I am in the middle of the first week course so bare with me.
I am confused. I was under the impression after the release of Chat GPT that AI could read tons of unstructured data and output specific summaries that are meaningful to humans. The first course explains that AI can really only handle an output that takes about 1 second of thought. Chat GPT seems to defy that.
Can someone help me understand if the 1 second rule still applies after the release of GPT? Thanks!
Hello @agamoran! Welcome to our community.
Which course are you talking about that “explains that AI can really only handle an output that takes about 1 second of thought”?
It all depends on the model. If a model is strong and robust, it will take less than a second to produce the answer, and vice versa. GPT 3 was trained on 175 billion parameters while GPT 4 was on 170 trillion.
Best,
Saif.
Thanks for the help. @saifkhanengr
It was a coursea course called “AI for Everyone”. Is a parameter a rule that governs the AI output or a piece of data it consumes? I am looking to create an application that would look at about 5-10 million records of unstructured data and provide a key summary of the data.
Data has nothing to do with parameters but as you increase the size of your model, the number of parameters increases.
If you wish to get insights from 5-10M records of unstructured data, I recommend you take this Unsupervised Learning course.
If you are new to AI, I will suggest taking this specialization, sequentially.
Best,
Saif.