Artificial Intelligence has an increasing impact on more and more industries. With its increasing impact, AI is applied in various different ways including the new AI-application also known as: AI built by AI. As the name gives away, an AI technology has the arsenal to create other AIs. A leading company in this new concept is Google, specifically their department Google Brain (Galeon, 2017). Google brain created AutoML in 2017 which is the main AI-application that serves as the controlling neural network. AutoML is able to create a “child” known as NASNet. NASnet is able to execute simple tasks like categorizing images and recognizing objects. With this new creation, Google already states that their new AI-application is a lot more efficient and powerful, outperforming most of the best human-designed systems. The accuracy rate by NASNet was 82,7%, which overall outperformed similar models that already exists by 3,1%. The AI built by AI can be used in various different industries, from the transportation industry (where it can improve safety by recognizing objects faster and reacting to it faster) to the health care industry (where it can help diagnose certain diseases by recognizing certain symptoms).
However, there are two main limitations to this concept. Firstly, the rapid pace of change is hard for society to keep up with. Therefore, it’s important that human beings act as the gatekeepers to determine the pace and to oversee the overall process. This gatekeeper-duty is a lot easier, now that the human engineers spend less time actually creating the systems. Secondly, the AI teaches its “child” itself meaning it is much easier to be biased (e.g. negatively associating certain ethnic and gendered identities with each other) (Neff, 2020). That’s why it’s very important to keep an eye on the ethical side of the AI systems. “The Partnership on AI to Benefit People and Society”, which includes members such as Google and Amazon, and various governments do a great job protecting the ethical side in the technological industry.
Galeon, D., 2017. Google’s Artificial Intelligence Built An AI That Outperforms Any Made By Humans. [online] Futurism. Available at: <https://futurism.com/google-artificial-intelligence-built-ai> [Accessed 25 September 2020].
Neff, G., 2020. Five Things You (Maybe) Didn’t Know About AI. [online] Google. Available at: <https://blog.google/technology/ai/five-things-a-to-z-ai/> [Accessed 29 September 2020].
The AI ‘child’ could and is improving functions in many industries. In the short term, it has major advantages. You are stating that governments do a great job in protecting. However, as I understand, this differs per country or for example the European Union. Is this true? And if it is, how can other countries be regulated that have less incentive to limit AI, because it could improve their servives?