Top 5 Takeaways from Axios AI+DC

At the Axios AI+DC Summit in Washington, D.C., the conversation around AI felt notably more mature than the hype cycle that has dominated much of the last two years. The focus was less on speculative possibilities and more on the conditions needed to make AI work in the real world: energy, workforce readiness, regulation, verification, public trust and practical deployment.

Here are the themes that stood out:

1. AI is now an infrastructure story

Some of the most urgent conversations were about the systems beneath the technology: power demand, data centers, grid capacity, and the physical infrastructure required to support AI at scale.

AI is increasingly tied to national readiness, industrial capacity and economic resilience. It is not just about who builds the best model. It is about who can support the energy and supply chain demands that come with widespread deployment.

That shift also changes the cast of characters. The AI conversation no longer belongs only to technologists. It now includes utilities, policymakers, labor, educators, defense leaders and companies thinking about long-term operational resilience.

2. Trust is becoming the real differentiator

If there was one undercurrent running through nearly every discussion, it was trust. Speakers returned repeatedly to deepfakes, identity fraud, privacy concerns, child safety and public skepticism about where AI is headed. Many cited that less than 20% of U.S. adults think AI will have a positive impact on the country over the next 20 years.

The implication is clear: technical progress alone will not determine adoption. Public comfort with AI is not keeping pace with the speed of innovation, and that gap creates real reputational risk for institutions trying to move quickly.

Vague references to “responsible AI” are no longer enough. Stakeholders increasingly want specifics: how content is authenticated, how people are protected from fraud, where human review sits in the process, what the limits are and how accountability is enforced.

In other words, trust is no longer a supporting message. It is becoming central to the product, policy and communications strategy itself.

3. The human question is still the hardest one

The promise of AI is enormous, but so is the anxiety surrounding what it means for work. Some speakers touted what AI will do for jobs – including the need for more electricians to build data centers. The conversation touched on the need for AI literacy, the role of upskilling and the reality that some job categories may shrink even as new forms of work emerge.

The question is not simply whether AI can perform tasks once done by humans. It is whether organizations and governments can help people navigate the transition in ways that feel fair and economically viable. In that sense, the future of AI may depend as much on education and workforce strategy as it does on model performance.

Notably, many of the strongest voices were not talking about a human-free future. They were talking about augmentation. One phrase that stood out was “human-machine teaming” or the idea that AI should support human judgment and improve decision-making rather than eliminate people from the process altogether.

That is an important reminder for organizations communicating about AI. It is no longer enough to speak only to efficiency and innovation. Stakeholders want to know how leaders are preparing employees and what opportunities will exist in the transition.  

As Padilla’s C-suite research found, nearly nine in ten leaders say their organizations

are adopting AI aggressively or selectively. However, employees are significantly less likely than leaders to view AI as a net benefit. They’re not necessarily opposed – it’s more that they’re unsure.

4. The policy challenge is balance

In the Space Race, it was the Soviet Union. In the AI race, we’re going head-to-head with China. And many of the speakers saw this as a race we can’t lose.

But how do you put guardrails in place without choking off innovation? That balancing act is easy to describe and hard to execute.

There was clear recognition that AI needs oversight, especially in higher-risk contexts. But there was also concern that poorly designed regulation could push innovators out of the U.S. or slow adoption while competitors move faster. The geopolitical dimension was impossible to ignore, with multiple speakers framing AI as a strategic race shaped not only by innovation, but by governance, economic policy and national security.

That leaves policymakers facing a difficult but necessary task: build rules that protect people without making it harder to compete. The most productive conversations were not framed as innovation versus regulation. They were framed around how to do both at once.

That is likely where the debate is headed next: less about whether guardrails are needed, and more about what smart, durable guardrails actually look like.

5. The strongest use cases are the most practical

Some of the most compelling examples discussed were not flashy or futuristic. They were operational.

The AI stories that landed best were those tied to everyday friction points: helping patients understand costs, improving claims and reimbursement processes, reducing administrative burden, supporting population health decisions, and using AI to make large amounts of information easier to navigate. Across sectors, the common thread was not novelty for its own sake. It was utility.

AI becomes more credible when it solves real problems that people already recognize. In healthcare especially, that means the most persuasive narrative may not be about futuristic transformation. It may be about quietly making care easier to access, easier to navigate and less administratively burdensome for patients, providers and systems.

The same principle showed up beyond healthcare, whether in fraud detection, coding, digital archives or content verification. The use cases that resonate most are the ones that remove friction, expand access and improve decisions in tangible ways.

Why it matters

For communicators, the lesson is straightforward: the most effective AI narratives right now are not the most ambitious. They are the most grounded.

They explain not just what the technology can do, but why it matters, who it helps and how the risks are being managed.

It’s also influencing, how we take in information – in the form of GenAI search results, for example.  

That is where trust gets built. And increasingly, trust is what will separate AI leaders from everyone else.