Video summary
Tyler Cowen argues that the widespread fear of artificial intelligence eliminating most jobs is highly unlikely and instead predicts a future characterized by near-full employment for those willing to work. He believes AI will act as a catalyst for creating new startups, projects, and industries rather than simply replacing human labor. Specifically, he highlights significant growth opportunities in sectors such as energy production, biomedicine, clinical trials, elder care, entertainment, coaching, and inspiration, suggesting that technology will expand the scope of economic activity rather than shrink it.
Despite this potential for job creation, Cowen points out a critical disconnect between technological advancement and government adaptation, noting that while national security agencies have begun meaningfully integrating AI into operations like military planning, most other branches of the U.S. government remain largely frozen in their current state. He identifies human writing as one specific area likely to become obsolete within the next decade because AI can produce content faster and smarter than humans; however, he draws a distinction between functional nonfiction, which machines will dominate, and personal memoirs by figures like Taylor Swift, where readers still value the authentic human voice behind the story.
A significant shift in Cowen's perspective concerns cybersecurity, which has recently emerged as an urgent issue driven by new AI capabilities from companies like Anthropic and OpenAI. These entities have developed programs capable of hacking into most systems humans have created, fundamentally changing the threat landscape. Although this development occurred only a year or two earlier than he had previously anticipated, Cowen emphasizes that even such a short delay is catastrophic because society is currently unprepared for these advanced cyber threats, making immediate attention to AI-driven security risks essential.
In conclusion, while governance structures struggle to keep pace with the rapid evolution of artificial intelligence, creating a sense of being "stuck," the overall outlook remains cautiously optimistic regarding economic vitality and employment availability. The key takeaway is that we must navigate this transition carefully by avoiding other poor decisions that could undermine progress, particularly in high-stakes areas like cybersecurity where new vulnerabilities are appearing sooner than expected. Ultimately, Cowen suggests that if humanity manages to adapt its institutions and focus on emerging opportunities without making critical errors, the future will be defined not by mass unemployment but by a transformed economy leveraging AI for unprecedented growth and efficiency.
Read the full video transcript
I fully expect we will have a future
with near full employment for anyone who
wants it, at least provided we do not
screw up on some other decisions.
Hello, I'm Tyler Cowen, professor of
economics at George Mason University and
an avid user of artificial intelligence.
If you speak to many people, they are
greatly worried that AI will take away
all of our jobs. This is an extremely
unlikely outcome. AI is going to enable
us to create new startups, new projects.
There will be incredible growth in the
energy sector, in biomedicine and
biomedical trials, more generally in
taking care of old people, entertaining
other humans, inspiring them, coaching
them.
If you compare the speed of
technological change in AI to changes in
governance, well, governance is mostly
frozen. Our national security
establishment in the United States and
the military has adopted AI in
meaningful ways, for instance, to plan
attacks in the war with Iran, but most
of the rest of our government has done
very little to incorporate AI into
making itself more efficient or more
productive. Right now, we are stuck.
So, what will be obsolete within, say, a
10-year time window? I think a lot of
human writing will be obsolete. The AIs
will do it better, they will do it
smarter. You might think you want to
read a human, but some human will put
his or her name on it. Now, some human
writing will continue. If Taylor Swift
writes a memoir, people do want to read
a memoir by Taylor Swift, but a lot of
nonfiction writing, I think that will be
left to the machines.
So, where have I changed my mind about
AI? Well, in the world of AI, it's very
important to speak in terms of exact
dates. And we are taping right now,
April 9th, 2026. So, what happened
earlier this week is that Anthropic
announced it had Muthos, which is a
program that has immense cybersecurity
implications. It seems it is basically
able to hack into most systems that
humans have created. OpenAI has let it
be known they have something of
comparable power. So, my belief revision
is that I now think cybersecurity is a
major AI issue, say a year or two before
what I previously thought. A year or
two, that may not sound like much, but
given that we're not at all prepared for
this, it coming a year or two sooner is
in fact a very big deal.