GPT-3- Fooled ya? Detective hat 'on'

I'm not much of a critic myself since I'm much too fascinated with how GPT-3 was introduced to the world. The bot entered the social media scene as a human, under the username u/thegentlemetre. It was unmasked when it caused a posting massacre on Reddit's subreddit r/AmITheAsshole. Posts typically describe a personal situation and ask the community to decide whether the original poster is the one at fault—issuing a judgment of “YTA” (you’re the asshole) or “NTA” (not the asshole).

This book covers how to deploy large language models in production. The book also provides detailed case studies of how companies built NLP products using a large language model. Overall, this book is a great resource for anyone who wants to learn more about building NLP products using large language models.

Getting GPT-3 to assist on a suicide helpline was the funniest bit in the whole book if you're into dark humour. I kept hoping to find more funny fails like this one, but that hardly makes for academic material. Adding memes and fail stories was a great touch. I would read a whole book of those alone.

There were a lot of terminologies related to the app development world. One particular term stayed with me, called Geofencing. I thought, what in the world is that. It's something to do with virtual boundaries and location based services.

The most engaging bit of the book was about toxicity scores and how tackling the challenge of Human Bias determines the quality of the generated text. Negative, positive and neutral can be pretty subjective to each individual but to come up with anti-bias measures they would have to define the sentiments objectively.

Time to have some fun with GPT-3. Stay awesome.