AI专家正在到来 — — 伪装者也随之而来。GLG专家网络首席执行官Gemma Postlethwaite表示：“任何了解AI的人”都受到试图参与AI交易的投资者的热烈追捧。还有大量深度伪造内容和愿意扮演角色的AI聊天机器人使用者。
AI experts are coming — and the pretenders too. “Anybody who knows anything about AI” is in high demand from investors trying to play the AI trade, said Gemma Postlethwaite, CEO of expert network GLG. And there are plenty of deepfakes and people using AI chatbots willing to play the part.
“我们必须能够提供不仅是他们所说的那个人，而且真正掌握他们所说知识的专家，”Postlethwaite对Semafor表示。为验证真实性，GLG正在进行额外的筛选和身份验证 — — 越来越多地通过摄像头进行，并设计问题来识破AI化身。她还说，他们还在解析转录文本，以发现“某人的专业知识开始偏离正轨”的情况。“可以把它想象成知识的数字指纹。”先验证，再信任。
“We have to be able to deliver experts that not only are who they say they are, but know what they say they know,” Postlethwaite tells Semafor. To test for authenticity, GLG is doing additional screening and ID verification — increasingly on camera with questions designed to trip up AI avatars. It’s also parsing transcripts to find “situations where somebody’s expertise is going off-piste,” she said. “Think of it like a digital fingerprint of knowledge.”Verify, then trust.
GLG还在将前沿AI实验室与学科专家联系起来，以帮助在敏感主题上训练模型，例如能够为处于危机中的儿童制定应对方案的儿童精神病学家，以及能够应对寻求操作手册的不良分子的核武器或生物武器专家。
GLG is also connecting frontier AI labs with subject-matter experts to help train models in sensitive subjects, such as child psychiatrists who can craft responses to kids in crisis and nuclear or bioweapon experts to respond to bad actors looking for instruction manuals.
