Craig Martell, Chief Digital and Artificial Intelligence Officer of the Pentagon, has expressed his concern on the potential of deploying AI tools such as ChatGPT. He notes that even though such tools are highly capable of responding to simple prompts with factual, articulate answers, they may not always provide accurate content. He reminds us that these tools lack a human understanding of context, as was seen when a reporter posed as a 13-year-old to get advice from an LLM.
Martell raises an important issue about how such tools could be utilized for malicious purposes, such as adversaries running influence campaigns or disseminating disinformation. He urges the industry not to merely sell AI services, but to also focus on development of detection functions to identify content generated by AI and distinguish it from human-authored content.
Therefore, counter technologies are needed to identify AI-generated content and, to ensure trustworthiness, the providers of the tech service must “build into it the right safeguards and the ability for us to validate” the information. Likewise, Craig’s priorities as head of CDOAI is to build infrastructure to meet the aspirations of joint-all domain command and control, JADC2, as well as to more effectively understand the data needed, data already existing and the data required to be shared.
Just the day before, Lt. Gen. Robert Skinner, the head of DISA, spoke highly of such AI tools and expressed his enthusiasm for the adoption of the new technology. Besides him, DISA’s Chief Technology Officer, Steve Wallace also stated the potential of leveraging such tools for back office capabilities and data labeling.
Nevertheless, Craig Martell is surprisingly not opposed to the AI deployment in mission-critical applications, albeit cautiously. His focus is to build tools which allow the systems to be supervised and monitored, in order to continue to bring the expected value.
Craig Martell has a wealth of knowledge in the field of artificial intelligence and machine learning, bolstered by his extensive experience in academia and managing machine learning at Lyft. His mission as Chief Digital and Artificial Intelligence Officer is to ensure the Department of Defense’s AI efforts are met and to make sure computer infrastructure and data operate without fail. He recently fed his thoughts on AI development and deployment during AFECA’s TechNetCyber event in Baltimore.
