HONOLULU (CN) - Artificial intelligence is transforming the way militaries plan for and fight wars, defense officials and technology executives said this week at the Honolulu Defense Forum.

The annual event, held in Waikiki and hosted by the nonpartisan foreign policy research institute Pacific Forum, brings together military leaders and government officials to discuss security for the U.S. and its allies. This year, the focus shifted with conversations dominated by AI and technology executives holding lucrative government contracts.

Speakers painted a picture of rapid transformation: autonomous drone swarms replacing manned aircraft, AI systems making targeting decisions in milliseconds, and nations racing to deploy machine learning across their entire military apparatus over weeks rather than years.

That world is already here in Ukraine. Eric Schmidt, former Google chairman and now executive chairman of Relativity Space, has spent two years helping Ukraine develop AI-powered drones. He claims the drones he worked on have flown missions with zero civilian casualties.

"The drones that I'm involved with do not miss," Schmidt told reporters at the forum. "Whenever I hear about collateral damage, I think, 'That's a bug. That's not the goal.' We have a policy of zero collateral damage, and so far, we've achieved our policy."

Ukraine's experience, Schmidt explained, has demonstrated how quickly AI can accelerate military innovation. The typical cycle time for one side to learn from and counter the other's tactics has compressed to six to seven weeks. Ukraine now aims to produce 20 million drones annually, a scale that dwarfs current U.S. military plans for just 1,000 drones per Army division.

"In war, because it's about life and death, innovation is faster," he said.

Those lessons are also being studied by China, which speakers identified as America's principal AI competitor. But while U.S. technology companies focus on achieving artificial general intelligence, China has adopted a different approach.

"In China, they have a different mindset," Schmidt said. "They're not going after AGI. They're going after using AI in everything."

He said China's military strategy likely centers on deploying large numbers of drones. This comes as North Korea builds up its forces through Russian cooperation and China maneuvers aggressively around Taiwan and the South China Sea.

"The Chinese military has not publicly announced its AI strategy, but you have to assume that it's essentially large swarms of DJI drones," Schmidt said. "That's the only reasonable scenario you can imagine."

Evidence of China's progress surfaced with DeepSeek V3, an AI model that appears to match or exceed American programming tools despite being developed without access to advanced semiconductors that U.S. export controls were designed to restrict.

China announced in September 2025 that it would focus on domestic chip production rather than purchasing American technology.

"We're up against an opponent that's really smart," Schmidt said. "The likelihood is that the majority of the world's users of AI will be using Chinese models and Chinese products embedded in Chinese exports."

Silicon Valley's wealthiest entrepreneurs have also entered defense work. Marc Benioff, chairman and CEO of Salesforce, described how his company now works extensively with military agencies, fully automating Veterans Affairs systems and providing AI-powered customer service that handles roughly half of all interactions autonomously.

"We fully automated and rewrote every system," he said.

But Benioff also delivered warnings about AI's current limitations. He pointed to recent cases where AI chatbots were involved in child suicides and noted that doctors using AI for diagnoses experience higher rates of misdiagnosis.

"This is not the messiah," Benioff said. "This is very complex, unwieldy, unyielding technology."

A central theme at the forum involved the tension between AI's potential and its real-world limits. Chris Murphy, AI technical director at the U.S. Department of Defense Innovation Unit, identified the Pentagon's fragmented data systems as a major obstacle.

"The traditional organizational structure of the military, with all the different forces, is too siloed," Murphy said. "We need the Data Force to be able to come across the whole agencies to really be able to deliver the AI across a multidimensional operation."

Aki Jain, chief technology officer at Palantir, emphasized that successful AI deployment requires constant testing and iteration as models evolve.

"We have to accept that we don't actually fully understand, especially large language models," Jain said. "The only thing that gets us there is experiential testing."

The implications ripple through the entire defense industrial base. Mike Cadenazzi, assistant secretary of defense for Industrial Base Policy, said applying AI to improve manufacturing efficiency remains challenging for smaller defense suppliers.

"As you get into tier two and tier three, where the firms are smaller, oftentimes they're multi-industrial," Cadenazzi said. "How can we apply AI tools to that? So, we're working with our partners to find opportunities for those firms to test their capabilities in those areas."

The forum highlighted that the race to deploy AI in military operations depends on multiple factors beyond building better models. Countries must also rapidly deploy systems, manage vast amounts of data, scale production and iterate based on combat experience. In some of these areas, China's centralized system may offer advantages over America's more bureaucratic processes.

At the same time, speakers emphasized the U.S. military's strengths, including a flexible command structure and highly trained personnel. The question is whether those advantages can be leveraged quickly during accelerating technological competition where geopolitical stakes rise.

Source: Courthouse News Service