Index / Writing / Page 5
Notes on building things.
Business, startups, product, software architecture and applied AI. 105 posts. Subscribe via RSS.
Page 5
49–60 of 105Know Who Owes You: An Invoice and Payment Sheet for Small Businesses
Invoice numbers that never repeat, a status the sheet works out itself, partial payments, an ageing report and polite reminders, all in Google Sheets.

AGI: Where We Are Now (October 2026)
What frontier AI can and can't do in October 2026, what lab leaders and skeptics claim, what surveys say, and a scoreboard of every bet on AGI.

The Other Bets: Neurosymbolic, World Models, Brains and Evolution
Cognitive architectures, neurosymbolic AI, world models, brain emulation, Thousand Brains, active inference, evolution and AIXI: the other roads to AGI.

Bet #5: Teach It to Think (Reasoning and Agents)
The reasoning bet on AGI: chain of thought, o1, DeepSeek-R1, ARC-AGI, METR's task-length trend, AI agents and the 'do they really reason?' debate, as of October 2026.

Bet #4: Just Make It Bigger (Scale)
The scaling bet on AGI: the Bitter Lesson, GPT-1 to GPT-4, scaling laws, emergent abilities, the 2024 data wall and where scale stands in October 2026.

Bet #3: Reward Is Enough (Reinforcement Learning)
From Samuel's checkers to AlphaGo's move 37 and today's reasoning models: the bet that trial, error and a reward signal can build general intelligence.

Bet #2: Let It Learn (Neural Networks)
From a 1958 Navy demo to the 2024 Nobel Prize: the 80-year bet that machines should learn from examples instead of rules, and where it hit walls.

Bet #1: Write Down the Rules (Symbolic AI)
The first bet on AGI: write intelligence down as symbols and rules. From Dartmouth and expert systems to Cyc and Deep Blue, and what survives today.

What Does "AGI" Even Mean?
Where the term AGI came from, the main ways people define it, from the Turing test to DeepMind's levels and ARC, and why the definition is worth billions.

Using AI With Customer Data: The Traffic Light Rule
What's fine to paste into an AI chat, what needs a ten-second edit first, and what never goes in. A simple rule for small businesses, plus a printable list.

World Models: Can AI Simulate Reality?
What a world model is, the evidence that AI builds one inside, the LeCun debate, and what Sora and Genie can and can't simulate yet.

Synthetic Data and Model Collapse
What happens when AI trains on AI-written text? Model collapse explained, why mixing in real data prevents it, and how labs use synthetic data well.
