• Written by Azhar Aliyar, Director

Some people object to using a domain tuned model for AsscherAi, arguing that it is too restrictive and will not be able to answer questions about other areas of the organisation, but this objection is based on a misunderstanding of what a general model and a domain tuned model genuinely commit you to. A general model commits you to a broad set of possible questions, but it does not guarantee that it will be able to answer them accurately, whereas a domain tuned model commits you to a specific set of questions, but it is more likely to be able to answer them accurately.

What each option genuinely commits you to

A general model commits you to a broad set of possible questions, which can be both an advantage and a disadvantage. On the one hand, it allows you to ask questions about any area of the organisation, without having to worry about whether the model has been trained on that specific domain. On the other hand, it means that the model may not have the same level of accuracy as a domain tuned model, which can be a problem if you need to rely on the answers to make important decisions.

In contrast, a domain tuned model commits you to a specific set of questions, which can be both an advantage and a disadvantage. On the one hand, it allows you to get accurate answers to questions about a specific area of the organisation, which can be very useful if you need to make decisions based on that information. On the other hand, it means that the model may not be able to answer questions about other areas of the organisation, which can be a problem if you need to get a broader view of what is going on.

The case where the other one is right

There are cases where a general model is the right choice, such as when you need to answer questions about a wide range of areas of the organisation, and you do not have the time or resources to train a domain tuned model for each one. For example, a category manager may need to compare two suppliers, and a general model can provide the necessary information to make that comparison.

On the other hand, there are also cases where a domain tuned model is the right choice, such as when you need to get accurate answers to questions about a specific area of the organisation, and you have the time and resources to train a model for that domain. For example, a shift supervisor may need to check the maintenance log for a specific machine, and a domain tuned model can provide the necessary information to do that.

The cost that only shows up later

One of the costs of using a general model is that it may require more maintenance and updating over time, as the organisation and its data change. This can be a problem if you do not have the resources to devote to maintaining the model, as it can quickly become outdated and less accurate.

In contrast, a domain tuned model may require less maintenance and updating, as it is specifically designed to answer questions about a particular area of the organisation. However, it can still be a problem if the domain changes over time, and the model is not updated to reflect those changes. You can learn more about how to maintain and update your model.

A way to decide without a six month evaluation

One way to decide between a general model and a domain tuned model is to think about the specific use cases you have for AsscherAi. If you need to answer questions about a wide range of areas of the organisation, a general model may be the right choice. But if you need to get accurate answers to questions about a specific area of the organisation, a domain tuned model may be the better option.

You can also think about the level of accuracy you need, and the resources you have available to devote to maintaining and updating the model. If you need high accuracy and have the resources to devote to maintaining a domain tuned model, that may be the right choice. But if you need to answer questions quickly and do not have the resources to devote to maintaining a model, a general model may be the better option. If you are still unsure, you can contact us for more information.

What this does not do

This article does not provide a comprehensive guide to using AsscherAi, but rather a specific discussion of the trade-offs between a general model and a domain tuned model. It also does not provide a set of rules for deciding which type of model to use, but rather a set of considerations to take into account when making that decision.

Most people get wrong the idea that a general model is always the right choice, or that a domain tuned model is always the right choice. In reality, the right choice depends on the specific use cases you have for AsscherAi, and the resources you have available to devote to maintaining and updating the model. By considering these factors, you can make an informed decision about which type of model to use, and get the most out of AsscherAi.