November 22, 2026

John Hill, CEO and Founder of Silico on Generative AI and its use cases

placeholder

Written by John Hill, CEO of Silico

Generative AI refers to the use of machine learning where an AI model, given some questions or information, generates an ‘answer’ or additional content. These models are trained on a large quantity of data and learn relationships and patterns within training material. This allows them to retrieve data they have been trained on. As models have become larger, generative AI is showing it can accurately synthesise different concepts together.

These generative AI models are also showing strength in not just taking input from human language and images, but also code and computer data formats such as JSON and XML. This form of AI can be used by anyone but is most effective with existing subject matter experts that can steer the AI models in the right direction with their questions and assess whether the information generated is accurate.

Generative AI’s use cases

Generative AI models have potential use cases anywhere knowledge is already known, but this information is large and unwieldy. Generative AI can be trained on information and then effectively answer questions or apply that knowledge to another context. The less information included in the original training material, the more review and guidance the generative AI will require.

The challenges associated with Generative AI

For businesses, the accuracy of results is significantly more important than in tasks like generating an image based on a prompt. For example, when a generative AI is tasked with turning a text prompt into a database query to return specific results, it must produce the correct database query and display results appropriately. If it queries the wrong fields or filters and aggregates incorrectly, the ‘wrong’ information may be provided to a decision-maker with costly consequences. Unfortunately, generative AI responses can confidently communicate incorrect results and lead their users on the wrong path.

Therefore, business decision-makers should be very thorough when it comes to assessing AI-generated insights. Generative AI can be valuable for quick experimentation, but any results should be sense-checked carefully.

How Generative AI will evolve

As training methodologies for generative AI improve, so will their accuracy and reliability. Each iteration has also improved its ability to be steered, which is directed towards a specific goal, such as producing a specific output or operating within certain constraints.

Recent advances such as GPT-4 have also shown them becoming multi-modal, taking in both text and images at the same time, allowing them to better understand traditional human documents.

The landscape of possible applications is still being explored. However, as models continue to improve, it seems likely that generative AI will find uses in assisting almost all knowledge-worker professions.