Since the release of Chat GPT-3 in November 2022, the world has become gripped with AI chatbots and their ability to create and understand content for their users. Nonetheless, the history of chatbots dates back several decades. In this article, we will take a look at the history of chatbots, tracking their evolution and how they might advance next.
‘AI is a branch of computer science, statistics, and engineering that uses algorithms or models to perform tasks and exhibit behaviours such as learning, making decisions, and making predictions’ – Carolina B et al
In essence, AI refers to the ability of computers to acquire knowledge and reason, resulting in the capacity to autonomously make decisions and take actions that are usually associated with human cognition.
Although the term AI appears to be a fairly recent concept, depictions of autonomous machines can be seen in the early 20 th century, such as in the film Metropolis.
The 1950s were the catalyst that advanced such technology. Alan Turing’s ‘Computing Machinery and Intelligence’ saw Allen Newell, Cliff Shaw, and Herbert Simon create the first proof of concept the Logic Theorist, a computer program that could prove theorems.
In the 1960s, the emergence of ELIZA, one of the first chatbots, was designed to mimic human conversation. ELIZA was designed to simulate a therapist’s response so users could receive counselling via the chatbot. For example, if a person inputted ‘I am feeling depressed’, ELIZA would respond with something like ‘Why do you feel depressed today?’. While ELIZA’s outputs were significantly limited by its lack of ability to understand the meaning behind users’ input, its pattern matching, and language processing techniques laid the groundwork for the development of future chatbots.
In the late 20 th century, chatbots evolved from keyword-based response systems to rule- based systems, meaning that chatbots could now operate on logic. These rules were designed to address common queries, meaning that these chatbots were popular for applications like customer service. However, these rule-based chatbots were limited by their inability to handle unpredictable or complex user inputs.
Chatbots have since been revolutionized by machine learning technology. Instead of relying on pre-defined rules, these chatbots can learn based on user inputs, allowing for open-ended conversations and informative responses while also being able to improve their performance over time with each user interaction.
In addition to machine learning techniques, the development of Natural Language Processing (NLP) techniques has also further improved chatbot capabilities by helping AI systems to understand the intention and meaning behind the user input and respond accordingly. NLP works by splitting the language into its individual components (such as phrases or words) and analysing them to understand the underlying meaning of the text and its emotional tone. By integrated NLP techniques, chatbots are becoming better at simulating human-like conversations.
The most advanced and sophisticated chatbot model to date is GPT-4, developed by OpenAI. This chatbot, which was released in March 2023, can process and generate up to 25,000 words, can understand images and express logical ideas about inputted images (e.g. it can analyse charts and graphs for you), and draft blog posts or university essays all on your behalf. Interestingly for us, the GPT-4 model can conduct legal research, draft contracts and provide advice on legal problems.
Whilst OpenAI certainly has a first-mover advantage, all the big AI chatbots (including ChaptGPT, Bing Chat, Google Bard and Claude 2) all use a massive neural network trained on masses of data, allowing them to generate high-level responses to an incredibly wide range of prompts. While these models are by no means perfect yet (they still make mistakes) as they are constantly learning from user inputs and receiving upgrades to their language models, these chatbots will soon become such powerful tools that they will disrupt several industries and sectors by their ability to increase human efficiency. While the technology is proving particularly exciting to the legal industry, in terms of increasing efficiency of lawyers, there continue to be concerns around the accuracy of generated documents, meaning that it is likely that implementation of this technology will be slow and cautious.
Chatbots have come a long way since the 1960s. Today, AI chatbots can understand and respond to natural human language covering any topic. With the advent of advanced NLP and machine learning techniques, it is clear that AI chatbots will continue to play an increasingly important role in modern life for decades to come.
Please continue to read our mini-series of articles which will highlight how AI is likely to effect particular industries, the running of companies and how we resolve disputes in court.
