How I integrated OpenAI ChatGPT API to Google Sheets & am not impressed just yet!

ChatGPT has been in rounds of conversations every where. This AI chatbot launched in Nov 22 has broken all records (of other platforms like Instagram and TikTok), with 100 million active monthly users. Such a tool in hands of a non-developer (like me :D) can be disruptive. But then, it is only as good as the response it gives. And, I would say at this moment, it is not that impressive.

ChatGPT API integration to Google Sheets

We have to give it to Google, for how they have made integrating to the ChatGPT APIs so seamless. Anyone can do it in 3 simple steps.

  1. Go to Google Workspace Marketplace and install the plug in GPT for sheets and docs.
  2. Go to the Open AI website, log in and generate an API key by clicking on “Create new secret key”

3. Head over to Google Sheets and Under “Extension”-> Click on Enable GPT functions. Also, Click on “Set API Key” and put in the API key obtained in Step 2. That’s it, you are all set!

What can you do with it!

Type ‘=GPT’ in the Sheet Cells and check out the list of commands you can use from. You can start by typing in =GPT(“Say hi”) to get a response back. Below is the list anyway (Source : Google Workspace Marketplace):

GPT to get the result of ChatGPT in a single cell
GPT_LIST to get multiple result in a column (one item per cell)
GPT_TABLE to get a table of items from a prompt
GPT_FILL to fill a range from examples
GPT_FORMAT to sanitize your spreadsheet data into the same format
GPT_EXTRACT to extract entities from your Sheets™ data
GPT_EDIT to edit your spreadsheet content
GPT_SUMMARIZE to summarize your spreadsheet content
GPT_CLASSIFY to classify spreadsheet content into a single category
GPT_TAG to apply tags to your spreadsheet content
GPT_TRANSLATE to translate your spreadsheet content
GPT_CONVERT to convert from table to csv to html to json to xml, etc
GPT_MAP to map values from two columns

We shot a few questions at the bot and checked the responses. Well, it is fascinating how the API plug in connects to the bot and gets us our responses without needing us to exit the App and going to the ChatGPT log in, however the response provided lacked the timeliness and accuracy. Let’s drill down to what we liked and what makes us skeptical about what this technology can do in hands of people who don’t understand that the responses can be baseless and outdated(as of this moment). Of course, in the hindsight we can’t ignore the fact that the bot is in testing phase, with access to limited data set.

What’s “Not so Impressive”!

1.You can NOT trust the response at all!

We asked the bot to give us top 10 logistics companies and their CEOs and the information provided was not the latest. You an get the same result through the command ‘=GPT_Table(“Give me top 10 Logistics companies and their CEOs”)’.

The information for latest FedEx CEO is incorrect

We tried to ask ChatGPT about their full form and history, we got an absolutely ridiculous response.

Here’s the response from the bot when we chose creative level 1 – “There is no full form of ChatGPT, as it is not an acronym or initialism for anything. Rather, ChatGPT is a platform for online chat rooms, where people from around the world can come together and communicate with each other in real-time.”

Here’s the response from the bot when we chose creative level 0- “ChatGPT stands for Chat Global Personal Technology. It is an online platform that provides a chat service for people to connect and communicate with each other. The platform was founded in 2018 by a group of tech enthusiasts who wanted to create a space where people could come together and share their ideas, thoughts, and experiences.”

2.It is too generic!

It is good for someone looking for a high level outline for something, draft of sorts. But you can not use it as it is. It is in no way – an accurate, in depth description of the topic you ask the bot to describe.

3. It is slow!

We used ChatGPT from their official open AI website and the response was much faster than the Google sheets plugin. The delay may be introduced due to the plugin making an API call and then parsing the response and providing it in the cell. The lag is for a few seconds when you can see the ‘Loading…’ comment on the sheets cell. Not complaining, as this is an early stage, but still – we are so used to having everything at our disposal in micro-seconds with a high speed internet on our smart phones.

What’s Good!

Seamlessness & ease of use :

The fact that a curious non-developer like me can simply generate an API key and start typing in excel-like formulas to get response from the bot is fascinating. All you need to do is punch in a formula and get a tonne of (some of it useless at this stage) information back.

Things are just starting. Such a technology in hands of the entire world – free of cost – can be potentially fruitful but also dangerously negatively disruptive at the same time. It’s just a matter of time when the bot will give accurate responses and the tasks it will be able to automate will be phenomenal – like report generations, dashboards from complex data models or even suggest which stock to buy next(which can only lead to more manipulations if its not controlled and accurate).

And now, the rivals are developing their own bot. Google is building Bard. The bots are learning with every conversation, at an unstoppable speed. Scares us more than it excites us!

What are your thoughts on how our world will look like with AI bots?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s