ChatGPT, Dall-E, Gemini, LLaMA, Stable Diffusion... the list goes on. Artificial intelligence has been working its way into all kinds of digital products and services over the last couple of years. Your smartphone, television and even your fridge probably now have some form of AI incorporated into their systems.

AI as it stands today generally fits into these broad categories:

1) Generative AI

Using AI models, like the text-based ChatGPT or Stable Diffusion for images, to create or modify content and images for use in your own systems and products. The quickest way in to this technology is via one of the paid services offered by the large providers like OpenAI. Working with this type of AI is largely a case of connecting to APIs and designing effective text based prompts to get the models to do what you need.

The image below is a fantastic example of a generative AI image from Dall-E, OpenAI's image model. The prompt to generate this image was 'A van Gogh style painting of an American football player'.

Image from https://labs.openai.com/ 

It's pretty impressive isn't it?

2) Machine Learning

Using AI on a more technical level to take advantage of model training or embeddings to build your own AI systems that are more tailored to your business.

An example of this would be using a transformer model to generate text embeddings for the customer help guides for your product. These embeddings could then be stored in a vector database and used to query your help guides based on a question from a user. The question is converted to embeddings and these are queried against your data, this can result in incredibly accurate matches for user queries, especially when the user doesn't know exactly what they want. This works by placing the user's question into the vector space where your help guides are embedded, it will be closest to the most relevant pages in your help guide and these are what it returns. The quality of the results is a huge leap from keyword based querying.

This is just one use case of machine learning and has huge potential within many areas, the possibilities really are limitless.

How can I use AI in my business?

If you're wondering how AI technologies can be used in your own business and projects the best place to start is by looking at the repetitive tasks that you or your employees are performing, could these be something that an AI could take over to free up your time? Does your product or service have areas where it could integrate AI? For example if you're running a T-shirt printing business could you bring generative AI in to your design process to give customers a better experience and more design choices?

It all sounds too good to be true, what's the catch?

The catch is that whilst AI is incredibly impressive in scope, it can't be relied upon to be factually correct, and it can't be relied upon to provide consistently accurate output. So using it to read your client's contracts and come up with legal advice is probably not a great idea... but it could be very useful at scanning through a contract with a well defined objective such as picking out potential errors.

Cowshed Works and AI

We're running a few AI and Machine Learning based systems in real-world apps so we've got some experience of this. Any new idea with AI is probably going to be a learning experience for everyone but we've got a solid grounding in the technology to help introduce it to your business. 

Get in touch if you'd like to discuss your AI ideas.

Photo by julien Tromeur on Unsplash