Can a small number of samples poison a human of any size (intellect?). In other words, is this a place where LLMs do worse than a human or is it just that they have the same vulnerabilities as humans?