Applying Cybernetics Principles for Responsible Generative AI

The landscape of artificial intelligence is constantly changing and becoming more integral to the way we operate as a society. Applying cybernetics principles seems to be a promising and worthwhile opportunity to maximize the potential of generative AI systems. What are the benefits and challenges of this marriage between two trending and significant technology topics, and how can it provide an avenue for helping us to understand better the complexity of AI’s creative and ever-evolving capabilities?

Cybernetics Explained

To consider if cybernetics principles are an asset to AI, we first need to understand what they are. Cybernetics is a term that was coined by pioneer and mathematician Norbert Wiener in 1943, from the Greek word for steersman or pilot. It can be defined as the study and science of the communications and controls of computer systems, and how they regulate themselves to work toward the goals that they were designed for based on the feedback they are programmed to receive.

The Rise of AI

It may be surprising to some, as artificial intelligence seems to be an idea that has come about much more recently, but in fact the theories which drive it have been around for quite some time. In 1945, Vannevar Bush published As We May Think, proposing a vision of a future where computers assisted humans. In 1950, Alan Turing submitted Computing Machinery and Intelligence, where he discussed the building of intelligent machines and how to measure their learning. A year later, Dietrich Prinz developed a chess-playing program that could determine the best move if it were two moves away from a checkmate.

So, while AI concepts are not new, our increasing technological capabilities provide the means to create and assess in a way that allows those ideas to become reality. Because of this, it is vital that we consider the best way to govern and monitor the behavior of these systems as to get the most out of what we are creating them for and to bridge any ethical issues that may arise.

Generative AI

Generative Artificial Intelligence, such as ChatGPT, can take available raw data and use it to generate images, text, and other content based on the prompts that are provided to it. For example, say “create a painting in an impressionist style of a cat riding a carousel” and it will use all of the data it finds across the internet to generate one. Using the AI Arta application, that prompt gave me the below image:

You might prompt ChatGPT to write a poem about potato chips, which gave me the very humorous one here:

One can surmise that access to this technology is beneficial for innovative content generation, but there still has to be a creative idea behind the prompt that it is fed.

Other uses and benefits of generative AI include using it to help diagnose patients of their ailments because it can take every bit of health data and provide possibilities in seconds. It can be used by businesses to instantly look at years of reports and research to provide summaries and ideas on how to improve customer success or the bottom line. It can help to design products that meet the needs of clients, personalize experiences for customers, and help to streamline processes so businesses can operate more efficiently.

Applying Cybernetics to Generative AI

When we look at all of the ways we can use Generative AI, we also need to consider the learning models that the tools themselves are based on. How are they programmed to communicate this information, where are they pulling data from, how do they regulate themselves to have the outputs make sense, and other factors. This is where applying cybernetics principles can optimize the technology.

By integrating these principles into the AI models, we can impose more regulation, security, and adaptability. Let’s look at what these principles are and how they might be applied:

Adaptability – Cybernetics emphasizes the importance of adaptability as a key survival trait. If a system can continuously learn from its experiences and use, adjusting its behavior in response to the different prompts it receives based on what data is most applicable to the situation, reliability increases.

Feedback Loops – Feedback is one of the most fundamental concepts of cybernetics. If a generative AI model is created with the right mechanisms for feedback, it will improve over time, learning from its own generative actions. With a continuous feedback model applied, it can learn from its errors so that future responses are more context-aware, accurate and valuable to the user.

Self-Regulation – Cybernetics holds that a system must be able to monitor its own function and adjust as necessary to minimize discrepancies. A generative AI model must also include the ability to regulate against biased content, inappropriate responses, and other considerations so that it adheres to certain standards.

Intelligent AND Ethical

Integrating cybernetic principles into the design-phase of the generative AI models could create more adaptive and ethical systems, reducing concerns about security risks, copyright infringement, biased or harmful content, and the like. It is important to understand that while technologies like generative AI have the potential to benefit us in many ways, there still needs to be a human component, especially when designing the initial models and analyzing and validating the responses for accuracy.

Those designing the models need to take responsibility for knowing what data sources the system is pulling from, ensuring that they don’t violate data privacy laws, give incorrect information, or generate offensive content.? Business practices must be in place to mitigate these concerns, with strict guidelines and processes for system development. Equally as important, there need to be human-imposed regulations in place to monitor inputs and outputs – automation is useful, but a machine isn’t going to know if it is generating something harmful.

Conclusion

The potential of generative AI technology could move us into an exceptionally innovative era of new ideas and insights that help us to accelerate transformation to progress our society and better our world. We do, however, need to apply the principles of cybernetics to design systems that can evolve and adapt in a way that is most responsible. Generative AI should be a supplement to human intelligence, created and implemented with ethics and governance in mind.