Photo by Mohamed Nohassi on Unsplash

Earlier this month, a video was posted of OpenAI‘s CPO, Kevin Weil, talking about how fast the cost of intelligence is dropping.

As he put it: “We’re able to distill smaller models at the same level of intelligence that previously required a much bigger model. Inference is getting cheaper. GPUs are getting faster. And you’ve got all of these trends converging.”

Just how fast is this happening?

According to Weil: “The cost of intelligence is coming down by something like 10X a year.”

Think about what that means.

We were all amazed when Chat GPT-3 was released on November 30, 2022. But a mere 30 months later, current AI models make those look like child’s play.

In fact, if you compared the two, you’d be shocked at how much better the GPT-4.o mini model is today. It reasons better, gives more accurate responses and hallucinates far less than GPT-3.

Yet, GPT-3 cost 100X what GPT-4.o mini costs to run today. That means AI isn’t just beating Moore’s Law, it’s blowing it away.

This means we could be closer to Artificial General Intelligence (AGI) sooner than we think…

Is AGI Around the Corner?
 

In 2005, the futurist Ray Kurzweil predicted a technological singularity by the year 2045.

This is the point when technological growth becomes uncontrollable and irreversible, resulting in unknown consequences for human civilization.

He predicted that an AI would enter a loop of successive self-improvement cycles, causing an explosion in intelligence, culminating in a powerful superintelligence.

This is also known as artificial general intelligence, or AGI.

The real fear is that humans have always been the apex predator of the intelligent food chain and we’ve never had to compete against anything smarter.

It appears Kurzweil underestimated how quickly we could arrive at AGI. In fact, research suggests that we’ve been consistently underestimating its arrival.

A year ago, Ark Research published this chart illustrating this fact. If the forecast error for AGI continues, we could be looking at a system much more intelligent than us by the end of next year.

If the current forecast is well-tuned, we are looking at 2030.

Keep in mind, this research was published even before intelligence costs fell dramatically. And also way before DeepSeek changed the calculus on what is possible.

I believe we are likely even closer to AGI than we think.

But how are humans faring with AGI right around the corner?

Are We Getting Dumber?
 

James Flynn is an intelligence researcher who, in 1980, published research about how people performed on IQ tests over time.

He observed that the average IQ score had been improving for generations, roughly three points per decade.

This rise in measured intelligence is now called the Flynn Effect, and it was thought to result from factors like better education, improved nutrition and more stimulating environments.

But recent studies suggest that American IQ scores are experiencing a phenomenon known as the “Reverse Flynn Effect”.

In other words, our IQ scores are going down.

A study published in the journal Intelligence analyzed 394,378 scores on intelligence tests taken between 2006 and 2018, revealing drops in IQ scores across multiple categories.

Interestingly, scores in spatial reasoning (3D rotation) generally improved from 2011 to 2018.

But scores decreased in verbal reasoning, visual problem-solving, analogies and computational/mathematical skills.

The decline was observed across age groups, education levels and genders, with the steepest drops among younger and less-educated test-takers.

Naturally, these findings are concerning. But they don’t necessarily mean that Americans are becoming less intelligent.

There are many valid reasons why our collective IQ could be dropping in these areas.

It could reflect changes in how people engage with information or perform on specific types of tests.

I’d wager that changes in education quality and emphasis on certain skills over others in the classroom plays a part too.

Even changes in perceived societal values might affect test performance. Heck, people just might not be motivated to perform well on these tests anymore.

But while all these factors are likely at play, the biggest potential factor that concerns me is our technological dependence.

I worry that as we continue to rely more on smartphones and AI to store and process information, we might be using our brains less for deep thinking.

In other words, as our machines are getting smarter, it’s quite possible we’re getting dumber.

What Happens in an AGI World?
 

Last month, Starbucks announced layoffs affecting over 1,000 corporate employees. Not its baristas. The company’s white-collar workers.

It was the latest addition to a troubling trend…

Unemployment for college-educated workers is rising faster than for other groups, and their wage growth is also lagging.

The big question is whether these are just short-term fluctuations, or if knowledge work is facing a more permanent shakeup.

Last May, unemployment rose above 4% for the first time in two years. And it has remained over that mark ever since.

Source: U.S. Bureau of Labor Statistics as of February 28, 2025

Of course, some of the more recent job losses are due to the Trump administration’s budget cuts at federal agencies and research institutions.

Johns Hopkins University, which depends on federal funding, announced 2,000 layoffs this month alone.

But many experts believe we’re witnessing a “meaningful transition in the way work is done,” as Carl Tannenbaum, the chief economist at Northern Trust recently told the New York Times.

He says: “I tell people a wave is coming.”

And that wave is already hitting some industries hard.

At this year’s Game Developers Choice Awards, the host highlighted “record layoffs” in the video game industry.

The finance sector is also slashing its workforce. Unemployment in finance-related jobs jumped about 25% from 2022 to 2024.

Wells Fargo, for example, has cut jobs for 16 straight quarters, with its home lending division seeing a nearly 50% reduction since 2023.

And the tech industry has been hit especially hard by layoffs for a few years now. According to layoffs.fyi, over 19,000 tech workers have already been laid off since the beginning of 2025.

Part of this is due to the modern corporate playbook shifting toward leaner cross-functional teams.

Amazon’s model — where employees juggle responsibilities across coding, marketing and other fields — is gaining traction. Companies like Starbucks and Nissan cite similar efficiency goals in their recent layoffs.

But I believe that the plummeting cost of intelligence is also a big factor here.

As I noted at the top, AI’s capability is growing exponentially while its price drops by a factor of 10 each year.

And as businesses continue to embrace AI, demand for knowledge workers will plummet. Especially if hires are expected to take on multiple roles, which will require AI to help streamline their workflow.

Just look at how AI is already affecting coding.

A recent study found that AI-assisted software developers boosted productivity by more than 25%, with the biggest gains among junior coders.

This is almost certain to erode the wage premium for experienced developers. It will also make some jobs redundant.

After all, novices could soon have AI do the bulk of coding for them.

And that brings us back to Kevin Weil of OpenAI. He just came out with a prediction that AI will surpass humans in competitive coding this year.

A mere two years ago, software engineers were in high demand. Many people thought coding was something that only highly intelligent people were able to do.

But what does it say about the value of human intelligence if AI can suddenly code better than humans?

Here’s My Take
 

As you can tell, I’m having a bit of an existential crisis about intelligence.

The optimist in me looks at the lowering cost of intelligence as a boon for humanity.

But I’m also a realist.

And as AI becomes cheaper and more powerful, it seems that the value of human intelligence in routine knowledge work is rapidly diminishing.

That means professionals in traditionally stable fields could be facing an uncertain future.

Harvard labor economist Lawrence Katz warns that while unemployment among college grads isn’t alarmingly high yet, “it may be in the next six months.”

It makes me wonder what kind of future my young daughters face. What will the job market look like in five years… a decade?

Once we achieve artificial superintelligence, will there even be a labor market at all?

All I know is that the rules of the game are changing.

And as AI continues to evolve, the economic value of human intelligence may be under threat.

---

More By This Author:

What’s Really Behind Elon’s Bid For OpenAi?
DeepSeek Accelerated America’s AI “Manhattan Project”
Is AI Coming For Your Job?