Social Innovation Lab Launches Innovation Forum AI Series: This Month on AI

“Is there anyone who doesn’t use Artificial Intelligence (AI) on a day to day basis?” The room remained silent when Kuldeep Bandhu Aryal, Senior Manager of BRAC Social Innovation Lab, raised this question to the attendees of the Innovation Forum held this month.

With AI evolving at an unprecedented pace, we recognised the need to connect with fellow colleagues on our respective understanding of the current scenario in the AI space. This year’s first Innovation Forum—“This Month on AI”—was designed to track and recap significant AI advancements in February and their implications for BRAC employees and development practitioners at large.

The forum started with attendees receiving handouts summarising major AI breakthroughs from February. Kuldeep then opened the discussion by revisiting the DeepSeek AI model and its R1 adoption, showcasing how DeepSeek efficiently distills knowledge into smaller, more accessible models. Acting as a “teacher” model, DeepSeek-R1 curates training data for student models like Qwen and Llama, reducing the need for high-powered GPUs and making AI solutions more cost-effective.

Another highlight was the launch of Google DeepMind’s PaliGemma 2 Mix, a vision-language model fine-tuned for multimodal tasks. One of its standout features is its availability in multiple parameter sizes, making it highly adaptable for image and text analysis while allowing developers to use their preferred framework.

Continuing the discussion, Kuldeep compared Grok-3, Elon Musk’s latest AI model from xAI, with OpenAI’s GPT-4o and Google’s Gemini. Grok-3 claims to outperform its competitors, particularly in its “Big Brain” mode, which is optimised for scientific research, multi-layered AI tasks, and complex problem-solving scenarios that require advanced inference capabilities.

A particularly relevant AI advancement for BRAC was the launch of Nvidia’s CorrDiff model—a generative downscaling model designed for high-resolution regional weather forecasting. Kuldeep highlighted its potential applications within BRAC’s Climate Change Programme (CCP), where access to precise climate and weather data is essential for programme planning and impact assessment.

To demonstrate the power of AI-driven research, Kuldeep then engaged the audience in a live experiment using OpenAI’s new Deep Research feature in ChatGPT. Within 20 minutes, ChatGPT generated an extensive research report on a selected topic, illustrating how AI can significantly reduce the time spent on research writing, allowing experts to focus on exploring new areas of inquiry.

The session concluded with a discussion on Anthropic’s Claude 3.7 Sonnet, the company’s first hybrid reasoning model, which has shown notable improvements in coding and front-end web development.

Following the AI recap, attendees participated in an interactive Q&A session, sharing thoughts and raising questions about how these innovations could shape their work.

The second half of the forum featured Emily Gosselin, Senior Manager of Partnerships at BRAC USA, who explored AI’s growing influence in fundraising. She highlighted how corporate and philanthropic investments in “AI for Good” are on the rise, with companies like Google.org, Mastercard Inc, Bezos Earth Fund, GitLab Foundation, and other private donors actively funding AI-driven social impact projects.

Emily underscored that AI is becoming the “new electricity”, presenting a major opportunity for BRAC to stay ahead of the curve. She also demonstrated how AI can streamline proposal writing, improve data management, and enhance fundraising efficiency. Concluding her session, Emily shared BRAC’s latest AI-related achievements and highlighted that BRAC Social Innovation Lab (SIL) and BRAC Health Programme (BHP) were accepted into Tech to the Rescue’s AI for Changemakers (AI4C) bootcamp – Health Impact Cohort and with support from BRAC USA and secured a private donor funding to develop BRAC’s AI strategy.

With AI advancing at an unprecedented rate, its use cases are expanding across all sectors. The forum reinforced a key message: BRAC and its employees must actively adapt to AI-driven technologies to maximise impact and drive innovation in the development sector.

As AI becomes increasingly integrated into fundraising, climate resilience, research, and programme management, it is clear that leveraging these advancements will be essential for shaping the future of development work.

Prima Sarker is a BRIDGE intern at the Social Innovation Lab who likes to smile her problems away and sees cooking as free therapy sessions.