What does the future of AI and machine learning hold? Will we push past the limitations of using ML models? Is Artificial General Intelligence possible, and can machines learn to grow, think, and act like humans, instead of only learning from provided data sets and scripts? While we may not have all the answers to these questions yet, we spoke with Charles Simon, founder and CEO of Future AI, award-winning, early-stage technology company, about Sallie—a new artificial entity for targeted Artificial General Intelligence (AGI) development. Charles Simon answered our questions about Future AI's latest artificial entity, what Future AI hopes to achieve during beta testing, some potential use cases for the technology, and how AI could potentially "evolve like a person".

devmio: Thank you for taking the time to answer our qusetions. Could you give us a short introduction to Sallie? How does it “evolve like a person”?

Charles Simon: Sallie is an artificial entity for our targeted Artificial General Intelligence (AGI) development. With a computerized “mind” which operates wirelessly to a mobile robotic sensory pod, Sallie learns in real-time with vision, hearing, speaking, and mobility, giving her the ability to draw conclusions – a critical facet of genuine thinking and a necessary component to ushering in AGI. Sallie’s small pods can utilize unlimited computer potential for local learning and adaptation. Future AI includes frequent software updates to enhance Sallie’s capabilities on a continuing basis as new AGI features are rolled out.

Users who opt into Future AI’s beta program will be participating in the most exciting project on the planet—to learn about the basis for human intelligence and implementing similar intelligence in a computer.

devmio: What do you hope to achieve with Sallie?

Charles Simon: Sallie has the beginnings of a mind. She can already see new things and deduce information about them from her previous experience, showing glimmers of true understanding. As development progresses, Sallie will gain more intellectual abilities and, over a period of years, will gain the abilities of a young child. In the same way that a three-year-old child learns, Sallie’s “brain” will interpret everything it knows in the context of everything else it knows.

Users who opt into Future AI’s beta program will be participating in the most exciting project on the planet—to learn about the basis for human intelligence and implementing similar intelligence in a computer.

devmio: How will Sallie work with virtual assistants such as Siri and Alexa?

Charles Simon: Unlike Siri and Alexa, which depend on immense data sets and predefined behaviors or scripts, Sallie has a fundamentally different purpose—to be a general intelligence. Although she doesn’t currently work with digital assistants, she has immense computing power available so this could be added in the future based on interest from our beta users.

devmio: What are some common use cases you can see for Sallie?

Charles Simon: Like a child, Sallie’s initial purpose is to learn and to be entertaining. She will learn the locations of household objects and be able to find them, play games, and interact verbally.

Initially, there will be three configurations of the Sallie distribution:

• 1. for household use and to play games ;
• 1. for children as a STEM game to teach the fundamentals of logic and learning; and
• 1. for college students as a senior-level robotics project.

devmio: How could Sallie potentially improve our lives, and who will it help? Are there any accessibility features?

Charles Simon: The things we learn from Sallie will impact virtually all aspects of our lives. Features of mobility, vision, and object avoidance can be useful in making self-driving cars and other autonomous robots safer. Features of context-sensitive speech recognition could be added to personal assistants. Features of general intelligence could be added to virtually any computer applications to make them more useful.

Although Sallie is fully controllable by voice or small keypad, her real accessibility value will come in a planned future iteration which will be somewhat larger and include an arm for carrying objects.

The things we learn from Sallie will impact virtually all aspects of our lives. Features of mobility, vision, and object avoidance can be useful in making self-driving cars and other autonomous robots safer.

devmio: What makes the sensory pods on Sallie different from other comparable sensors?

Charles Simon: The key to the sensory pod is its simplicity and the amount of feedback between the pod and the “mind.” The computer running Sallie’s mind can access the raw data from all the sensors on the pod (such as motor speeds and position, accelerometers, etc.). This is similar to the way the human brain can get information from any part of the body.

It is also in stark contrast to most robots which attempt to do as much as possible autonomously and can’t transmit as much information. With this architecture, a small amount of processing can be done on the pod itself, more on the wireless controlling computer, and even more in the cloud.

The current favorite son is machine learning and while it has tremendous power, it doesn’t perform functions the way biological brains do and doesn’t seem to be the pathway to general intelligence.

devmio: What do you hope to learn from the beta tests?

Charles Simon: We hope to learn what aspects of Sallie users find unique and entertaining. We will also focus on the more mundane physical reliability of motors, servos, battery-life under actual use, etc.

devmio: Does Sallie tie into your work on Brain Simulator II, an open source AGI toolkit, and if so, how?

Charles Simon: Yes, Sallie is an application which was created using the Brain Simulator II. Many parts of the BrainSim are embedded in it. Most of Sallie’s capabilities are created in BrainSim modules.

Developers who dedicate themselves to a single platform or application often face a significant learning curve if that particular technology shifts away.

devmio: Do you have any tips or advice for machine learning programmers?

Charles Simon: I always suggest maintaining a broad interest in many aspects of software development and always learning because technologies shift. Languages change. OS services change. Even the concept of what a computing platform is may change. Because of these shifts, developers who dedicate themselves to a single platform or application often face a significant learning curve if that particular technology shifts away.

devmio: Have you always been interested in the idea of artificial intelligence and machine learning? How do you feel about how far we’ve come in terms of technological progress?

Charles Simon: I’ve been interested in artificial intelligence since college during the very first explorations into chess-playing programs. Since then, the AI field has been through boom-bust cycles as various techniques have come into vogue and then, as the limitations of the technology become more well-known, funding has dried up.

The current favorite son is machine learning and while it has tremendous power, it doesn’t perform functions the way biological brains do and doesn’t seem to be the pathway to general intelligence.

At Future AI, we are pursuing a graph structure which is fundamentally different and significantly more biologically plausible. We’ve already seen some success with this approach and look forward to solving some of the problems which machine learning has yet to solve.