Last Friday I attended ‘Insight out’ - an AI workshop organized by @Tech4GoodCommunity and OASIS.
TL;DR: When Rinjhu from T4G asked me how I felt about AI in the morning, I said “I used to be very suspect, I am opening up but still kinda suss”. By 3 pm I couldn’t hold my excitement at what I had built!
That picture should do the talking.
My biggest learnings/ take-aways:
- Know what you want to use AI for and what you don’t - It was useful to hear from K and the others in the room on where they think AI is beneficial and where they would never use AI. (e.g. many said they wouldn’t use AI for writing at all, or they would check every line of AI generated code, if it was a product they were pushing out to the public)
- Be cognisant of the biases it may have - much more potent in social or anthropological work v/s technical work.
- Original data and insights are more useful than ever - LLMs are as good as the data that’s going into it. So if we stop feeding it primary data and ground truthing these models will become useless very soon. This makes the work on orgs getting ground data even more important!
- Using agentic tools (e.g. Antigravity or Claude code) is where you can create a lot of value (v/s using just the chat bot). It’s a big unlock for non-techies, like me. It’s made the gap between ideas and execution way smaller. Of course, let’s not fall into the trap of building inane stuff just because we can. It doesn’t take away that we must first be clear about what’s the problem we are solving and whether it is worth solving, but if used well, this can be a great tool.
- You’ll get up to speed faster than you think. Every group there managed to actually create some working model/ dashboard/ analysis by the end of the session. Super cool!
Why did I get so excited?
I believe there are connections between seemingly disparate events that most of us miss, and surfacing them is important for us to recognize the second and third order effects of what’s happening around us. Even for those working in climate, the ones working in energy may not see the connection with water or the ones working in waste may not see the connection with biodiversity. But I have struggled to make it real for people.
I tried and managed to build a ‘Web of connections’ that showed these hidden connections between different news stories from the broad climate/ environment realm. I just took stories from 4 days and I was quite blown with some of the connections it suggested!
Warning: Not all of this will be right. This still needs to be checked by a human and validated with ground truthing, but it does give a good direction and set of hypotheses to validate!
Post the workshop, I launched a project I have thought about for 3 years! A simple page that published 1 climate news story every day and talks about why it affects our daily life. The whole flow is set up using an AI agent with my oversight and interventions at different points. It’s called 'Times of Climate Change’
A big thank you to @Tech4GoodCommunity , Oasis, Rinjhu, Varshini, Akhil and the whole team for putting this together and to @knadh for leading us through this and being such a cheerleader for all our projects!