Jun 24, 2024

AI for Small Businesses: Introduction

We are kicking off a multi-part series with the goal of taking a pragmatic look at artificial intelligence and how small businesses can use current AIs in everyday business without investment.

Right now, there is no getting around the topic of "Artificial Intelligence." AI dominates the news in the IT industry, and digital tools and services are increasingly marketed as AI-supported. When AI appears in the headlines, it is usually "Generative AI" (GenAI) based on "Large Language Models" (LLM). Such news often involves generalist chatbots like ChatGPT, Microsoft Bing/Copilot, Google Bard, or Claude.

Generated with DALL-E via ChatGPT

What do Chatbot, GenAI, and LLM mean?

GenAI refers to any AI capable of creating content such as texts, images, or musichttps://link.springer.com/article/10.1007/s12599-023-00834-7. Modern GenAI learns this ability by being "trained" with a huge amount of data. If this data is text, it is referred to as a "Large Language Model" (LLM)https://dl.acm.org/doi/pdf/10.1145/3624724. Chatbots are then the face of these LLMs: they are the applications that allow you to interact with an LLM.

A prominent example explains it well: ChatGPT is a chatbot application that uses the LLM "Generative Pre-trained Transformer" (GPT). ChatGPT is also a GenAI capable of creating entirely new texts.

The Current Hype

The latest hype cycle for AI began with the release of ChatGPT in November 2022https://arstechnica.com/information-technology/2023/11/chatgpt-was-the-spark-that-lit-the-fire-under-generative-ai-one-year-ago-today/. While the underlying technology (the GPT models) had been available for some time, ChatGPT marked a major leap in natural language processinghttps://www.sciencedirect.com/science/article/pii/S266734522300024X. Almost overnight, complex AI was accessible to a broad audience and amazed us with complex conversations.

The release of ChatGPT triggered a race for GenAI: Many players (some of whom had already been developing their own LLMs in parallel) rushed to bring their own LLMs and chatbots to market. Microsoft released Copilot (based on ChatGPT), Google announced Bard and Gemini, Anthropic made Claude publicly available, and Meta (Facebook) offered limited access to the LLM "LLaMA". The list could go on.

The Strengths of GenAI

All these new LLM-based products share a number of positive features. They are good at:

  • Understanding and processing natural language.
  • Aggregating and summarizing information.
  • Generating creative texts.
  • Translating or paraphrasing language.
  • Simulating various roles and their linguistic patterns.

The Weaknesses of GenAI

LLMs also have several weaknesses. They are poor at solving mathematical tasks or performing numerical analysis. They tend to "hallucinate" when asked for factual information. This is particularly dangerous as they (thanks to their actual strength, creative writing) can present objectively false answers as facts very convincingly. Moreover, LLMs are not designed to always have the latest data for their answers: they "know" nothing of new developments until they are trained with relevant information.

What You Should Also Consider

In addition, there are critical aspects beyond inherent weaknesses of LLMs that should be considered.

Data security and privacy: It is not transparent how the data you provide to chatbots is processed, stored, or used later. Your inputs (including confidential texts) may be used to train the AI and could thus find their way to the public.

Legal issues: Currently, it is a contentious issue which data developers of LLMs are allowed to use to train the AIs. At the same time, the copyright of content created with GenAI falls in a legal gray area.

Ethical aspects: In terms of social and societal issues, LLMs may reinforce biases. LLMs are also increasingly used for criminal activities.

Environmental impact: Training and operating LLMs is very energy-intensive and creates a large environmental footprint.

Practical Benefits for Small Businesses

From the strengths of GenAI, exciting use cases emerge for small businesses. Specifically, we see great potential in the following use cases:

  • Creative writing and generating images: You don't have to write or translate every report by hand.
  • Ideation: Need a sparring partner for a new business idea? Chatbots can take on many roles to help you refine your idea.
  • Research: New chatbots are sufficiently good at searching for and aggregating information from the internet.
  • Summarizing documents: You don't have to read the latest 197-page guideline in full. Let an LLM handle that.

These use cases have the potential to save a lot of time on everyday, mundane tasks, thus creating more room for the development of your core business. This is a goal we also pursue with our consulting services. For this reason, we will discuss each use case in detail in the following posts and provide you with practical examples of how you can benefit from the hype without investment. We will start next week with the most obvious case: creating texts and images.