There has been much attention and hype since the launch of the research phase of ChatGPT in November 2022 and how it can answer what appears to be any question that is posed to ,it in an amazing amount of depth and intelligence.
So what is ChatGPT?
ChatGPT is a chatbot interface powered by large language models (LLM) and natural language processing (NLP) developed by OpenAI.
By asking questions to it, it is able to reply by combining 175 billion parameters of data, and predicting what would be the most suitable response to that question. It can also read and understand a number of programming codes used in analytics. In theory a lot of what we currently might ask google it can look to do and more.
How might you use it in Analytics?
At the moment it looks like in the world of analytics it is most useful in automating repetitive tasks and helping with best practice approaches.
If you are adept at breaking down a project into specific tasks, it can help with progressing through those. Some, but not extensive uses are:
- Automating data gathering, cleansing processes and create dummy data
- Help define data structures in terms of fields in a database or rows and columns in a spreadsheet
- Write code for modelling E.g python code for simple linear regression
- Suggest how to construct charts, suggest report structures, create training material and research purposes for best practice. E.g how to use the vlookup in xls
It does have its flaws…
- Sometimes it can get it wrong, so the user couldn’t solely rely on it to answer big questions without having some knowledge on what the expected outcome could be
- For analytics you can only upload text files at the moment rather than any format of dataset
- When we tried to go into it, it was showing at capacity so wasn’t available to even use
Will analytics roles change as a result?
Where analytical work is repetitive in nature, the chances are in the near future technology like ChatGPT could automate those, removing the need for that element of the job.
However, it cannot replace human decision making, problem solving, leadership, strategy, team building and know the complexities that face individual companies to be able to answer that with a NLP or LLM response.
It would be wise to know and understand what it can do to help compliment existing roles, to know what could be automated to make way for other areas of the role it cannot replace. Worth a try to see how it could help with your day to day…..if its not at capacity when you go to try!