Happening now: Join staff writer Ian Bogost for a live Q&A about his latest stories; his course, Ordinary Extraordinary; and more. Click on Discussions to enter! For AI doomers, signs and portents are everywhere. When OpenAI introduced ChatGPT, almost four years ago, it couldn't do basic arithmetic. Now the company has unreleased models that are more fluent in higher mathematics--the language of the universe--than all but the most highly trained mathematicians.

Experts say claims AI could kill all humans through biological weapons development had no scientific basis. Experts say claims AI could kill all humans through biological weapons development had no scientific basis. There have been some shocking claims in recent days about AI safety: we face a 10% chance of doom; AIs are worse than nukes; a "botnet" threatens the entire internet; it's all a big tech psyop. Below, we look at six claims and reactions to them. 'In 6-12 months a swarm could be capable of taking over the entire internet with a persistent botnet - potentially causing hundreds of billions of dollars in damage' - Dario Amodei, chief executive of Anthropic, 12 September 2026 Amodei's concerns should not be entirely dismissed but require a large shaker of salt, said Gary Marcus, an AI sceptic and emeritus professor at New York University.

'I tried to hold the company to its ethical commitments against supplying AI for military use. When Google broke those commitments, I resigned at significant financial cost so that I could publicly document Google's broken promises.' 'I tried to hold the company to its ethical commitments against supplying AI for military use. When Google broke those commitments, I resigned at significant financial cost so that I could publicly document Google's broken promises.' I worked at Google DeepMind.

This position paper argues for two claims regarding AI testing and evaluation. First, to remain informative about deployment behaviour, evaluations need account for the possibility that AI systems understand their circumstances and reason strategically. Second, game-theoretic analysis can inform evaluation design by formalising and scrutinising the reasoning in evaluation-based safety cases. Drawing on examples from existing AI systems, a review of relevant research, and formal strategic analysis of a stylised evaluation scenario, we present evidence for these claims and motivate several research directions.

Reinforcement learning (RL) for large language models (LLMs) is dominated by the cost of rollout generation, which has motivated the use of low-precision rollouts (e.g., FP8) paired with a BF16 trainer to improve throughput and reduce memory pressure. This introduces a rollout-training mismatch that biases the policy gradient and can cause training to collapse outright on reasoning benchmarks. We show that the mismatch is non-stationary and acts as a double-edged sword: early in training it provides a stochastic exploration bonus, exposing the gradient to trajectories the trainer would otherwise under-sample, but the same perturbation transitions into a destabilizing source of bias as the policy concentrates. To solve this, we propose Adaptive Importance Sampling (AIS), a correction framework that adjusts the strength of its intervention on a per-batch basis. AIS combines three real-time diagnostics, namely weight reliability, divergence severity, and variance amplification, into a single mixing coefficient that interpolates between the uncorrected and fully importance-weighted gradients, suppressing the destabilizing component of the mismatch while preserving its exploratory benefit. We integrate AIS into GRPO and evaluate it on the diffusion-based LLaDA-8B-Instruct and the autoregressive Qwen3-8B and Qwen3.5-9B across mathematical reasoning and planning benchmarks. AIS matches the BF16 baseline on most tasks while retaining the 1.5 to 2.76x rollout speedup of FP8.

Adaptive importance sampling(AIS) uses past samples to update thesampling policyqt.

Even superforecasters are guessing that they'll soon be obsolete. To live in time is to wonder what will happen next. In every human society, there are people who obsess over the world's patterns to predict the future. In antiquity, they told kings which stars would appear at nightfall. Today they build the quantitative models that nudge governments into opening spigots of capital.

Onthetheory side,wediscuss howtotailorthevelocity fieldtothetargetandestablish general conditions under which the proposed estimator is a perfect estimator with zerovariance.

Forthis10] propose extensiontothe15], which roundingthevistokrandomly in Algorithm (10).

In this case we observe that using UHA leads to higherELBOs.