Federal AI Forum<\/a> that agencies do not need to wait for perfect AI systems. Instead, they need to build the right foundations around the technology so they can use it with confidence and scale what works.\u00a0<\/p>\nBrian Frutchey, CTO of Clear Fracture, said that starts with making sure that the decision-maker, whether human or AI, has the context it needs before taking confident action.\u00a0<\/p>\n
\u201cAI agents quietly fail not because they lack reasoning power, they fail because they lack grounding,\u201d he said. \u201cAn agent operating on vague instructions and unorganized situational awareness is just improvising. An improvisation leads to hallucinations, compliance failures and security risks. If your start and endpoints are fuzzy, agentic speed just gets you lost faster.\u201d\u00a0\u00a0\u00a0<\/p>\n
NOAA is already applying that approach in the field. Stephen Bieda, acting deputy director of the NOAA Center for AI, described AI systems that help sift severe-weather reports, identify wildfire hotspots and accelerate weather prediction.\u00a0<\/p>\n
The agency\u2019s AI-based wildfire detection has given wildland firefighters an average of 17 minutes of lead time before a 911 call, according to Bieda. The system is designed to alert forecasters, who then evaluate the information rather than allowing AI to make the final decision.\u00a0\u00a0\u00a0\u00a0<\/p>\n
NOAA is also using AI to reduce the computational resources required for global weather prediction, while continuing to incorporate large amounts of observational data. The agency is working to make its hundreds of petabytes of data more accessible for AI applications while developing policies and risk controls to support responsible adoption.\u00a0\u00a0<\/p>\n
The same focus on building responsible practices is emerging around AI spending. Sidney Epner, a product manager at Apptio, said organizations are learning that managing AI requires more than tracking tokens\u00a0and added he saw a 4.7-fold increase in model spending over six months across his customer base. \u00a0<\/p>\n
Epner said it\u2019s important for organizations to have the right insight into their AI spending to connect their investments with measurable outcomes. \u00a0<\/p>\n
\u201cConnecting that spend to the outcomes, whether it\u2019s the cost per case processed, the incident resolved or mission outcome-not just what you spend, but whether it was worth it. Once you can prove it, the conversation changes,\u201d he said. \u201cIt stops being about minimizing AI spend, and it starts being about optimizing it for cost, sure, but also for performance, risk and mission value, and for federal teams, that also includes data sovereignty, security review and procurement cycle.\u201d\u00a0<\/p>\n
For the 4th Marine Logistics Group, the path to broader adoption is being built around people. Chief of Staff Col. Scott Wilson described how reservists with civilian expertise in AI and innovation helped him develop an AI-enabled command climate survey prototype in three hours. \u00a0<\/p>\n
As a result, surveys that previously took months due to low participation rates and other obstacles, now had a quicker turnaround, higher participation and a\u00a0lower barrier to entry, Wilson said. Wilson and his innovation team are currently working to roll out the tool, PULSE Check, to other units to assist commanders in making real time command climate analysis.\u00a0\u00a0<\/p>\n
\u201cWe\u2019re hoping that in the next year, next time I come talk at this forum, we\u2019re going to be able to show how we\u2019ve rolled out the forward deployed engineer concept, how we have enabled AI innovation and modernization at the component MLGs, and we\u2019ve been able to accelerate their decision-making and their resource optimization using AI tools that are available to us now. It\u2019s incredible what\u2019s out there. We\u2019re making a lot of progress,\u201d Wilson said. \u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"Federal AI leaders say grounding, human oversight, cost visibility and workforce expertise are essential to scaling AI effectively.…\n","protected":false},"author":2,"featured_media":140793,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[179,24,25,203,204,69274],"class_list":["post-140792","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","tag-agentic-ai","tag-ai","tag-artificial-intelligence","tag-data","tag-defense","tag-noaa"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/140792","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=140792"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/140792\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/140793"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=140792"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=140792"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=140792"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}