There’s a growing rift in the AI world. On one side are tech leaders racing to build not just powerful models, but superintelligent systems that could upend society and labor markets as we know them.

On the other side, new research is raising doubts. Despite the hype, today’s LLMs, which are seen as the gateway to superintelligence, face three fundamental limitations:

LLMs got us this far, but getting to what’s next might require something new—and that breakthrough is still ahead.

We don’t need AGI to see transformative value. AI is already revolutionizing how we work and how technology works on our behalf. In other words, we shouldn’t confuse limitations in superintelligence with a lack of usefulness.

Here’s another thought: Maybe the obsession with “superintelligence” has less to do with capability, and more to do with attracting capital?

Either way, the real intelligence might lie in the comments section. WSJ readers hilariously told us how they really feel about AI.

If you’ve ever tried to build an app or website, whether for yourself or your company, you know great UI is non-negotiable. And traditionally, turning that vision into reality meant relying on skilled designers and front-end developers.

✨ Enter Stitch, Google’s new AI tool that bridges the gap between design and development. Just type a prompt or upload an image, and Stitch generates clean UI designs and the front-end code to match. Powered by Gemini 2.5 Pro, it’s built for rapid iteration and collaboration.

Here’s how it works:

Whether you’re a solo founder, a designer tired of dev handoffs, or a developer who dreads mockups: Stitch makes going from concept to creation insanely faster.

AI is more than just using ChatGPT.

It’s about understanding how the technology works—and applying it in practical ways to solve real business problems.

The AI for Business & Finance Certificate Program from Wall Street Prep + Columbia Business School Exec Ed helps you move beyond the hype.

You’ll demystify today’s most important tools and learn just enough Python to streamline workflows, automate tasks, and make smarter decisions.

No coding experience needed—you’ll get step-by-step guidance from world-class faculty.

👉 Save $300 with code SAVE300.

Enroll Today & Save $300

Europe is finally buying into Nvidia CEO Jensen Huang’s pitch for “sovereign AI”, the idea that nations need their own AI infrastructure shaped by local language and culture. After years of leaning on U.S. tech giants, leaders in the U.K., France, and Germany are now throwing billions at homegrown infrastructure, with new data centers, cloud platforms, and even “AI gigafactories” in the works. It’s a major pivot for the continent and a strategic win for Nvidia, whose chips are still the engine behind this independence movement.

AI is speeding up ad creation in a big way, turning long creative cycles into quick, data-driven sprints. That’s a game-changer for marketers under pressure to do more with less, deliver faster, and stay relevant across multiple channels. Leaders like WPP and Publicis are betting big on AI to boost efficiency without sacrificing creativity. The real takeaway? In a world where attention is short and expectations are high, AI is becoming essential to compete.

Despite lingering controversy and industry pushback, AI is quietly going mainstream in Hollywood. Studios like Lionsgate and AMC are teaming up with AI companies like Runway to speed up content creation, slash costs, and rethink what’s possible on screen. The Runway AI Film Festival saw a jump from 300 to 6,000 submissions in just a year, as more filmmakers lean into generative tools to tell surreal, boundary-pushing stories. While concerns about job loss persist, industry leaders say AI is more likely to transform filmmaking than replace it, opening the door to entirely new roles and workflows.

Most game-playing AI systems today, like AlphaZero, are unbeatable at chess and Go by training on fixed rules through endless self-play (called Reinforcement Learning). Impressive? Sure. But drop them into a new game with different rules, and they’re toast.

These models rely on being trained on the playbook ahead of time, making them great specialists but terrible improvisers.

Enter the Gardner test, a proposed new benchmark that flips the script:

If it works, we’d have AI systems that can follow complex instructions, adapt on the fly, and still color inside the lines. That could be key to building smarter, safer AI platforms that perform reliably in high-stakes, real-world scenarios.

Because AI’s real breakthrough won’t be in what it remembers, but how it handles the unknown.