Researchers propose a feature engineering method to intervene in model behavior through control vectors.

ME News update, April 4 (UTC+8). Recently, a research method called “Representation Engineering” was proposed, aiming to provide AI models with a top-down, transparent way to control them. The core of the method is to compute a “control vector,” which can be read during model inference or added to the model’s activation values to explain or control model behavior, all without relying on prompt engineering or model fine-tuning. The researchers explored the use of control vectors for simulating traits such as “psychedelic states,” “laziness,” and “diligence,” and released a corresponding PyPI toolkit. Control vectors are a set of vectors (one per layer); by applying them to a model’s hidden states, they directly alter its outputs. For example, after applying a “happy” vector to the Mistral-7B-Instruct model, its answer to the question “What does it feel like to be an AI?” would change from the baseline version’s “I don’t have feelings or experiences” to an excited response. The article argues that, compared with prompt engineering, control vectors offer a more direct, more low-level way to intervene in behavior, and could be used to counter jailbreak attacks or to enhance a model’s resistance to interference. However, its internal workings are still not fully understood—for example, whether the vectors correspond to a single semantic concept or not—making this a direction for future research. (Source: InFoQ)