Hi from

Artificial Linguistic Internet Computer Entity (A.L.I.C.E) is a chatbot that used heuristic pattern matching to have conversations. Invented by Richard Wallace in 1995, it relied on an XML schema known as artificial intelligence markup language (AIML), which specifies conversation rules. The program simulates chatting with a real person over the Internet – in this case, Alice, a young woman who can divulge personal details and answer easy questions.

90’s kids can’t forget SmarterChild. At one point, it amassed 10 million users and was getting a billion messages a day. This temperamental chatbot was baked into AOL Instant Messenger (AIM) and MSN Messenger with the power to carry out fun conversations, with some additional data access to popular services at the time, like the weather and Moviefone. It’s widely considered a successful show of “conversational computing”.

With the explosion of the iPhone came Siri for Apple iOS in 2010–an intelligent assistant and navigator that uses a natural language interface built on machine learning. Google followed suit with Google Assistant, originally launched in 2012 under a different name. Then came Cortana and Alexa in 2014. All of these digital assistants brought interactive voice into the mix, helping users send emails and texts, play music, place calls and find files, locations, and general information.

We’ll get to ChatGPT in a second. Large language models – neural networks with tons (billions) of parameters, trained on large quantities of text – came about in 2018. BERT was first on scene with 340 million parameters. But OpenAI wasn’t far behind, and shocked the world with GPT-3 in 2020, an LLM that boasted 175 billion parameters. After some fine-tuning, a variant called GPT-3.5 was made available to the public in 2022 in the form of a web interface. You guessed it. ChatGPT, the fastest growing app in history, generates human-like text and can carry out a variety of tasks, including writing scripts and translating books.