When humans envision future applications of artificial general intelligence, we tend to think of pop culture examples like Jarvis from Iron Man — the ultimate automated assistant. We can easily imagine how helpful that kind of technology would be, but we still write it off as complete science fiction.
At Mind AI, we are building a reasoning engine that will eventually be as capable as Jarvis. But before we get there, we will be able to put our patented approach to AI to use in many exciting applications that will bring about superhuman capabilities.
How Mind AI works
Mind AI is a reasoning engine. Unlike today’s machine learning algorithms, Mind AI does not learn by recognizing patterns in massive data sets. Mind AI takes in knowledge piece-by-piece in the form of ontologies. An ontology, according to Merriam-Webster, is “a particular theory about the nature of being or the kinds of things that have existence.” The genius of Mind AI is its ability to use the ontologies in its knowledge base to correctly interpret new information.
Future applications built around the Mind AI reasoning engine will rely on four categories of ontologies: Global, Local, User, Session. Global ontologies are the foundational ontologies that make up Mind’s base intelligence and ability to learn, for example, “Humans are mortal,” or “green is a color,” or “gravity makes objects that have mass fall to the ground.”
Local ontologies are domain-specific and are sourced from volunteer contributors around the world. User ontologies are pieces of knowledge specific to the user that are contained within the application that is using the reasoning engine and are not shared with other users. Session ontologies are pieces of knowledge supplied during one session that Mind can factor into its reasoning, but it does not retain that knowledge after the session ends. These categories allow Mind AI to be infinitely customizable while still protecting individual privacy.
The initial use cases for Mind AI will be based on its ability to read and interpret texts. As long as Mind AI has enough general knowledge about the relevant subject matter (from crowdsourced local ontologies), it will be able to read and comprehend any text and hold unscripted conversations about that text with a human. These conversations can happen in real time using natural language — no coding or special formatting required. Unlike a human, Mind AI will have perfect recall of everything it has ever learned relevant to the topic at hand.
This capability has applications in a wide variety of industries. We’ll outline a few of them here, to illustrate the incredible potential of this technology.
5 potential use cases of Mind AI
Automated call centers for small and large businesses
The call center industry is a multi-billion dollar industry, with some estimates putting it around $200 billion dollars. Many businesses have experimented with ways to use artificial intelligence to improve customer service, but they all come down to creating a decision-tree — ask the customer standardized questions and provide a limited set of standardized responses. Even when voice-recognition technology is used, the customer experience is still typically negative. We all want the freedom to talk to a real person using natural language, asking whatever questions we want.
A customer-service bot run on Mind AI would be completely different. How do you train a human to work in customer service? You provide reading materials about the products they are going to support. Mind AI will be able to read a customer service manual and comprehend it just as well as a human customer service rep. It will also be able to hold conversations about the material in the manual using natural language (in any spoken language) — not a script or a decision-tree.
E-commerce shopping assistants
Amazon’s Alexa is already a very popular shopping assistant, but it is not a reasoning engine. It’s a voice-operated keyword search engine with a limited number of commands it can recognize. It can provide an overwhelming amount of data about a product, but it can’t help you make sense of that data.
Mind AI could improve the shopping experience by answering complex questions about products, helping a user compare different products based on both the specifications and user reviews, and making personalized recommendations. Those recommendations would be based on user and session ontologies that the user provides access to, like personal data, purchasing history, or social media accounts. Drawing on these sources, Mind AI could take into account the user’s personal style and preferences instead of relying on generic values and trends.
Due diligence in asset management
Venture capitalists and investors spend a lot of time reading for market research and due diligence. Before they decide to invest in a company, they need to read a mountain of materials about the company — analysts’ projections, patents, founder bios, industry trends, etc. If the investor had access to Mind AI, they could feed all of the materials they would have read into the reasoning engine, and then they could hold a conversation with Mind AI about the research instead of reading each document. This kind of collaboration could dramatically cut down the time it takes to do due diligence.
Virtual education assistant
One of the most important skills children develop during school is the ability to read complicated texts and understand and interpret them. In a traditional educational setting, students read the text, and a teacher, who is very familiar with the text, leads a discussion about the text. When students don’t immediately grasp the central theme of a piece of literature, the teacher explains it, showing which passages support that theme.
Mind AI could act as a one-on-one reading comprehension tutor, teaching students how to understand and interpret texts. A student could ask, “What is the central theme of the first chapter of this book?” Mind AI would reply based on its interpretation of the text, and the student could then ask, “How do you know that?” Mind AI would then be able to reveal its reasoning, pointing to the passages that support that theme, and explaining why.
Medical assistant
When doctors are faced with medical mysteries, they often seek out additional educational materials. They look for clinical studies that might shed light on the patient’s condition. This involves a massive amount of reading. Mind AI could help the doctor find and interpret the relevant research faster, through a conversational process.
In this case, Mind AI would need a strong foundation in medical terminology, which could be supplied by ontologists with a background in medicine. Once it had a basic understanding of medical terminology, Mind AI could read the same textbooks that human medical students read, becoming an invaluable medical assistant. And when Mind AI makes recommendations, it would be able to trace its reasoning path, giving the human doctor a full list of studies and research results backing up its conclusions.
Looking forward to powerful collaborations
These are just some of the many early applications we have imagined for our reasoning engine. These tools will enhance each individual’s ability to learn, to make smart decisions, and to get personalized help on-demand. We are proud to be developing these tools in a way that will make them accessible and affordable to all.
As development continues and the reasoning engine becomes more intelligent, the possible applications are endless, limited only by the imaginations of the community members guiding and contributing to its growth.
By supplying ontologies in a specific domain, community members influence which kind of applications can be created. We would love to hear what kind of applications our followers would like to see using this powerful reasoning engine. Share your ideas on our Telegram channel, or Tweet at us, or email us directly at info@mind.ai.
To stay up-to-date on our progress, sign up for our email list or follow us here on Medium.
