AI and it’s quest to find General Purpose understanding of ‘ideas’ seem to me interesting and tricky at the same time. Ideas are rooted in their context of place, people, culture. This is how we have been learning about the world, people and their stories and cultures.
Now we have a machine which is attempting to find General Purpose understanding of things. Two critical questions pops up, one is is there anything called “General Purpose” understanding of anything? This questions cannot be generally discuss other wise it will be an irony, we could take up scenarios and discuss. For example this is the slide fore day before, General understanding of Musical Genres in AI models:
The questions can now be asked contextually. That once machine has learned this, how will continue to diversify? Will machine not dictate its’s understanding on anything new that is coming in? For example, an experimental sound may be put in to a genre, and there may be many genre which are yet to be defined, yes, we have many questions but let’s wait machine is learning. We may have to explore the tools in many cases o learn about it, and that’s something to keep exploring, taking inspiration from Machine, how it is continuously learning.
GPT-3 – Aakash Wani!
Aakash Wani in Hindi language means is the loud and mysterious voice from the sky, that’s how exploration in GPT3 felt like! You feed GPT3 text with some pattern and gives you back some stories after reading between the lines of your input text. This mysterious but insightful tool makes our team really curious. You can see in following image our first input and output and our reactions :
And we enjoyed this and every output met our curiosity :
Poem Fusion as input in GPT- 3 :
Jo, Dibyendu and I were working as a group. Jo had got a pem written by a friend named Ishani, Dibyendu had a poem written by Tagore and I had a poem written by a friend named Piyush.
We took one one paragraph from each poem. Defined our models in a structure of premise and poem. In premise we mentioned whos poem it is and who has got it to GPT3 and in the poem we input few lines. There we go. Checkout input and output which we received from GPT 3.