Use Cynefin to understand when to use (and not use) Generative AI

Since the launch of ChatGPT on 30 Nov 2022, there has been many hot discussions around what kind of role generative AI tools like ChatGPT play. Some said it’s going to take over people’s jobs, some said it will act primarily as an assistant to augment people’s capability, some said it would be useful because of its rather low accuracy rate and possible issue with copyrights, some even said it will make us mentally and intellectually lazy too.

None of these are wrong with the specific context. Given to the ultimate goal of Generative AI is to make it more like people especially approaching complex problems. I found it useful to apply Cynefin framework here to help understand better when to use and not use generative AI.

What’s generative AI?

Generative AI uses algorithms to create new, realistic contents from training data based on patterns learned from vast amounts of unlabeled datathrough self-supervised learning. It involves training the model on large amounts of unlabeled data, allowing it to discover patterns and relationships within the data and generate new content that resembles it. This is often used for tasks like generating images, text, or music. Yet most involves a hybrid approach together with supervised and semi-supervised with specific scenarios. Worth to note that, supervised learning limits the creativity and novelty of the generated content

What’s changed by generative AI?

Generative AI fundamentally changed:

• people’s access to knowledge, without having to learn everything by ourselves. On one hand, it lowers the bar for access to vast amounts of existing knowledge for good and for ill deeds; on the other hand, it may change how people share knowledge in the future. There’s a possibility organizations can take advantage of this technology to build their moat wall.
• Refocus on the type of work that humans should perform. This is another motivation to rethink about what type of work we should focus on and what type of work we could leverage tools like generative AI.

The Cynefin framework

The Cynefin framework, named after the Welsh word for “habitat,” acts as a compass for navigating different problem spaces. It categorizes situations into five domains based on the relationship between cause and effect: clear (simple), complicated, complex, chaotic, and disorder.

Each domain demands distinct approaches:

• in clear situations, follow established rules;
• in complicated situations, analyze relationships to find the right answer;
• in complex situations, experiment and adapt to emergent patterns;
• in chaotic situations, act decisively and learn as you go;
• and in disorder, pause, assess, and identify the dominant domain before acting. By understanding the context through Cynefin, you can choose the most effective approach to any challenge, maximizing the chances of success.

When to use and not use Generative AI

Here is a brief summary around the impact of Generative AI for different domains: