layout: post title: Challenges around General AI

There are several ways to look at general AI

1. A solution based on AI is consisting of several components, technologies and associated data sets. General intelligence could be an agent with the capability to define required data and procedures to form and integrate a solution for a problem. Replacement of researchers and developers, a challenge.
2. Typical challenges in business are to solve a problem or to find a problem for a technology. AI is a good example for the 2nd, many corporations and KMUs are in search of problems to solve with AI. So another type of general AI could identify or invent problems and hand it over to 1., which can only solve problems. To find a relevant problem is at least as much of a challenge than solving a problem. To define would general AI mean combining 1. and 2.?
3. Third challenge is credibility. The typical method to predict behavior and results of AI is to simulate and evaluate in virtual worlds. There are at least as many virtual worlds as there are autonomous and complex automation projects, because it’s to expensive and to dangerous to test in the real world. Who would test general AI? How would general AI reach credibility and, consequently, social status to enforce general intelligence? Like Turing masturbation, a challenge. To define again would general AI mean combining 1., 2. and 3.?