has been proven to provide a 50%-300% increase in shopper engagement and it often provides a 20% or greater revenue lift for the online stores that use it.
By contrast Meta (née Facebook) recently trained a large language model on the world’s scientific papers. As experts interacted with it, it rapidly became clear that there was a serious problem. Meta’s AI was so inaccurate and misleading in every topic, but yet convincing in a way that would fool any non-expert, that it led to a massive outcry from the scientific community. Meta took it down after two days. https://www.cnet.com/science/meta-trained-an-ai-on-48-million-science-papers-it-was-shut-down-after-two-days/
Like the military or scientists, businesses do not want their brands or operations compromised by inaccurate information. They want AI that is controllable. That’s what DARPA originally funded us to do, and it’s our unique capability. Gamalon’s technology is available for licensing.