ChatGPT’s implications for the AI industry
Updated: Mar 26
ChatGPT, developed by OpenAI, is a language model that has gained significant popularity since its release in 2022. The model can converse with users about almost anything and has enormous potential for various sectors. We investigate how ChatGPT may represent a turning point in AI's application in industries like customer service, healthcare, and teaching.
ChatGPT, the state-of-art large language model developed by OpenAI, is revolutionizing various industries and shifting perceptions of how we should interact with technology. In five days, the service managed to blow past the milestone of a million users. In comparison, this feat required ten months for Meta and three years for Netflix to match. ChatGPT was created by OpenAI, a company founded in 2015 by Sam Altman and funded by Y Combinator's startup acceleration program.
The ability of ChatGPT to answer questions posed to it as if one was talking to a human has enormous potential for different sectors. Such language models blur the boundaries of what qualifications employers should demand, considering that these models give almost everyone the ability to write clearly, even for legal documents.
ChatGPT is built on OpenAI's GPT-3 large language model, a 175 billion parameter model trained on over 570GB of internet text. Some have likened it to the human brain, masses of interconnected neurons that collectively learn to identify patterns in data and make predictions accordingly. Released in 2020, OpenAI fine-tuned GPT-3 from 2020-2021 to serve specific tasks, creating GPT-3 Dialog, GPT-3 Finance, and GPT-3 Legal.
Building on the GPT-3 model, ChatGPT was developed through reinforcement learning algorithms which learn the environment of relevant data subjects for a set of tasks to be achieved and reinforce this pattern to maximize a cumulative reward. This approach allows the chatbot to generate human-like text to natural language questions and maintain an engaging conversation. As such, ChatGPT can supplement skill gaps in areas like customer service by answering queries, providing product recommendations, and handling troubleshooting tasks.
ChatGPT also has significant applications for advertising, healthcare, education, finance, news, and government. For instance, medical coding is an integral component in a medical procedure, where spoken or written medical information is turned into codes that can be used for billing and record-keeping. ChatGPT can automate this process, saving time and reducing errors in medical documents.
The challenges facing ChatGPT
Despite ChatGPT's potential, its application in industry is challenging. Since ChatGPT is a black-box model, human verification of its decisions is difficult or impossible. This calls its decisions into question as the model may not possess the specific knowledge and expertise required for it to make accurate predictions in niche industries. Furthermore, ChatGPT is only trained on events up to 2021. As a result, queries about events that have occurred in the past year will be met with limited knowledge, and the model may output false or confused information.
Concerns about the quality of ChatGPT's responses to events after 2021 raise larger questions about what data is used to train the model. Its users are required to agree to unclear terms and conditions that govern how OpenAI will use their data. Similarly, since ChatGPT is closed source model fed with undisclosed data, it is hard to determine whether the model is biased toward certain viewpoints or is trained with a political agenda. Regulations will inevitably be published to address these concerns, but for now, the data privacy and security concerns surrounding ChatGPT must be emphasised.
In conclusion, there is no doubt that ChatGPT is the first step in a rapid acceleration in the broader use of AI. It is likely that we will look back at this moment in the same way we look back at the 2000s dot-com era. With advancements in many industries like healthcare, transportation, and even teaching, ChatGPT is the starting point for open AI to be incorporated in the future. The competition to produce the best large language model is now well underway, with Bing's new chatbot features and Google making their appearance with BARD. Only time will tell which AI bot will dominate households in the future; until then, enjoy the transition into the new digital era.