London Futurists Anticipating 2026
Jan 7, 2026
The hosts reflect on significant developments from 2025, including the rapid improvements in large language models and the 'jagged edge' of AI. They categorize attitudes towards AI into three camps: accelerationists, doomers, and scouts. The conversation touches on rising cyber risks linked to AI, an investment bubble, and OpenAI's vulnerabilities. Additionally, they explore AI's potential in handling everyday tasks, the political implications of faster information, and the urgent need for equitable job redistribution amid increasing automation. Optimism for 2026 is highlighted.
AI Snips
Chapters
Books
Transcript
Episode notes
LLMs Show A Jagged Progress Curve
- Large language models (LLMs) are rapidly improving and now exceed humans in many narrow tasks.
- They still lack a form of common-sense capability, producing a 'jagged edge' of strengths and weaknesses.
From Forecasts To Levers
- Move from forecasting to identifying levers that can change outcomes rather than just predicting them.
- Focus effort on parts of power that are amenable to change to actually improve the world.
Personal Zoom SDK Malware Attempt
- David Wood described a Telegram-to-Zoom scam where attackers tried to get him to install a fake Zoom SDK.
- Several acquaintances installed malware and lost crypto, showing targeted social-engineering risk.
