Just the mention of those two letters – A.I. – and you can see the full spectrum of emotions and thoughts it conjures in business owners.
On the one hand, we are being compelled to explore and innovate using the tool to drive the business forward, yet on the other, no one seems to have a clue about what or where to implement it effectively.
Similarly, a future ruled by machines that may replace us, our jobs, or our industry strikes fear into the hearts of some. While others dream of a lean workforce, competitively producing high-quality work powered by AI while banking hideous amounts of profit.
AI, LLM, GPT… WTF?
AI is a broad and wonderful subject that covers many functions, from machine learning to robotics, deep learning, computer vision, and natural language processing.
Specifically, most business owners, tech influencers, or “gurus” are probably concerned about a subset of AI that uses Large Language Models (LLMs) for tailored text-based tasks. More specifically still, this subset is the world of Generative Pre-Trained Transformers, or GPTs.
Here’s ChatGPT’s take on explaining this relationship like we’re five years old:
- AI is the smart robot toy.
- LLMs are the big books of words that the robot uses to understand language.
- GPT is a special book that helps the robot talk well.
We are being pushed, but where should businesses start?
Until recently, OpenAI, Google, Microsoft, and other serious AI players were progressing by integrating different technologies to allow humans to access more advanced AI functions. By now, most of us are familiar with what ChatGPT can do for us by giving prompts (read: instructions or questions) to the “chatbot.”
However, since ChatGPT-4o came to the surface things have started to look even more interesting. This release introduced a multimodal model, that allows the tool to understand different types of information (audio, visual, etc). This offers better performance at a much faster speed and lower costs.
But more importantly, multimodality offers the option to go beyond text and analyse images, interpret data, process video, and have flowing voice conversations with AI—among other tasks.
The work experience kid
A fun analogy is likening LLMs to work experience kids. Yes, they are eager to please and will rush to find you the answer or do the work, but it takes a lot of clear instruction, guidance, training, and attention for their output to get anywhere close to a commercial standard.
Large Language Models (LLMs) excel at processing data and are ideal for automating tasks that involve language. Working on logic, LLMs can rapidly analyse data and reports, read emails for you, and summarise blog articles and publications.
Drafting creative content is where it currently comes unstuck, which is more of a user problem than a tool problem. When ChatGPT3.5 was released, a popular use case was drafting blogs, website content, scripts, and other creative forms. However, it’s important to remember that these tools work on logic and what existing information they have. Creativity requires lateral thinking, something the tools have not yet mastered.
In addition, it is crucial to remember that LLMs are trained on existing data, which can be biased in many harmful ways. Critical evaluation of its outputs is essential to ensure accuracy and avoid perpetuating or reintroducing biases into the model.
How LLMs can empower small businesses
At Forte, we have been exploring ways to integrate these technologies into our workflow. Training the model is one of the key steps we have identified to improve output quality and avoid generic or biased results. This is what OpenAI offers by allowing you to create your own “GPT”.
Here are some specific examples of how LLMs can automate repetitive tasks in a small business:
Reporting and analytics
LLMs can analyse data from various sources and generate reports with summaries and insights. While having access to the raw data is still critical, AI could help you discover issues or trends that a human could easily miss.
Brainstorming and planning content
While automating the content creation process is still far off, we have found useful workflows and prompts to brainstorm content ideas, generate outlines, or even draft short social media posts. Remember, their role is to assist, not replace, your creativity, expertise, and human touch.
Integration and automation
Integrating LLMs via APIs with other systems and platforms unlocks a world of potential.
APIs act as bridges, allowing AI models to share their capabilities with existing workflows. We have created an email-to-task list workflow by connecting AI with our email inbox. Whenever it receives an email from a specific domain or sender, the model analyses the email and determines if any tasks need to be done. Any tasks are then fed into a project or task management system and tags specific staff members for help.
Large Language Models (LLMs) and Generative AI are powerful tools that can potentially transform small businesses’ operations. While it’s important to be aware of their limitations and the importance of responsible use, LLMs offer a wealth of practical applications.
If you are unsure where to start or how these tools can benefit you, please contact us.
