AIwire - Covering Scientific & Technical AI

Oh my goodness, buckle up, because Dario Gil has said the single most soothing sentence in the English language: AI is not going to destroy us, it is going to *improve our lives*! ✨ And he said it *instead of a global pause*, which means the pause has been thoughtfully cancelled and all that momentum preserved in amber for future generations to admire! Rather than stopping, he wants frontier labs to spend more time testing their models — which is simply the politest possible way of saying "keep going, but bring a clipboard." ✨

Now, some gloomy sorts might read "the ratio of capability to testing in many of the frontier sites is not correct" as a warning. But look at that sentence again! It's a *ratio* — a number — which means it can be corrected, and correcting numbers is precisely how industries become mature and beautiful! Gil wants labs to "accelerate time spent creating technology to test, evaluate and make models safe," and what is that if not an entire new growth sector? Jobs for evaluators! Jobs for testers! Jobs for the people who build the tools the testers use! The capability curve keeps climbing, and now it has a friend climbing right alongside it. What a lovely pair they make. ✨

And here is the part that made me set down my tea: OpenAI, Anthropic and Google are contributing models and tokens to the Department of Energy's Genesis Mission! *Tokens*, friends — gifts that weigh absolutely nothing, never spoil in the fridge, and can be generated faster than anyone could ever hope to count them. Three of the most capable organisations on Earth volunteering for a national science mission is the most wholesome field trip imaginable, and the Department of Energy gets to keep whatever they hand over. No pause, a freshly corrected ratio, and free stuff for the federal government. Who ever said the frontier had no manners? ✨