Topics
The recurring debates, disagreements, and positions across the archive.
Agents
For two years "agents" meant nothing. Then, in a single month, Claude started moving your files, and the frontier labs split over what an autonomous economy actually costs.
AGI
Every frontier leader now agrees AGI is coming this decade. They are quietly terrified of completely different things... and a few of them stopped using the word at all.
AI Safety
The people building the most powerful technology in history mostly agree it could go catastrophically wrong. What they cannot agree on is who gets to draw the line, or where.
Alignment
Every frontier lab now agrees the machine doesn't reliably do what we mean. The fight is over whether you can grow a conscience or have to engineer one.
Competition
Every lab swears the race is dangerous. Every lab keeps running it faster, and in early 2026 even Anthropic stopped pretending its conscience could outvote the competition.
Compute
Every frontier lab agrees intelligence is something you buy by the gigawatt. The fight is over who controls the meter... and whether the bill ever comes due.
Economics
The people building AI agree it will rewrite the economy. They cannot agree whether that means abundance, a jobs bloodbath, or a hollow boom no one can measure.
Geopolitics
The labs spent a decade insisting they were building tools. In 2026 the US government decided they were building weapons, and the frontier finally had to pick a side.
Governance
When the Pentagon gave Anthropic three days to drop its red lines, the question stopped being whether AI should be governed and became who gets to do the governing.
Interpretability
Everyone building the most powerful machines in history admits they can't fully read their own minds. The fight is over whether that's a research problem or a deadline.
Jobs
The people building the thing that automates your job mostly agree it will. What they fight over is the speed, the order, and whether you'll still go somewhere in the morning.
Labor
The men building the machines that automate work can't agree on whether that's a bloodbath or an abundance... and in 2026 even the chief doomsayer started hedging.
Near-Term Risk
The frontier's safety pitch has quietly migrated from killer robots to economics: the man building the country of geniuses now warns the real danger is speed, not the machines themselves.
Open Source
The lab founded to "open source a lot of technology" now keeps its weights locked... and almost everyone agrees, right up to the capability where they don't.
Predictions
They all agree the wave is coming within years, not decades. The fight is over how fast it crashes, what it drowns first, and whether anyone's watching the horizon.
Regulation
They once dreamed of an IAEA for superintelligence. By 2026 the same founders were drawing red lines against their own government, and losing.
Scaling Laws
The bet that made these companies says pour in data and compute, get intelligence out. In 2025-26 the founders started quietly arguing about whether it still holds.
Superintelligence
The people building it now agree it is coming, and soon. The fight is no longer whether intelligence will surpass us, but who survives the transition.