📨 AI for Social Impact Deep Dive: AGI

✍🏼 A Note From the Editor

Welcome to an existential deep dive on artificial generalized intelligence, or AGI. Here's the thing: nobody can quite agree on what AGI actually is, when it will arrive, or what it might mean for our species. So let's do our best and examine what we know about AGI and what it might mean for our future.

🧐 The Definition Problem: What Even Is AGI?

At the most basic level, AGI is a hypothetical stage in machine learning development where an AI system can match or exceed the cognitive abilities of human beings across any task. Unlike the AI we use today—which excels at specific tasks like image recognition or language translation—AGI can generalize knowledge, transfer skills between domains, and solve novel problems without task-specific reprogramming.

But without a clear definition, it can be difficult to interpret announcements about AGI or claims about its risks and benefits. We also have to consider who is making these claims, and what possible self-interest (or company interest) they are serving.

🔬Different Hypotheses

1. Just Keep Scaling: We just need bigger models trained on more data with more computing power. These folks think the transformer architecture that powers ChatGPT will scale its way to AGI.

2. New Designs: Large language models are impressive, but they're limited. We need entirely new architectural approaches, possibly combining neural networks with symbolic reasoning systems that can understand cause and effect.

3. Evolution and Emergence: Intelligence arose through billions of years of evolution. Maybe we should replicate that process: create systems that improve themselves through competition and novelty-seeking rather than being directly programmed.

4. The Singularity: It’s only a matter of time before your brain merges with AI. Advancements in artificial intelligence will lead to machines surpassing human intelligence, resulting in a profound transformation of human capabilities and the merging of human and machine intelligence.

5. Maybe Never?: AGI might be ill-defined, far away, or impossible with current methods. We are so focused on distant AGI scenarios that we are ignoring the immediate harms from current AI: bias, misinformation, labor exploitation, environmental costs.

🦾 So, Now What?

Expert timelines vary on if/when we will reach AGI, reflecting different assumptions about scaling laws, hardware limits, algorithmic breakthroughs, and whether current deep-learning approaches can generalize far enough. And don’t forget that economic and competitive incentives play a significant role in AGI too, as AI companies benefit from portraying themselves as close to major breakthroughs. Critics caution that this creates hype cycles that may distort a real assessment of genuine progress and limitations.

🎸 Times Are A-Changing

The bottom line: there is no way to accurately predict if/when we will reach AGI. But we do know that things are already beginning to change, so it’s time to consider: how might AI disrupt your work?

This isn’t about stoking fear, it’s an invitation to practically examine where we are and where we are headed, and lean into a new future of work. As they say, “the only constant in life is change.”

👋🏼 About AI for Social Impact

I’m Joanna, and I’m on a mission to help folks in the social impact sector understand, experiment with, and responsibly adopt AI. We don’t have time to waste, but we also can’t get left behind.

Let’s move the sector forward together. 💫

♥️ Spread the Love

Spread the love and forward this newsletter to anyone who might benefit from a dose of AI inspo!