ai-forecasting
The Quiet Power to Decide What Gets Built
Somewhere right now there is an engineering team that has spent a decade building on Erlang. It is a language famous for one thing: keeping programs alive. Phone networks run on it. It was designed so that one part of a system can fall over and the rest keeps humming. That kind of resilience is…
The Day the Machine Stopped Asking and Started Buying
Mastercard spent this week wiring rails for a customer it has never had: a piece of software that can reach for your card on its own. Decrypt’s account of Mastercard enabling AI agent payments with Coinbase and Ripple shows the plumbing running straight through crypto infrastructure, which tells you the shape of the thing. The…
The Week Scale Stopped Being the Answer
One of the people who helped invent the modern AI model just changed teams. CNBC reported that Noam Shazeer, a co-lead on Google’s Gemini, left for OpenAI, taking with him a name that appears on the original paper that made all of this possible. On its own that’s a transfer-window headline, a star moving from…
The Race Everyone Is Watching Is Not the Race Being Run
Amazon’s AI chief said this week that the company has fallen behind OpenAI and Anthropic, but sees a path to catch up in the coming year. CNBC’s account of how Amazon lagged the frontier labs and expects to close the gap is a calm piece of reporting about a calm and honest admission, and it…
When the Expensive Thing Gets Cheap, Watch Where the Value Goes
A coding model from a lab called Z.ai now matches the best paid system on long, complicated programming tasks and runs at roughly a sixth of the cost. VentureBeat’s report that the open-weights model beats a frontier system on long-horizon coding benchmarks for a fraction of the cost is remarkable less for the benchmark than…
Eight Hours, A Hundred Pages, and the Quiet Problem of Too Much
A research agent from Sakana AI now writes a hundred-page report in eight hours. Not an outline. Not a draft for a human to finish. VentureBeat’s write-up of an ultra deep research agent that produces hundred-page reports in eight hours describes a finished document: sourced, structured, the kind of thing a consulting team used to…