The approach improves performance but complicates inspection by researchers.
The Information reports that OpenAI’s forthcoming Astra model employs a technique that makes its reasoning harder for researchers to inspect. OpenAI says the model marks a step up in capabilities such as coding and operating applications on a computer. The publication states that an innovative technique improved the model’s performance and could cut costs. At the same time the method complicates efforts to detect dangerous behavior according to the article.
17.9K
1 post, first seen 22h ago
The approach improves performance but complicates inspection by researchers.
The Information reports that OpenAI’s forthcoming Astra model employs a technique that makes its reasoning harder for researchers to inspect. OpenAI says the model marks a step up in capabilities such as coding and operating applications on a computer. The publication states that an innovative technique improved the model’s performance and could cut costs. At the same time the method complicates efforts to detect dangerous behavior according to the article.
The Information@theinformationOpenAI’s upcoming Astra model uses a technique that can make its reasoning harder for researchers to inspect. The approach could improve coding performance and cut costs while complicating efforts to detect dangerous behavior. Read more: https://thein.fo/463LwGV
The Information@theinformationOpenAI’s upcoming Astra model uses a technique that can make its reasoning harder for researchers to inspect. The approach could improve coding performance and cut costs while complicating efforts to detect dangerous behavior. Read more: https://thein.fo/463LwGV
