Has anyone used GPT-3 to load data from a database table and summarize the data into words. For example load a database table about people which contains data about address and it would summarize:
X% of people live in NY and has increased by X.
1 Like
This is a duplicate topic.
@logankilpatrick , please delete or close this topic. Thanks!
Please continue here.
https://community.openai.com/t/openai-gpt-summarizing-database-tables/80826
1 Like
Thanks @ruby_coder . Here is what I’m looking to do. Lets say I have a database table that’s looks like this:
Now assume this is a larger table with multiple records.
I want to be able to teach OpenAi this data and summarize it such that. I can say, “Summarize this data,” and it will say something to the effect of:
X% of people have a license in NY and has increased by X this month (year, etc)
Basically have OpenAI learn my data so I can ask it any type of question?
1 Like
The short answer is no.
Even if you use the fine-tuning API method, you cannot “teach” a generative AI to run calculations for you like you want, I am sorry to inform you @chirag.shah285
You will need a different approach.
Sorry to inform.
Interesting I was in the Playground messing around with the data I provided and its doing basically what I want:
Its unclear to my how can I scale this to load all my data
That is because you just “got lucky” because if I test it using the API, text-davinci-003 “makes up” a different answer.
As I tried to explain to you, these generative AI models do not actually do calculations. They generate text. They auto-complete. They make things up to generate a completion.
It’s easy to be fooled into thinking that a generative AI is performing calculations whenit is just “generating blah, blah” generating a sequence of tokens based on the current tokens.
This is one of the “dangers” of generative AI.
Hope this helps.
2 Likes
@chirag.shah285 . There’s no direct way to connect GPT to a database on OpenAI’s UI as far as I’m aware of. There are however, ways you can interact with your data tables using AI Agents. Aigents run on specific tasks on data and connect to Large Language Models on the backend for reasoning, search and other language tasks.
Leaptable has built an AI Agent orchestration framework which you simply connect to your database, spawn a number of agents, connect a language model - could be OpenAI, Co:here or Anthropic and run tasks such as summarizing your data. Leaptable also offers a managed, hosted version which you can simply upload a CSV to and then you can being interacting with your data using a Prompt dialog.
1 Like