Artificial Intelligence is a machine learning that mimics human intelligence and behaviour. The machine reasons like human being to perform its operations. It uses big data for modelling and predictions.
Generative AI is a subset of Artificial Intelligence, that generates set of data from big data base. What it basically does is that it extracts information from big data which can be texts, images, pictures, videos, audios, etc. to perform a specific task.
Basically, there are three types of Generative AI data classifications:
- 1. Binary classification : This classifies data into two (yes or no; relevant and not relevant; spams and not spams; frauds and no frauds, etc). This can be done with Pruned Decision Tree Modeling.
- 2. Multi-class classification : This classifies data into positive, negative and neutral. This can be done with sentiment classification analysis.
- 3. Multi-label classification: This is used for complex tasks. It is a machine learning technique that predicts multiple labels for a given input. This means that it produces multiple outputs. For instance, a movie can be classified as documentary, science fiction, action and comedy.
The type of Generative AI classification to use depends on the intent of the researcher. A researcher dealing with a case of transactional fraud, will definitely use binary classification to predict whether there is a fraud or not. It is the prerogative of researchers to have deep understanding of how each classification works in order to professionally utilize them to solve specific tasks.
Specifically, this article focuses on ChatGPT model of Generative AI. ChatGPT is a generative AI model that provides answers to questions on request. It responds to human interactions like any other online search engines. ChatGPT is a text-based model, generating human-like text responses in conversational settings, generate contents and can be used to schedule meetings, among others.
ChatGPT-3 was developed by OpenAI in year 2020 and upgraded to version GPT-3.5 in year 2022. The latest version GPT-4 was released in 2023 with innovative and advanced features. Nonetheless, version GPT-5 is work in progress and is expected to be launched soon.
ChatGPT extracts information from big data, simplify the information in a layman language and enable users to define the limit of the information. For instance, if you ask ChatGPT a question, it will bring out lengthy responses or voluminous texts. You can further ask the model to limit the texts to certain number of words. This will be accurately done without missing any value from the original responses. If the feedback is not too clear or cumbersome to comprehend, you can further ask the model to present it in a layman language. This will be done by simplifying the sentences to a level of easy comprehension. One good thing about ChatGPT model is that it is user-friendly and allows users to continuously dig deep into the enquiry by asking the model for additional information that are directly linked to the original question.
ChatGPT is a useful tool for conducting desk research of any kind. I have personally used it to research on topical issues that required secondary data. It will amaze you the quantum of useful information I generated that were relevant to the topics. ChatGPT requires less skill to navigate through the processes of optimizing the information generated. What is required from users is to have a clear mission or intention on what to search for. A visionless approach will lead to vague responses, which could be termed as fatigue. This should be avoided in the use of ChatGPT.
Another caution to take note of is the issue of plagiarism, particularly for academic purposes. It is unethical to pull out the work of another person from the internet and use it directly as if you are the original writer without any form of reference. You must be creative in adapting the information extracted from ChatGPT model for academic write-ups.
Limitations of ChatGPT
ChatGPT model only uses information in the public domain. Some of the information in the public domain may be obsolete because they are not usually validated from time to time. It beholds on researchers to carefully validate the information to avoid misleading decisions. Sometimes, a deep stick primary data should be gathered to add value to the secondary data. For instance, if you request ChatGPT to list out the brands of beverages in Nigeria market, it will provide you with a list of all the available brands of beverages in Nigeria that are domiciled in the public domain whether the brands still exist or not. It may not take into account new brands that are not yet in the public domain. This will affect the quality of the data.
ChatGPT does not obey mathematical instructions, hence it cannot be used for computation. The best ChatGPT provides is statistical formulas and instructions on how to apply them. The actual computation cannot be done with ChatGPT. You need machine learning to do this. The texts generated from ChatGPT can be used as inputs for machine learning that will train the data and convert them to numbers. Once the data is converted to numbers, they can now be used for mathematical modelling and predictions.
Conclusively, ChatGPT is a powerful Generative AI Model with highly innovative features that enable users to navigate through big data, that are practically impossible with human efforts. There are a lot to gain from ChatGPT for day-to-day business activities. Grab the opportunity and utilize it to gain competitive advantage