A #101Guide to why is everyone so excited about Generative AI and #ChatGPT?
Like most, I have been obsessing over #ChatGPT.
Lets learn more about why the hype in 5 based on a lovely article that I will link in comments
#1 Is this already in use? Generative AI has been around for a while now – Grammarly, for example, already puts generative AI into use
#2 Why the hype now and how is it different from other ML based recommendations and models used earlier?
This begs the question about why we use machines and in what aspects are they better than us? There are three areas one could look at around judging that –
1. Perception
2. Optimization
3. Cognition
Example of perception is a facial recognition algorithm that can ‘extract patterns very quickly’
Optimization refers to the ability of an algorithm to assess multiple options (quickly) to arrive at decisions – ex. crunching data to identify the most optimal route
An area that AI has struggle (and still does to a large extent) is the third one – Cognition – ‘ to be able to make sense of things and understand them in context’.
Something that stands in stark contrast, when my then 5 year old understood context better than my Google Home device.
And that is what we have started to impact with Generative AI – to not just ‘understand the language, but the story’.
And that is the difference between searching keywords in an engine vs asking a question on ChatGPT
#3 Can you talk layman language and tell me what it implies?
I’d highly recommend the book – ‘The Moon is a Harsh Mistress’ by Robert Heinlein because it probes this topic further through Sci-Fi. As ‘Mike’ the AI that has ‘come alive’ and its implications to society
#4 How does Generative AI work?
“Generative AI is an umbrella term for a number of ML methods — including Large Language Models (LLMs) and Generative Adversarial Networks (GANs)”
Most ML algorithms used mainstream today are ‘discriminative’ –
1) Recommendation Systems : Things similar to what you have purchased
2) Classification : This is an image of a cat
3) Clustering : Here are 5 tweets about Obama
And this is based on input data that it is trained on.
Generative AI, can create data leveraging what it pays ‘attention’ to amongst other things, thereby creating closer to ‘human-quality’ output.
#5 What should we be wary about?
For all the hype, the tech is still early stage. There are umpteen challenges associated with it as well around security (deep fakes), data privacy to copyright issues.
More interestingly, the ethical implications of the tech.
The future though, is what we make of it 🙂
What are your thoughts on Generative AI?
https://sifted.eu/articles/what-is-generative-ai/
********************************************
#reviewswithranjani #GenerativeAI
#Technology | #Books | #BeingBetter