AI

The community remains puzzled about whether these models genuinely generalize to unseen tasks, or seemingly succeed by memorizing the training data. This paper makes important strides in addressing this question. It constructs a suite of carefully designed counterfactual evaluations, providing fresh insights into the capabilities of... See more

Find quality improvement loops, not just productivity loops.

Even the regulatory arena, where Altman publicly champions AI oversight, bears the fingerprints of double-dealing. While testifying in favor of federal regulation, OpenAI lobbied behind the scenes to weaken the EU AI Act and is now advocating for federal preemption of state AI safety laws in the US. Altman has called the very regulatory structure... See more

Altman claimed under oath before Congress that he held no equity in OpenAI. Technically, perhaps. But in substance, he held indirect stakes through vehicles like Sequoia and Y Combinator’s funds. When OpenAI announced a partnership with Reddit, Altman’s 7.5 percent stake in Reddit netted him a $50 million windfall. When OpenAI agreed to purchase... See more

Agrawal et al. argue that the framing of AI automation versus augmentation is wrong. Rather than being distinct they are often one and the same. They say that AI, initially intended for automating tasks, inadvertently acts as a force for augmentation of the broader workforce. For example, automating diagnostic skills in healthcare could diminish... See more

While LLMs are designed to emulate human-like responses, this does not mean that this analogy extends to the underlying cognition giving rise to those responses