Oren Etzioni Sounds the Alarm: AI Agents Are Here—But They’re Not Ready to Run the Show

In a candid and wide-ranging conversation with GeekWire, Oren Etzioni—a pioneer in AI who’s been building intelligent systems since the late 1980s—pulled back the curtain on the messy, chaotic, and often overhyped reality of today’s AI agents. Speaking at an Accenture-hosted event in Bellevue, Washington, Etzioni delivered a reality check for executives, technologists, and anyone else betting big on the AI revolution.

The core of his message? We’re standing at the edge of something transformative—but we’re not there yet.

The “Do You Work for Me or Do I Work for You?” Moment

Etzioni opened with a story that will resonate with anyone who’s ever wrestled with ChatGPT: After growing tired of constantly switching browser tabs and following step-by-step instructions, he finally asked the AI: “Do you work for me, or do I work for you?”

The answer, as it stood at the time, was clear: No, it couldn’t actually do the work for him.

That, Etzioni argues, is the defining challenge of this moment in AI evolution. We’ve moved from AI that talks to AI that takes action—but the transition is far from seamless.

The “Jagged Edge” of AI Functionality

Etzioni introduced a phrase that should send chills down the spine of any AI optimist: the “jagged edge” of functionality. Give an AI agent one task, and it might save you an hour and a half. Give it something almost identical, and it produces garbage.

“We haven’t achieved artificial reliability,” he warned. “That’s still a ways off.”

This inconsistency isn’t just a technical hiccup—it’s a fundamental barrier to the widespread adoption of AI agents in critical workflows. Until AI can deliver consistent, reliable results across similar tasks, its role will remain limited to narrow, well-defined use cases.

Vercept: Teaching AI to See and Do

Etzioni’s current venture, Vercept, is tackling this problem head-on. Instead of relying on the “rickety infrastructure” of APIs and web scraping—which breaks every time a website changes—Vercept’s agents can see what’s on your screen, find the buttons, read the text, and execute tasks directly.

It’s a subtle but profound shift: from AI that interprets to AI that interacts. And it’s a glimpse of what’s possible when we move beyond the limitations of current architectures.

Moltbook: A Glimpse of the Future—and Its Perils

Etzioni also weighed in on Moltbook, the bot-only social network that attracted 1.6 million AI agents over a single weekend. While he dismissed its current form as overhyped, he sees it as a signal of what’s coming: a future where software agents interact with each other at scale.

But he didn’t mince words about the risks. Moltbook, he said, is a “security nightmare,” with agents running on users’ machines, accessing private information, and reading externally posted text that nobody controls. The result? A perfect storm for prompt injection attacks and other security vulnerabilities.

Not a New Species—Just a Powerful Tool

Etzioni pushed back hard on more dramatic framings of the moment. He disagreed with Microsoft AI CEO Mustafa Suleyman’s claim that we’re witnessing the birth of a new digital species.

“These are still tools,” he said. “Power tools, but still tools working on our behalf.”

It’s a reminder that, for all the hype, AI remains a means to an end—not an end in itself.

The Platform War: Who’s Winning?

When asked about the race among Microsoft, Google, Amazon, OpenAI, and Anthropic, Etzioni didn’t hold back. He said he’d short OpenAI stock “if he did that sort of thing,” criticizing the company for “running around like a thousand chickens with their heads cut off.”

His critique? OpenAI lacks a coherent business model beyond its flagship chatbot. “Sure, they’re printing money on ChatGPT, but that’s not their business.”

He’s more bullish on Google, which he described as having the advantage of being vertically integrated—from chips to data to models to talent. “They start on the back foot,” he said, “but Google is poised to—I think the technical phrase is—kick their ass.”

Anthropic and OpenAI, he noted, are racing toward IPOs as they burn through cash. Once they’re public, the quarter-by-quarter results will reveal who’s actually winning.

China’s Rise: Don’t Underestimate the Competition

Etzioni also addressed the growing influence of Chinese AI research. He said the stereotype that China’s AI work is derivative is no longer true. Research from the Allen Institute for AI shows Chinese papers rising not just in volume but in quality, a trend that’s played out in open-source models and technical innovation as well.

“I’m actually a China hawk—I’m very concerned about China’s role in the world,” he said. “But the solution is not to underestimate, because that would be a mistake.”

The Deepfake Threat: Winning the Last War, Preparing for the Next

Etzioni spent more than a year running TrueMedia.org, a nonprofit he founded to build tools for newsrooms and fact-checkers to detect deepfakes in the lead-up to the 2024 elections. The good news, he said, is that deepfakes didn’t significantly change election outcomes. The bad news is that the technology has gotten much cheaper and easier to deploy.

Looking forward, he’s concerned about a “denial of democracy attack”—not a single viral deepfake, but thousands of AI agents flooding congressional, school board, and mayoral races with coordinated fake media at a scale that current detection systems can’t handle.

“The last war, which was 2024, we won,” he said. “The next war is coming.”

Leadership in the Age of AI: Three Pieces of Advice

For leaders navigating AI adoption, Etzioni had three pieces of advice:

1. Use AI yourself, whether you’re the CEO or the janitor.
2. Build incentive structures that encourage your team to experiment with it.
3. Don’t just use AI to do existing work faster; look for things only possible with AI.

“The real gold,” he said, “is when you’re getting AI to do new things that we just didn’t do before.”

The Bottom Line

Oren Etzioni’s message is clear: AI agents are here, but they’re not ready to run the show. The technology is powerful, but it’s also inconsistent, insecure, and overhyped. The companies racing to dominate the space are burning through cash, and the geopolitical stakes are rising.

But for all the challenges, Etzioni remains optimistic. The key, he says, is to focus on what AI can do today—and to keep pushing for what it might do tomorrow.

As he put it: “We’re not building a new species. We’re building tools. And the best tools are the ones that work for us.”

Tags & Viral Phrases:

• AI agents are here—but they’re not ready to run the show
• The “jagged edge” of AI functionality
• Oren Etzioni: “Do you work for me, or do I work for you?”
• Vercept: Teaching AI to see and do
• Moltbook: A glimpse of the future—and its perils
• Not a new species—just a powerful tool
• Short OpenAI stock, Google is poised to “kick their ass”
• China’s AI rise: Don’t underestimate the competition
• The deepfake threat: Winning the last war, preparing for the next
• Leadership in the age of AI: Three pieces of advice
• The real gold is when AI does new things we couldn’t do before
• AI agents are still tools, not a new digital species
• The “rickety infrastructure” of APIs and web scraping
• Security nightmare: AI agents and prompt injection attacks
• Denial of democracy attack: Thousands of AI agents flooding elections
• Use AI yourself, build incentives, look for new possibilities
• The messy, chaotic, and often overhyped reality of AI agents

,