How to avoid AI disasters : the existential guide to exponential futures

AI is a continuous and narrowing balancing act. Companies unleashing powerful new AI models have to balance safety and speed, but the markets only reward them for the latter. Alignment remains science fiction.

Like all technology, AI can be used for good and bad. But when bad could mean auto-hacking a power grid, AI-scamming bank networks or designing new pandemic viruses using only a prompt, freedom enables disasters. Here we have to balance the narrowing path between potentially disastrous freedom with the dystopia of a surveillance state. 

Sure, mankind has handled transformational tools before. But not a tool that can create infinite new tools. AI is ultimately the last tool we have to invent, but it will also be mankind's ultimate test.

What happens when AI exponentially empowers everyone but also keeps polarising and radicalising us for maximum social media engagement? AI feels democratizing but is ultimately a winner-take-all technology. What happens when AI makes the wealth gap peak? 

What happens when nobody can distinguish real content from deepfake? How will the legal system work when evidence, alibis and implications can be generated? Can Universal Basic Income feed an unemployable world?

This book lists more problems than solutions, reflecting another unsettling imbalance. Some problems are fixable but some are so systemically deep that the best we can do is understand and talk about them before it s too late.

About the author: Eric Langenskiöld is an awarded AI keynote speaker, startup investor and serial entrepreneur.