With the recent release of OpenAI’s ChatGPT, the topic of artificial intelligence and its ever-growing applications in the public’s daily lives has become a trending topic. Within just 5 days of its public release, ChatGPT has amassed over 1 million users and has reignited the public interest in artificial intelligence. ChatGPT’s ability to generate human-like text and to perform a wide range of language tasks such as question-answering, conversation, and summarization has sparked discussion on the creative uses of AI, as well as the consequences and morality of doing so.
What is ChatGPT?
But first of all, what is ChatGPT? Normally, we could simply search Google for this answer, but why not try asking ChatGPT itself? Here is the answer it provided:
ChatGPT is an AI-powered language model developed by OpenAI. It uses cutting-edge machine learning algorithms to generate human-like text based on the input it receives. Essentially, it can answer questions, write stories, provide explanations, and more, all based on patterns it learned from large amounts of text data. It's like having a digital friend or personal assistant that can understand natural language and respond in a way that sounds like a real person. The goal of ChatGPT is to make it easier for people to interact with technology in a more natural and intuitive way.
And if you wanted to explain ChatGPT to your 5th grade child, you could simply ask it to do exactly that!
ChatGPT is like a super smart computer friend that can answer your questions and talk to you. It was made by some really smart people called scientists who taught it how to understand and talk like a human. You can ask it anything, like what's the capital of a country, or tell it to write a story for you, and it will try its best to give you an answer or write a story that makes sense. It's like having a helper who can understand what you're saying and respond in a way that sounds like a real person!
This simple question-answering is merely scratching the surface of ChatGPT’s potential. With its ability to generate human-like text, ChatGPT has become a valuable tool for many tasks including content creation, as it generates articles, summaries, and text with ease. Within the 3 months since ChatGPT was released, it has become one of the most widely used AI language models with a plethora of general use cases including language translation, homework completion and essay writing, chatbots, and many more use cases to be explored.
Here are a few other use cases of ChatGPT that could make life and work easier.
Alternative to using Google for research
Think about all of the times you have used Google search, whether it be for work or for personal matters. ChatGPT makes general information retrieval a much better experience than sifting through tens of hundreds of Google search results for the information we want. People can now ask ChatGPT and get the information they need in a single, easily digestible result. The threat of ChatGPT has even made it onto Google's radar as they have recently issued a “code red” to reprioritize their efforts into AI development and products. Google has even announced the release of their own AI language model, Bard, to rival Microsoft-backed OpenAI’s ChatGPT, despite the criticism from employees indicating that it was not ready as a product yet.
As amazing as ChatGPT may be, it still has a long way to go before it can destabilize the monopoly that Google has on the market share of search engines. ChatGPT comes with the caveat that it is only as good and reliable as the data that it has been trained on. As a result, ChatGPT can often-times generate misinformation or biased results and people should remain wary when using ChatGPT for retrieving factual information.
Providing structure to unstructured text or data
Another extremely useful functionality of ChatGPT is its ability to quickly parse unstructured texts or documents, identify key entities, categorize them according to predefined categories, and even determine the overall sentiment of the subject.
Here is an example of ChatGPT identifying the entities within a short story and classifying them.
This process called Named Entity Recognition (NER) is incredibly useful with many notable real-world use cases including:
- Healthcare: Extracting key patient information from medical records
- Customer support: Categorizing customer support requests, complaints, and questions and filtering them by priority keywords
- Human Resources: Summarizing and filtering applicant resumes
However, the code it generates is not completely free from bugs and other errors depending on the complexity of the requirements and prompts that were provided to it. Because of this, it will always be necessary to have an engineer review the code that it generates. While ChatGPT won’t be replacing engineers anytime soon, it can definitely be a productivity boost as more of a coding “assistant”.
ChatGPT is also capable of generating SQL, which is typically used for managing and manipulating data within relational databases. Relational databases are widely used in various industries including finance, healthcare, e-commerce, and more.
Typically, companies would go to their IT department or analysts in order to extract the information necessary to have their questions answered and to understand how the company is performing. With AI becoming more and more sophisticated, this process which would normally take days to weeks, would become near instant as anyone could simply ask their questions and ChatGPT could output the required SQL to retrieve that information.
Although ChatGPT is able to generate SQL queries, it has not been trained specifically to do so, thus the accuracy and desired outcome of the SQL may not be as expected or may be completely wrong. The correct SQL query often requires an understanding of the relevant details and context of the company’s database such as the database schema and tables. In the image above, ChatGPT is able to generate a SQL query to calculate the monthly revenue from sales in 2022 to 2023, but it has no knowledge of the actual tables and columns within our database. It creates the SQL query based on assumptions of data that may or may not exist thus the query would most likely fail.
For this reason, ChatGPT will not work properly out of the box for production-ready use cases.
How Athenic AI can fill this gap
Athenic AI is an AI-powered question-to-SQL generator that is universally compatible with any database, data warehouse, CRM, ERP, and other applications including Shopify, Stripe, Splunk, and more. Athenic AI allows anyone from sales, marketing, management, human resources, customer support, and other non-technical teams to be able to retrieve the information that they need on-demand and in near instant time. Users just simply connect to their data sources and can start using Athenic AI out of the box. Under the hood, Athenic AI utilizes semantic search and GPT-3, an even larger, more general purpose AI language model created by OpenAI, in order to understand your question and generate the correct SQL query within the context of your company data.