Predictive AI: In the Shadow of the ChatGPT Spotlight

Friday 16.02.2024

Regardless of whether the conversations take place in a business or private context: whenever the topic of Artificial Intelligence (AI) comes up, it's actually referring to ChatGPT. Or one of its siblings, which are based on the same principles. However, in the shadow of the ChatGPT spotlight, there is much more to discover!

ChatGPT and tools like DALL-E, Midjourney, or even GitHub Co-Pilot belong to the family of so-called generative AI. As the name suggests, these systems generate new content (such as texts, images, audio/video, program code, etc.) based on patterns in the underlying training data. And these tools are really good at it. In some cases, the results generated by AI are even better than those created by human hands.

However, there are use cases where generative AI systems are not the right approach. In such cases, other AI paradigms are used, such as Predictive AI.

While generative AI creates new content, predictive AI evaluates existing data to make statements about their unknown properties (for example, spam or not spam) or predictions about future events and developments (tomorrow will be sunny or cloudy). Predictive AI systems are significantly more limited in their applicability, but they can generally solve their specific task much better than a Swiss Army knife from the family of generative AI could.

Karakun CEO Elisabeth Maier has examined both paradigms in her software column in Inside IT. Read the full article.

More details about Karakun AG