Reinforcement learning with the help of neural-guided search consumes huge computational resources to achieve remarkable performance.

We show here that this approach can effectively be used to make a large pretrained network equivariant.

Chinese technology giant ByteDance has pledged to curb a controversial artificial intelligence (AI) video-making tool, following threats of legal action from Disney and complaints from other entertainment giants. In the last few days, videos made using the latest version of the app Seedance have proliferated online. Many have been lauded for their realism. Disney's lawyers accused ByteDance of committing a virtual smash-and-grab of their intellectual property, including superheroes from Marvel, Star Wars and various cartoons. On Monday ByteDance told the BBC that the company respects intellectual property rights and we have heard the concerns regarding Seedance 2.0.

We introduce and study the problem of detecting whether an agent is updating their prior beliefs given new evidence in an optimal way that is Bayesian, or whether they are biased towards their own prior. In our model, biased agents form posterior beliefs that are a convex combination of their prior and the Bayesian posterior, where the more biased an agent is, the closer their posterior is to the prior. Since we often cannot observe the agent's beliefs directly, we take an approach inspired by information design . Specifically, we measure an agent's bias by designing a signaling scheme and observing the actions the agent takes in response to different signals, assuming that the agent maximizes their own expected utility. Our goal is to detect bias with a minimum number of signals. Our main results include a characterization of scenarios where a single signal suffices and a computationally efficient algorithm to compute optimal signaling schemes.