A Leading AI Expert Just Warned of an Incoming "Flood of Intelligence"—And the Math Behind It Is Staggering

Ethan Mollick, a prominent voice in AI and innovation, just published an important essay titled “Prophecies of the Flood.” His thesis? We’re seeing urgent new signs that superintelligent AI may not be decades away—it might be just around the corner.

Mollick isn’t alone. A growing number of researchers from top AI labs are sounding alarms (or ringing bells, depending on your perspective) that we’re nearing a fundamental breakthrough in AI. And these rumblings aren’t the typical tech hype—many are coming from people inside major labs who appear genuinely convinced we’re on the brink of something huge.

So, what exactly are these so-called prophecies? And how should businesses and individuals make sense of the frenzied conversation around superintelligence?

I got the answers from Marketing AI Institute founder and CEO Paul Roetzer on Episode 130 ofThe Artificial Intelligence Show.

Mollick’s “Signals of the Flood”

Mollick calls out multiple indicators that advanced AI might be arriving sooner than most people expect:

1. OpenAI’s o3 model. This new model reportedly hit an 87% accuracy on a test that even PhDs, using the internet, only averaged 34% on—if they ventured beyond their specialty. It’s also solving high-level math problems and crushing ARC AGI benchmarks.
2. Early AI agents. Mollick points to Google’s Gemini and its Deep Research feature, highlighting how he asked it to analyze 173 websites and generate a 17-page report in minutes. That’s basically a personalized AI researcher at your beck and call.
3. Researchers sounding the alarm. Where typical tech hype might come from marketers or evangelists, Mollick notes these signals are coming from within AI labs—a major difference from the usual cycle of big product promises.

However, despite the drumbeat around superintelligent systems, Mollick reminds us to keep a level head. Even if AI labs reach something resembling “AGI,” mass integration into businesses and society often lags behind the lab breakthroughs.

He also warns that while AI researchers focus on alignment and safety, less attention seems to go toward how society should adapt to (or shape) this powerful technology.

Why Most People Don’t "Get It" Yet

Roetzer points to a telling tweet by Vedant Mishra, a DeepMind researcher with past AI stints at OpenAI and HubSpot. Mishra says only a “few hundred people in the world” truly understand the scope and speed of what’s coming, because you have to simultaneously grasp:

• Rapid algorithmic improvements
• Recursive self-improvement via reinforcement learning
• Tens of billions of dollars committed to AI data centers (“intelligence factories”)

Either those experts are all wrong, Mishra says, “or everything is about to change.”

The Math That Changes Everything

To get a sense of how much is already changing—and how unprepared businesses are—Roetzer then laid out a simple but startling example: 

Using today’s AI capabilities—no future breakthroughs required—you can likely unlock a 10% efficiency gain in nearly any knowledge worker’s day right now. 

What does that mean? Some back-of-the-napkin math:

1. A single employee working 176 hours per month saves about 18 hours at a 10% efficiency boost.
2. In a 10-person company, that’s 180 hours a month total—equivalent to a full extra employee.
3. In a 1,000-person organization, you’re effectively freeing up 100 employees worth of time each month.
4. Scale that to 10,000 employees, and its 1,000 full-time equivalents of monthly productivity, without hiring a single additional person.

And that’s if you only get 10%. Realistically, Roetzer sees 20–30% as completely attainable with better training and adoption—today, right now, with off-the-shelf AI tools.

How do you do that? Roetzer recommends these steps:

1. Map out tasks. Take one person’s role, break it down into 3–5 major tasks they perform repeatedly.
2. Build 3-5 custom GPTs. These AI “assistants” handle chunks of the person’s workload—like drafting content, analyzing data, or summarizing research. Build 1 (no coding required) to assist with each of the person’s major tasks.
3. Realize the savings. Even a modest 10% efficiency jump translates to major productivity gains when multiplied across your entire company.

Now, imagine what becomes possible if superintelligent AI becomes a reality.

Why This Matters (Even If You Don’t Care About “Superintelligence”)

But, at the end of the day, this matters to your business right now whether or not we reach superintelligence, says Roetzer.

Whether your organization aims to reduce headcount or expand output, ignoring AI’s efficiency gains is a huge missed opportunity. Competitors who adopt quickly will gain the upper hand.

The biggest barriers are lack of AI literacy and internal inertia, not technological limitations. The tools already exist; using them is the challenge.

As Mollick and the labs chase superintelligence and advanced AI agents, plain-old generative AI is quietly reshaping companies from the inside out.

“Everybody has this choice,” says Roetzer. “You can keep doing what you're doing, or you can accept that you can change things and you accelerate your AI literacy and capabilities.”

Bottom Line: Don’t Wait for the “Flood” to Hit

Even if you’re not convinced superintelligence is right around the corner, a transformation of the workforce is already underway. As Mollick, Altman, and countless insiders keep warning, ignoring AI now could be like ignoring the early signs of a flood.

A 10% efficiency gain today—with the potential of doubling or tripling that in the near future—could make or break organizations in the next 1–2 years. Whether it leads to workforce reductions or an explosion of new initiatives, the result is the same: AI is changing the calculus of business faster than any technology in living memory.

Mollick’s question isn’t if the flood is coming—it’s whether we’ll be ready when the waters start to rise.