Day 2 of the cohort had sessions by Digital Futures Lab and Tattle on Responsible AI and AI safety. The sessions covered topics like AI over-reliance, how do we identify them and think of possible ways of avoiding them.
During the session led by Digital Futures Lab, we nudged participants to imagine how users would leverage AI in specific situations. For example, how much would a “Male teacher, mid-40s; 15 years of experience teaching middle school students in a semi-urban government school” rely on AI when severely crunched for time?
One of the discussions explored how much a woman in her late 30s, part of an SHG group, might rely on AI outputs in real-life scenarios. For instance:
- An AI tool gives a diagnostic suggestion that conflicts with her intuition.
- An AI-powered micro-investment app is recommended to her, and its usefulness is validated by her peers.
The responses ranged from under-reliance to over-reliance, sparking an engaging conversation about how personal biases shape our judgments. Many counterarguments highlighted the importance of stepping into the shoes of the woman in these scenarios before answering, which shifted the perspective meaningfully.
We also saw a demo by Tattle, which has built an AI safety tools to flag harmful prompts, prevent unsafe content from being engaged with, and detect abusive messages. One key challenge they highlighted was the difficulty of contextualizing this for multi-language slang, especially when such slang could signal dangerous situations.
We also heard from Inqui-lab foundation and Simple Education Foundation and the progress that they have made in the first 1.5 months of the program.
The second half of the day focused on teams working in groups with their mentors. Since there were 4 NGOs focused on building similar yet slightly disparate solutions for the education space, a round table was encouraged where each of the 4 NGOs came together to brainstorm on a singular solution.
Overall the workshop combined sessions, peer interactions, and focused work with mentors—helping uncover synergies across use cases, sparking collaboration, and fostering peer learning within the ecosystem.
This blog is written by Aditya Bhatkal from Dasra
