近日，OpenAI针对公众关注的“AI威胁论”展开了一项实验评估，旨在验证人工智能技术在生物武器制造领域的应用风险。实验结果表明，GPT-4虽能轻微提升制造生物威胁的信息获取能力，但整体风险较小。
在这场实验中，OpenAI采用了被誉为“最强语言模型”的GPT-4，对其在生物武器制造领域的应用潜力进行了深入探讨。实验结果显示，尽管GPT-4能在一定程度上提高制造生物威胁的信息获取能力，但这种提升相对较小，且可控。这意味着，在合理监管下，GPT-4等技术并不会对生物武器制造带来显著风险。
事实上，针对AI技术可能被恶意利用的担忧并非空穴来风。然而，此次实验结果为公众注入了一剂“强心针”，表明在充分认识到AI技术潜在风险的同时，通过有效的监管和管理，我们有能力确保AI技术的安全与应用。
英文翻译：
News Title: GPT-4 Enables Controllable Risk in Bioweapon Manufacturing
Keywords: GPT-4, bioweapon, risk assessment
News Content:
Recently, OpenAI conducted an experimental assessment to verify the application risks of artificial intelligence technology in the field of bioweapon manufacturing. The experiment showed that although GPT-4 can slightly enhance the information acquisition capabilities of manufacturing biothreats, the overall risk is small.
In this experiment, OpenAI used GPT-4, which is known as the “strongest language model,” to delve into the application potential of artificial intelligence technology in bioweapon manufacturing. The experiment showed that although GPT-4 can improve the information acquisition capabilities of manufacturing biothreats to some extent, the enhancement is relatively small and controllable. This means that with reasonable supervision, technologies like GPT-4 will not bring significant risks to bioweapon manufacturing.
In fact, the concerns about the potential malicious use of AI technology are not groundless. However, the experiment results inject a “booster shot” into the public, indicating that we have the ability to ensure the safety and application of AI technology by fully recognizing its potential risks and implementing effective supervision and management.
【来源】https://www.cls.cn/detail/1587372
Views: 1