Our primary project concerns the creation of a functional AGI system. It consists of a software and a hardware part:
- AGI Software – A trainable AGI system that adapts to its hardware
- AGI Robot – A robotic embodiment for the AGI system
Sub-systems and possible spin-off projects include:
- Computer Vision
- Image Synthesis
- Sound Separation
- Speech Recognition
- Speech Synthesis
- Robot Control
Objectives
The project is about creating a system that demonstrates intelligent behaviour, and that can interact with the real world in a sensible way. Therefore, it needs intelligence, embodiment and initiative.
Intelligence
In our project, we use the term intelligence as the capacity to learn, reason, plan, solve problems, think abstractly, comprehend ideas and understand and use language. The final objective of this project is to create computer software that exposes the above capabilities at a similar level as an average human being and in some areas even can surpass it.
Embodiment
In order to give the system a similar reference frame as a human being, it will be given similar sensors and actuators (cameras, microphones, thermometers, gyroscopes, robot arms, motors, speakers, etc). Idealy, this will lead to the development or purchase of a full-featured autonomous robot.
Initiative
Most AGI systems are like chatterbots; you ask them a question and they respond to that question. However, intelligence usually does not sit around waiting for someone to ask questions, but acts out of its own accord. Therefore, the possibility to take initiative is a major design goal for our AGI system.
Focus
Although the greater picture involves both hardware and software, the project at hand primarily focuses on the software side of the development, being the AGI system. The goal is to create software that exhibits intelligent behavior and that can interact with the real world in a sensible way. Hardware necessities (robotic or other hardware to interact with the real world) will be developed within a project to be executed in parallel, or will be externalized under strict supervision. If full robot development results to be unfeasible, the project will continue with more limited hardware.
It is important to note that it is not the intention to create a perfect simulation of the human mind. The AGI system will obviously have possibilities as well as limitations that a human being has not. Also, the system can do without a lot of things that a human needs due to its biological build-up, whereas it will need other things that a human can do without. The idea is to have a system that lives up to the expectations of strong AI, and that can be used for the tasks it is created for (designs may differ depending on the imposed application).
Although the final objective of artificial general intelligence is highly ambitious, the project has other objectives that, although no less ambitious, are more easy to accomplish in less time. These are spin-off technologies that can be applied in a wide variety of applications. Examples include computer vision, speech synthesis, speech recognition and robot control. Theses spin-offs that are created during the development of the AGI system will supply the project with valuable return on investment.
Schedule
A speculative planning for this project (an initial full-featured commercial AGI system) would cover roughly 8 years. However, an actual schedule cannot be made until after the first year, in which the project will be prepared (including project definition, feasibility study, preliminary investigation, planning, market study, competitors). Furthermore, the first year will be used to outline a strategic and marketing schedule, as well as to consolidate the company (funding, management, advisors, protection of intellectual property, infrastructure).
In order to be there when the market starts to grow to interesting proportions, it is essential to start early. This will ensure that by the time it happens, you have it all: technology, experience, suppliers, infrastructure, patents, prototypes, spin-off and a history in the field.
I expect that in the year 2020 the technology in the field will be such that there are prototypes for several components. In the year 2023 there will be (relatively expensive) prototype toys and industrial applications, in 2025 there will be toys at a reasonable price, and in 2030 the intelligent robots will be part of the world much the same way as computers are now.
Our long term objective:
Create a working AGI system.
Our short term objective:
Create spin-off technology and products.