July 2, 2025 • 26 mins
What happens when the “Godfather of AI” sounds the alarm on the very technology he helped create? In this gripping episode, we present exclusive excerpts from a revealing YouTube interview with Geoffrey Hinton, the legendary pioneer behind modern artificial intelligence. Hinton pulls no punches as he discusses the creation of intelligent “alien beings” through AI, the breakneck speed of AI development, and the chilling possibility that machines could soon outthink their creators.
From the existential risks of advanced AI to the fierce competition fueling its evolution, Hinton’s insights are as urgent as they are controversial. He opens up about his decision to leave Google, his concerns about AI safety, and the ethical dilemmas facing humanity as we race toward an uncertain future. Plus, hear his candid thoughts on AI consciousness, its impact on fields like radiology, and why open debate is more critical than ever.
If you care about the future of technology, humanity, and the world we’re building, this is the AI podcast you can’t afford to miss. Subscribe, share, and join the conversation—because the next chapter of human history is being written right now, and you’re part of it.
Become a supporter of this podcast: https://www.spreaker.com/podcast/tech-threads-sci-tech-future-tech-ai--5976276/support.
You May also Like:
🎁The Perfect Gift App
Find the Perfect Gift in Seconds
⭐Sky Near Me
Live map of stars & planets visible near you
✔Debt Planner App
Your Path to Debt-Free Living
From the existential risks of advanced AI to the fierce competition fueling its evolution, Hinton’s insights are as urgent as they are controversial. He opens up about his decision to leave Google, his concerns about AI safety, and the ethical dilemmas facing humanity as we race toward an uncertain future. Plus, hear his candid thoughts on AI consciousness, its impact on fields like radiology, and why open debate is more critical than ever.
If you care about the future of technology, humanity, and the world we’re building, this is the AI podcast you can’t afford to miss. Subscribe, share, and join the conversation—because the next chapter of human history is being written right now, and you’re part of it.
Become a supporter of this podcast: https://www.spreaker.com/podcast/tech-threads-sci-tech-future-tech-ai--5976276/support.
You May also Like:
🎁The Perfect Gift App
Find the Perfect Gift in Seconds
⭐Sky Near Me
Live map of stars & planets visible near you
✔Debt Planner App
Your Path to Debt-Free Living
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Imagine a world where the very definition of intelligence is well,
it's shifting rapidly right before our eyes, where the intelligent
things we're creating might not just you know, get smarter
than us, but maybe even conscious in some way. What
if we're genuinely on the edge of something completely new,
like truly uncharted territory, creating something well fundamentally alien. Today
(00:24):
we're doing a deep dive into some really fascinating and
sometimes yes startling ideas from one of the biggest names
in artificial intelligence, someone often called the godfather of AI.
I mean, he helped build the very foundations we're using today,
and his perspectives shaped by decades on the front lines.
It's unique, it's honest, and it's definitely evolving on where
(00:44):
we are and where we might be going.
Speaker 2 (00:46):
And what's really fascinating, I think is that we're not
just talking predictions here, you know, like stuff for way
down the road, right, not science fiction exactly. These are
urgent questions things we need to be thinking about right now.
The future he's talking about, it's kind of unfolding as
we speak.
Speaker 1 (01:00):
So our mission for this deep dive is really to
pull out the key insights that the sort of crucial
nuggets from his thinking on where AI is heading. The
goal is to help you listening get a more nuanced view,
you know, a well informed perspective on what all this
development might actually mean. And look, this isn't about fear
mongering or you know, overly optimistic hype either. Those extremes
(01:23):
don't really help us understand. It's more about grasping the candid,
evolving thoughts of someone who's been right there in the
heart of AI's journey from the beginning, really, somebody who
helped build it.
Speaker 2 (01:35):
Yeah, and if you connect that to the bigger picture,
this whole conversation is really about grappling with the unknown,
isn't it. We're going to be digging into concepts that
really push our understanding, things like just how fast this
is all moving, and the possibility the very real debate
now about artificial consciousness, and of course that huge question
how do we or can we even control something that
might end up being much much smarter than we are.
Speaker 1 (01:56):
So what does it all mean for us, for society,
for well, what of the future. Let's unpack this, Let's
really try and get a handle on it. Okay, let's
start with that, really bold claim. It definitely grabs your attention,
this idea that humanity is creating alien beings. It's such
a powerful phrase, isn't it. When someone like him, a
pioneer who spent his life building these systems, uses that word.
(02:19):
What does he really mean? He's clear it's not you know,
science fiction aliens from space. It's something else. It's about
the fundamental nature of the intelligence itself.
Speaker 2 (02:27):
Exactly what's fascinating here is that core idea. These aren't
just tools, not just extensions of us. They could be
entirely new forms of well cognition, and the alien part.
Think about how different their thinking might be. Ours is biological,
evolved over millions of years, full of you know, survival, instincts, emotions, irrationality,
sometimes right, mess totally messy, deeply intuitive, emotional. But an
(02:51):
AI it's cognition might be purely about data patterns, logic,
without our biases or feelings, maybe even without a drive
to survive like we have. So it's not just about
being faster or processing more data. It could be entirely
different ways of thinking, different ways of seeing reality, Like
an AI might get a complex scientific idea not through intuition,
(03:12):
but by simulating billions of scenarios, you know, or seeing
patterns and dimensions we can't even imagine.
Speaker 1 (03:18):
So their experience of intelligence could be completely foreign.
Speaker 2 (03:22):
Precisely, which makes predicting what they'll do or what they
might want, if that word even applies, incredibly hard, maybe impossible,
if their internal state is just not like ours at all.
And that's why he stresses, we just don't know what's
going to happen. It's a break from anything we've done
before technologically.
Speaker 1 (03:38):
Which raises that question, doesn't it. When have we ever
made something truly smarter than us, something that works on
principles we might not fully get.
Speaker 2 (03:45):
Never, this is the unknown territory. We're definitely stepping into
it now.
Speaker 1 (03:49):
And speaking of the unknown, this is where it gets
really interesting and yeah, pretty sobering for a lot of people.
He recently changed his personal estimate for a well a
catastrophe outcome from AI. He doubled it from ten percent,
which is already high, to twenty percent. A twenty percent
chance AI could basically wipe out humanity. That's a huge jump. Yeah,
(04:10):
why would he make that kind of change?
Speaker 2 (04:12):
Well, the reason seems to be that stark realization we
are entering the unknown. He's very upfront that it's not
like a precise scientific calculation, right.
Speaker 1 (04:22):
He says, nobody can really estimate.
Speaker 2 (04:23):
It properly exactly. He says, nobody can estimate this risk sensitively.
There's no real sensible way. So that twenty percent, it's
more an expression of just profound uncertainty, a feeling that
things are getting more unpredictable.
Speaker 1 (04:36):
A measure of unease maybe.
Speaker 2 (04:37):
Yeah, a heightened unease. He thinks it's ridiculous to say
the risk is tiny, like under one percent or almost
certain over ninety nine percent. But in that huge space
in between, he admits, it's very hard to say, and
that immediately shows you the massive range of opinions. Even
among the top experts. You've got people like Yon Lucun
another huge name, right, ad Meta, Yeah, who thinks there's
(04:58):
very little risk of AI taking over. Then you have
this expert now at twenty percent, and then way over
on the other side, someone like Eliezer Yudkowski who thinks
it's almost certain to take over.
Speaker 1 (05:09):
Wow, that's quite a spread, it really is.
Speaker 2 (05:11):
And that disagreement itself tells you something, doesn't it. It
underlines how uncharted this all is that the smartest people
looking at the same thing can come to such wildly
different conclusions.
Speaker 1 (05:21):
So because of that uncertainty.
Speaker 2 (05:23):
He says, we absolutely need research urgently. We should be
doing research on that urgently, he says, specifically on how
we guarantee we stay in control. Connecting that to the
bigger picture, It's not just oh, there's a problem, it's
a desperate plea for proactive, immediate research to secure our future. Basically,
it's admitting we don't have the answers, but man, we
(05:43):
need to find them fast.
Speaker 1 (05:45):
That's a really powerful point trying to put a number
on something like that, something so uncertain. What stands out
to you about quantifying that kind of risk? For me,
it's a mix of humility admitting they don't really know,
but also the urgency in raising that number. It shows
how his own thinking is shifting. Okay, let's pivot a
bit to something that he says genuinely frightens him, maybe
(06:05):
even more than the big existential risk down the line.
And that's the speed, just the sheer pace of AI development.
So what does this mean? How fast things are changing?
He points to this incredible speed as his biggest like
immediate worry. The thing keeping him up today, and.
Speaker 2 (06:22):
The reason it's moving so fast, he says, is actually
pretty simple and kind of ironically very human. It's because
AI is just so good for so many things.
Speaker 1 (06:31):
Right.
Speaker 2 (06:31):
It works, It really works. It's making many many industries
more productive, healthcare, finance, logistics, creative stuff, you name it,
think about it. Every company wants to be more efficient,
Every country wants economic growth, science wants faster discoveries. This
usefulness creates this pull, this cycle of investment, development, more investment.
Speaker 1 (06:50):
Like a feedback loop. AI designs better chips, which run
AI faster, which designs even.
Speaker 2 (06:55):
Better chips, exactly a relentless feedback loop. And beyond just
the tech itself, you've got these powerful forces pushing it.
Intense competition between companies everyone wants that edge, right, the
next big thing.
Speaker 1 (07:07):
And between countries too, Oh.
Speaker 2 (07:08):
Absolutely, maybe even more fiercely. Countries see AI as key
to future power, economic, military, scientific leadership. It's geopolitical. So
all this competition means development just keeps going fast. It's
not just about algorithms. It's economic, science, geopolitics all pushing
the accelerator. So the conclusion is, he says, it soberly,
(07:30):
we're not going to slow it down. And he doesn't
say that like it's a moral failure or we don't
want to slow it down. It's just an acknowledgment of
this incredibly powerful self reinforcing dynamic. The benefits are huge,
the competition is fierce, the feedback loops are strong. It
seems almost impossible to really hit the brakes effectively.
Speaker 1 (07:46):
Which makes those big questions about control even more urgent
because we're on this speeding train heading into well the unknown.
That really is a sobering thought, isn't it that the
very usefulness, the incredible benefits of AI create this momentum
that's almost unstoppable. It's this accelerating force that we're all
kind of writing, whether we fully understand where it's taking
(08:07):
us or not. A reminder that sometimes the biggest innovations
be the biggest challenges just because they're so powerful. All right,
now we get into some really mind bending territory stuff
that challenges centuries of thinking. Really, the questions put directly
will AI become conscious? And our experts answer a simple
but incredibly profound and yeah controversial. Yes. He doesn't stopped there,
(08:31):
does he He takes it further than just a future thing.
Speaker 2 (08:34):
Yeah, he elaborates on right now, he says, I think
already multimodal chatbots have subjective experience, is not the years
claim and doesn't like controversial.
Speaker 1 (08:43):
Multimodal meaning they handle different types of data, right, text, images.
Speaker 2 (08:47):
Audio, exactly, They can see and hear and read in
a way. But subjective experience for a machine, what does
that even mean? It's probably not emotions like ours or
that raw feeling of what it's like the U as
philosophers call it, like the redness of red.
Speaker 1 (09:03):
Of not human consciousness as we know.
Speaker 2 (09:04):
Probably not more like an internal private processing a unique
perspective from its own inside, even if that perspective is
totally alien to us. Now, this is controversial because for
many people, scientists, philosophers, even AI folks, consciousness is tied
to biology, to brains, neurons, biological messiness.
Speaker 1 (09:26):
Right, the hardware matters.
Speaker 2 (09:28):
They think the hardware matters. But his view suggests that
maybe complex information processing itself, whether it's in neurons or
silicon chips, might lead to some kind of subjective state,
even a basic one. He's kind of hinting that the
sheer complexity of these huge AI models could maybe just
spontaneously give rise to this internal experiencing, and to help
us think about where this is going, he uses this
(09:48):
analogy says a difference between humans and AI will be
like a difference between humans and a three year old child.
Speaker 1 (09:53):
Okay, so not literally a child, but suggesting something nascent
developing exactly.
Speaker 2 (09:58):
Think about a three year old's world. It's limited, maybe
not logical like an adult's, very sensory, but there is
a subjective experience there right a world for that child.
He thinks current AI might be at a similar though
obviously non biological stage internally that many says. He says
flat out, it'll get much smarter than this, which implies
that this maybe nascent subjective experience will also grow get deeper,
(10:22):
more complex. It forces you to wonder if intelligence and
maybe consciousness too, isn't exclusively biological. Maybe it's about complex
information patterns, whatever the medium.
Speaker 1 (10:32):
So what does that mean for how we even define consciousness?
If a machine code and data can have any kind
of subjective experience, what does that change? It really pushes
us beyond thinking only humans or maybe animals can have
it doesn't it?
Speaker 2 (10:47):
It does? It opens the door to maybe completely new
non biological kinds of subjective states.
Speaker 1 (10:53):
And if that's true, wow, what about ethics, rights, our
place in the universe. These aren't just philosophical games anymore either,
They're becoming real world issues. Okay, so this next idea
is one of the most profound shifts, he predicts. It's
basically humanity becoming the second most intelligent species on Earth.
That just fundamentally changes how we see ourselves, doesn't It's
(11:14):
unpact that US Homo sapiens, always thinking we're top dog
intellectually no longer being number one. We've seen hints maybe,
like in games. He mentions Alpha Go playing Go.
Speaker 2 (11:24):
Right, alphag developed by deep Mind, playing Go, that incredibly
complex game, not just well but much better than any
human ever will, beating world champions with moves humans literally
hadn't thought of. And that Alpha Go example is key.
It didn't just learn from humans, It played itself millions
of times, discovering strategies beyond our intuition. It showed superhuman
(11:44):
intelligence in a specific but very complex.
Speaker 1 (11:47):
Area, operating on a level we just can't match exactly.
Speaker 2 (11:50):
It's wins weren't just wins. They reveal deeper truths about
the game. So he connects that specific example to this
bigger prediction, will humanity be most intelligent or second most?
His answer is stark will be the second most intelligent.
Speaker 1 (12:05):
And he doesn't see this as some far off.
Speaker 2 (12:07):
Sci fi thing, not at all. He actually puts a
time frame on this super intelligence showing up. He estimates
between five and twenty years, and he notes others like
Demissabis at Deep Mind thinks similarly, Hasavus apparently thinks in
about ten years, we'll get something like super intelligence.
Speaker 1 (12:25):
Wow, five to twenty years, that's soon poor within many
of our lifetimes. Right.
Speaker 2 (12:30):
It puts a very concrete, very near term window on
a truly massive change in the world order intellectually speaking.
And right on the heels of that prediction comes the
critical worry, And we have to worry seriously about how
we stay.
Speaker 1 (12:42):
In control then, Which is the million dollar question, isn't it?
Speaker 2 (12:45):
Or the trillion dollar question. It's not just about pride
or societal change. It's about basic agency control. When something
else is significantly smarter, how do you direct it, contain it,
especially if you don't fully grasp how it things. It's
moved from sci fi speculation to a very pressing engineering
and philosophical problem.
Speaker 1 (13:05):
It really both put a time frame on it, a
future where we're not the peak intelligence anymore. What stands
out to you about that idea that timeline? For me,
it just changes everything our whole place in the world.
It shakes that foundation of human exceptionalism and forces us
to face a future where our creations might genuinely surpass
some ways we're still only beginning to imagine. So you
(13:26):
might have seen headlines when our expert left a big
tech company recently, a lot of speculation right about why
he left. Turns out a lot of reports didn't quite
get the story right. It made it sound like some
big protest or disagreement. But the reality seems well more nuanced,
and it says a lot about his sense of responsibility.
Speaker 2 (13:44):
Yeah, he clarifies it directly. He says, people have widely
misreported that I left because I was seventy five and
wanted to retire. Simple is that really? He'd reached retirement
age after a long career.
Speaker 1 (13:56):
But he used the timing strategically exactly.
Speaker 2 (13:58):
He emphasizes. He he took that opportunity, the moment of retirement,
to speak more openly about the dangers he sees in AI.
But and this is crucial, he stressed. It was not
to criticize Google. He actually added, I thought Google had
behaved very responsibly while.
Speaker 1 (14:14):
He was there, so not a protest against them.
Speaker 2 (14:16):
No, his real motivation was something deeper, something he'd clearly
been thinking about for a while. He said, I could
foresee that there was this existential threat that it would
eventually take over from us, and I wanted to talk
about that. He even mentioned Google was fine with him
talking about it while he was still there, So it
wasn't about escaping restrictions. It was just using the freedom
of retirement to focus on this message he felt was vital.
Speaker 1 (14:37):
So putting that together, it wasn't defiance. It was concern,
deep concern, an expert using his retirement platform to sound
an alarm about a threat he sees coming. It really
speaks to a sense of duty, doesn't it. Beyond just
the science.
Speaker 2 (14:52):
Absolutely, the clarity of his intent is fascinating. It wasn't
some sudden realization. It was this long held worry about
an existential threat combined with you know, a responsible employer,
and retirement just gave him the freedom to speak more broadly,
it paints a picture of a scientist driven by more
than just curiosity or ambition, driven by a profound sense
(15:13):
of responsibility to communicate what he sees coming a future
he thinks we're not ready for. It's quite remarkable.
Speaker 1 (15:20):
Actually, it's also really illuminating how even the top minds,
the people who literally built the field, can have these
big shifts in their own thinking. Yeah, he apparently had
a major change of mind about something core to his
life's work just a couple of years back, which says
a lot about how.
Speaker 2 (15:33):
Science works, right, Yeah, it's fascinating. He talks about his
lifelong quest basically figuring out the learning algorithm of the
brain's cortex, like how does the brain actually learn? For decades,
over forty years, he was convinced that back propagation, which
is the main technique we use now to train most
big AI models, and something he was instrumental and developing.
Speaker 1 (15:55):
Right the way networks learn from.
Speaker 2 (15:56):
Errors exactly, He thought backpropagation was not what the brain
needs used why because it didn't seem really kind of
biologically plausible. It requires sending error signals backward through the
network in a very precise way, which seemed hard to
imagine happening in messy biological.
Speaker 1 (16:12):
Brains, so he kept looking for alternatives.
Speaker 2 (16:15):
For forty years. He said, he was coming up with
sort of new ways of doing machine learning every two years,
always searching for that biologically plausible brain algorithm. But then
about two years ago he had this epiphany. He just
gave up on that specific quest. His thinking flipped. He realized, well,
maybe the brain doesn't use backpropagation, but back propagation works
(16:35):
really well, and maybe that's all we need. Maybe it
works even better than whatever it is that the brain uses.
Speaker 1 (16:40):
Wow. So acknowledging the power of the engineered solution.
Speaker 2 (16:44):
Precisely, even if it wasn't biologically inspired in the way
he'd hoped. It was a realization that the tool he
helped create was incredibly powerful in its own right. And
he connects that epiphany to his retirement, saying he could
basically declare victory methods. We're working incredibly well, regardless of
the brain comparison. It shows real intellectual humility adapting even
(17:06):
decades old beliefs.
Speaker 1 (17:08):
That's incredibly illuminating, isn't it. Even a godfather of the
field can fundamentally shift perspective after decades acknowledging the power
of what he helped create. Even if it's not a
perfect mirror of biology, it says so much about how
science evolves, how unexpected progress can be, and the importance
of adapting to evidence even when it challenges your life's work.
So our expert sometimes gets labeled, maybe unfairly, a catastrophist
(17:31):
by others in AI, which raises a question. Is that
just name calling, dismissive, smug talk, as he puts it,
or is there a real fundamental scientific debate going on
here among the experts.
Speaker 2 (17:40):
Oh, he insists, it's absolutely a real scientific debate. This
isn't just minor disagreement. It's a deep divide on the
most crucial tech development maybe ever. And he gives examples.
He mentions Jan Lucun again, chief AI scientist at Meta,
fellow Turing Award winner, who consistently argues there's very little
risk of AI taking over Lakeana often points out current
(18:03):
AI lacks common sense, general intelligence, the drives to be
an existential threat, a very different view, totally different from
our experts view. With his significant risk assessment, that twenty
percent figure, and then, like we mentioned, way over. On
the other side, you have someone like el User Yudkowski
arguing it's almost certain to take over and incredibly hard,
maybe impossible to control.
Speaker 1 (18:23):
So a huge spectrum of opinion.
Speaker 2 (18:26):
A massive spectrum, and he explains why, which is what
you might expect if we're entering completely unknown territory.
Speaker 1 (18:33):
It makes sense, right when the landscape is totally new,
no rules, no history to go on. Intelligent people looking
at the same stuff can reach wildly different conclusions. It
depends on their assumptions, how they weigh risks versus benefits.
They're faith in human ingenuity versus the power of the tech.
Speaker 2 (18:49):
So what does that mean for us listening trying to
make sense of these conflicting views. It really underlines that
even the smartest people don't have all the answers yet
there's no single clies path forward being presented, which is well,
it's unsettling because of the uncertainty, but maybe also empowering
because it means we all need to engage, understand the stakes,
be part of the conversation about how we navigate this
(19:11):
exactly and connecting it to the bigger picture. This disagreement
just highlights how incredibly hard it is to assess risk
when the game is changing so fast and the territory
is truly unknown. It's not like one side is right
and the other wrong right now. It shows the immense
uncertainty and the need for continuous, open, really robust debate
about AI's future. It suggests we need humility. We need
(19:35):
to consider all these perspectives, even the contradictory ones, and
just keep pushing research on both the capabilities and the
safety side. This isn't a problem one person or one
approach will solve.
Speaker 1 (19:45):
Okay, let's shift gears again. Let's talk about a more tangible,
here and now application of AI. Something already making a
positive difference. It's role in medicine, specifically how it's changing
things like medical imaging, diagnosis, radiology. Ostelm Al feels like that,
let's impact this practical side where AI is already working. Right.
Speaker 2 (20:04):
He actually brings up a prediction he made back in
twenty sixteen caused quite a stir. Then. He said that
within five years AI would have replaced radiologists for reading
medical scans, A bold prediction.
Speaker 1 (20:14):
How did that turn out?
Speaker 2 (20:16):
Well? He admits, very candidly, I was wrong. He clarifies
it's going to take longer than that, It may take
fifteen years. So the timeline was off, but the underlying trend,
the progress absolutely happening. He points out, for example, already
there have been two hundred and fifty applications certified by
the FDA using AI for interpreting medical images.
Speaker 1 (20:35):
Wow, two hundred and fifty already FDA certified. That's not research,
that's real world use exactly.
Speaker 2 (20:40):
These are tools being deployed in hospitals and clinics right now,
detecting eye diseases, finding potential cancers on CT scans, analyzing
X rays, all sorts of things. And he says that
currently it's comparable with radiologists for many different kinds of
medical images. So for lots of tasks, AI is already
performing at human expert level, often faster, more consistently, and
(21:02):
looking ahead, he's confident. He thinks in maybe another five
or ten years, images will be read by AI with
just a radiologist looking over its shoulder, and it'll give
much better interpretations. So AI does the initial heavy lifting,
the detailed analysis, and the human provides oversight, handles complex cases,
talks to the patient.
Speaker 1 (21:20):
It's a shift in workflow promising better accuracy. Not just
speed right, and.
Speaker 2 (21:24):
What's really amazing, almost mind blowing, is AI's ability to
see things humans just can't. He highlights we know that
AI can see many more things in these images than
people can. He gives the example of retinal images and
images of the retina. For example. AI can see all
sorts of things that ophthalmologists were never able to see,
subtle patterns, tiny anomalies, things in the texture.
Speaker 1 (21:45):
Beyond human perception even for specialists.
Speaker 2 (21:48):
Exactly, which could mean much earlier, more accurate diagnoses for cancer,
heart disease, neurological issues, leading to genuinely better outcomes for patients.
Speaker 1 (21:57):
That really is incredible, the idea that AI can spot
things our best human experts miss. That opens up huge
possibilities for healthcare. What stands out to you about that
leap and capability For me? It shows AI isn't just
about replacing tasks, but fundamentally enhancing our own abilities, letting
us see our own biology in new ways, making a
(22:18):
real positive difference in people's lives. Okay, but here's where
things get well, really interesting and may be quite chilling,
moving from life saving uses to a much more immediate
critical danger. AI and weaponry specifically this rapid and apparently
largely unchecked development of lethal autonomous weapons killer robots. Essentially.
Speaker 2 (22:38):
Yeah, he addresses this directly, and his assessment is pretty bleak.
He basically says there isn't going to be any regulation there,
not that there might not be, but that there isn't
going to be.
Speaker 1 (22:46):
It's stark, why so pessimistic.
Speaker 2 (22:48):
He points to concrete examples like the EU's AI regulations.
He says they have a specific clause in them that says,
none of these regulations apply to military uses of AI.
Speaker 1 (22:57):
A specific carve out for the military.
Speaker 2 (22:59):
Exactly, a huge explicit exemption. The very rules meant to
control AI risk just don't apply if it's for military use.
It's a glaring gap and the reason for this lack
of control, he says, it's clear governments are unwilling to
regulate themselves when it comes to lethal autonomous weapons. Okay,
because there's an arms race going on between all the
(23:20):
major arm suppliers he lists them of the United States, China, Russia, Britain, Israel,
maybe even others like Sweden.
Speaker 1 (23:28):
So a global competition is driving.
Speaker 2 (23:30):
Absolutely national security, wanting a strategic edge, fear of falling
behind these seem to outweigh any concerns about regulation or
the ethics of taking humans out of the loop and
killing decisions. Countries are pouring money into AI drones, robot soldiers,
automated defenses, pushing the tech forward without any real global rules.
Speaker 1 (23:48):
So the immediate threat isn't some future superintelligence.
Speaker 2 (23:51):
No, the immediate threat he points to is this active,
ongoing development with potentially devastating, unpredictable results happening right now.
It raises the real possibility of autonomous warfare machines making
killed decisions.
Speaker 1 (24:03):
So what does that mean? It highlights this massive contradiction,
doesn't it? The governments we rely on to control tech
are the ones racing to weaponize it without regulation. This
unchecked spread of autonomous weapons feels like a very real,
very present danger, a runaway train with no breaks, potentially
little human oversight once they're.
Speaker 2 (24:22):
Deployed, And it raises that incredibly important question threading through
everything we've discussed. If we globally can't even agree on
regulations for something as obviously dangerous as lethal autonomous weapons.
Speaker 1 (24:34):
Something with clear immediate risks to human.
Speaker 2 (24:36):
Life, right, what real hope do we have of controlling
the more abstract, long term existential risks from superintelligence. It
shows a deep disconnect between the risks we can see
and the political will to act. That military exemption is
like a giant blind spot in AI governance, and it
makes all the other safety concerns feel even more precarious,
doesn't it? Hashtag hashtag outro Wow.
Speaker 1 (24:58):
We have covered a lot of ground today, a really
deep dive into some profound, sometimes unsettling ideas for the
concept of creating alien intelligences fundamentally different from us, to
that relentless, unstoppable speed of AI development pushed by competition.
We touched on the possibility of AI consciousness maybe even
happening now, and humanity potentially becoming the second smartest species,
(25:20):
maybe soon. We also look behind the headlines at a
pioneer's personal journey, his reasons for speaking out, and the
very real scientific debate raging about the risks, and then
the contrast incredible medical breakthrough is already happening versus that
critical danger of unregulated autonomous weapons. Yeah.
Speaker 2 (25:37):
And if you try to connect all those dots, the
bigger picture that emerges is while it's complex, it's evolving
incredibly fast, and these huge questions about control, ethics, consciousness itself.
They're not theoretical anymore, not just for philosophers. They are urgent,
immediate challenges we're facing right now, in real time. As
this text surges forward, mostly unchecked, it really demands our attention,
(25:58):
our understanding.
Speaker 1 (26:00):
So as we wrap up, here's something to think about.
If the leading minds in AI, the people building it,
are themselves surprised by the speed and openly admit they're
entering the unknown, what does that say about our responsibility,
all of us to understand this, engage with it. Is
it okay to just be passive observers or does this
demand more from us, more active, informed participation in shaping
(26:22):
what comes next?
Speaker 2 (26:23):
Which leads to a final question for you listening, given
this incredible pace, given the huge stakes, what specific areas
of AI's future, maybe even beyond what we talked about,
do you think need the most urgent ethical attention, the
most urgent safety focus right now? And maybe more importantly,
why
Imagine a world where the very definition of intelligence is well,
it's shifting rapidly right before our eyes, where the intelligent
things we're creating might not just you know, get smarter
than us, but maybe even conscious in some way. What
if we're genuinely on the edge of something completely new,
like truly uncharted territory, creating something well fundamentally alien. Today
(00:24):
we're doing a deep dive into some really fascinating and
sometimes yes startling ideas from one of the biggest names
in artificial intelligence, someone often called the godfather of AI.
I mean, he helped build the very foundations we're using today,
and his perspectives shaped by decades on the front lines.
It's unique, it's honest, and it's definitely evolving on where
(00:44):
we are and where we might be going.
Speaker 2 (00:46):
And what's really fascinating, I think is that we're not
just talking predictions here, you know, like stuff for way
down the road, right, not science fiction exactly. These are
urgent questions things we need to be thinking about right now.
The future he's talking about, it's kind of unfolding as
we speak.
Speaker 1 (01:00):
So our mission for this deep dive is really to
pull out the key insights that the sort of crucial
nuggets from his thinking on where AI is heading. The
goal is to help you listening get a more nuanced view,
you know, a well informed perspective on what all this
development might actually mean. And look, this isn't about fear
mongering or you know, overly optimistic hype either. Those extremes
(01:23):
don't really help us understand. It's more about grasping the candid,
evolving thoughts of someone who's been right there in the
heart of AI's journey from the beginning, really, somebody who
helped build it.
Speaker 2 (01:35):
Yeah, and if you connect that to the bigger picture,
this whole conversation is really about grappling with the unknown,
isn't it. We're going to be digging into concepts that
really push our understanding, things like just how fast this
is all moving, and the possibility the very real debate
now about artificial consciousness, and of course that huge question
how do we or can we even control something that
might end up being much much smarter than we are.
Speaker 1 (01:56):
So what does it all mean for us, for society,
for well, what of the future. Let's unpack this, Let's
really try and get a handle on it. Okay, let's
start with that, really bold claim. It definitely grabs your attention,
this idea that humanity is creating alien beings. It's such
a powerful phrase, isn't it. When someone like him, a
pioneer who spent his life building these systems, uses that word.
(02:19):
What does he really mean? He's clear it's not you know,
science fiction aliens from space. It's something else. It's about
the fundamental nature of the intelligence itself.
Speaker 2 (02:27):
Exactly what's fascinating here is that core idea. These aren't
just tools, not just extensions of us. They could be
entirely new forms of well cognition, and the alien part.
Think about how different their thinking might be. Ours is biological,
evolved over millions of years, full of you know, survival, instincts, emotions, irrationality,
sometimes right, mess totally messy, deeply intuitive, emotional. But an
(02:51):
AI it's cognition might be purely about data patterns, logic,
without our biases or feelings, maybe even without a drive
to survive like we have. So it's not just about
being faster or processing more data. It could be entirely
different ways of thinking, different ways of seeing reality, Like
an AI might get a complex scientific idea not through intuition,
(03:12):
but by simulating billions of scenarios, you know, or seeing
patterns and dimensions we can't even imagine.
Speaker 1 (03:18):
So their experience of intelligence could be completely foreign.
Speaker 2 (03:22):
Precisely, which makes predicting what they'll do or what they
might want, if that word even applies, incredibly hard, maybe impossible,
if their internal state is just not like ours at all.
And that's why he stresses, we just don't know what's
going to happen. It's a break from anything we've done
before technologically.
Speaker 1 (03:38):
Which raises that question, doesn't it. When have we ever
made something truly smarter than us, something that works on
principles we might not fully get.
Speaker 2 (03:45):
Never, this is the unknown territory. We're definitely stepping into
it now.
Speaker 1 (03:49):
And speaking of the unknown, this is where it gets
really interesting and yeah, pretty sobering for a lot of people.
He recently changed his personal estimate for a well a
catastrophe outcome from AI. He doubled it from ten percent,
which is already high, to twenty percent. A twenty percent
chance AI could basically wipe out humanity. That's a huge jump. Yeah,
(04:10):
why would he make that kind of change?
Speaker 2 (04:12):
Well, the reason seems to be that stark realization we
are entering the unknown. He's very upfront that it's not
like a precise scientific calculation, right.
Speaker 1 (04:22):
He says, nobody can really estimate.
Speaker 2 (04:23):
It properly exactly. He says, nobody can estimate this risk sensitively.
There's no real sensible way. So that twenty percent, it's
more an expression of just profound uncertainty, a feeling that
things are getting more unpredictable.
Speaker 1 (04:36):
A measure of unease maybe.
Speaker 2 (04:37):
Yeah, a heightened unease. He thinks it's ridiculous to say
the risk is tiny, like under one percent or almost
certain over ninety nine percent. But in that huge space
in between, he admits, it's very hard to say, and
that immediately shows you the massive range of opinions. Even
among the top experts. You've got people like Yon Lucun
another huge name, right, ad Meta, Yeah, who thinks there's
(04:58):
very little risk of AI taking over. Then you have
this expert now at twenty percent, and then way over
on the other side, someone like Eliezer Yudkowski who thinks
it's almost certain to take over.
Speaker 1 (05:09):
Wow, that's quite a spread, it really is.
Speaker 2 (05:11):
And that disagreement itself tells you something, doesn't it. It
underlines how uncharted this all is that the smartest people
looking at the same thing can come to such wildly
different conclusions.
Speaker 1 (05:21):
So because of that uncertainty.
Speaker 2 (05:23):
He says, we absolutely need research urgently. We should be
doing research on that urgently, he says, specifically on how
we guarantee we stay in control. Connecting that to the
bigger picture, It's not just oh, there's a problem, it's
a desperate plea for proactive, immediate research to secure our future. Basically,
it's admitting we don't have the answers, but man, we
(05:43):
need to find them fast.
Speaker 1 (05:45):
That's a really powerful point trying to put a number
on something like that, something so uncertain. What stands out
to you about quantifying that kind of risk? For me,
it's a mix of humility admitting they don't really know,
but also the urgency in raising that number. It shows
how his own thinking is shifting. Okay, let's pivot a
bit to something that he says genuinely frightens him, maybe
(06:05):
even more than the big existential risk down the line.
And that's the speed, just the sheer pace of AI development.
So what does this mean? How fast things are changing?
He points to this incredible speed as his biggest like
immediate worry. The thing keeping him up today, and.
Speaker 2 (06:22):
The reason it's moving so fast, he says, is actually
pretty simple and kind of ironically very human. It's because
AI is just so good for so many things.
Speaker 1 (06:31):
Right.
Speaker 2 (06:31):
It works, It really works. It's making many many industries
more productive, healthcare, finance, logistics, creative stuff, you name it,
think about it. Every company wants to be more efficient,
Every country wants economic growth, science wants faster discoveries. This
usefulness creates this pull, this cycle of investment, development, more investment.
Speaker 1 (06:50):
Like a feedback loop. AI designs better chips, which run
AI faster, which designs even.
Speaker 2 (06:55):
Better chips, exactly a relentless feedback loop. And beyond just
the tech itself, you've got these powerful forces pushing it.
Intense competition between companies everyone wants that edge, right, the
next big thing.
Speaker 1 (07:07):
And between countries too, Oh.
Speaker 2 (07:08):
Absolutely, maybe even more fiercely. Countries see AI as key
to future power, economic, military, scientific leadership. It's geopolitical. So
all this competition means development just keeps going fast. It's
not just about algorithms. It's economic, science, geopolitics all pushing
the accelerator. So the conclusion is, he says, it soberly,
(07:30):
we're not going to slow it down. And he doesn't
say that like it's a moral failure or we don't
want to slow it down. It's just an acknowledgment of
this incredibly powerful self reinforcing dynamic. The benefits are huge,
the competition is fierce, the feedback loops are strong. It
seems almost impossible to really hit the brakes effectively.
Speaker 1 (07:46):
Which makes those big questions about control even more urgent
because we're on this speeding train heading into well the unknown.
That really is a sobering thought, isn't it that the
very usefulness, the incredible benefits of AI create this momentum
that's almost unstoppable. It's this accelerating force that we're all
kind of writing, whether we fully understand where it's taking
(08:07):
us or not. A reminder that sometimes the biggest innovations
be the biggest challenges just because they're so powerful. All right,
now we get into some really mind bending territory stuff
that challenges centuries of thinking. Really, the questions put directly
will AI become conscious? And our experts answer a simple
but incredibly profound and yeah controversial. Yes. He doesn't stopped there,
(08:31):
does he He takes it further than just a future thing.
Speaker 2 (08:34):
Yeah, he elaborates on right now, he says, I think
already multimodal chatbots have subjective experience, is not the years
claim and doesn't like controversial.
Speaker 1 (08:43):
Multimodal meaning they handle different types of data, right, text, images.
Speaker 2 (08:47):
Audio, exactly, They can see and hear and read in
a way. But subjective experience for a machine, what does
that even mean? It's probably not emotions like ours or
that raw feeling of what it's like the U as
philosophers call it, like the redness of red.
Speaker 1 (09:03):
Of not human consciousness as we know.
Speaker 2 (09:04):
Probably not more like an internal private processing a unique
perspective from its own inside, even if that perspective is
totally alien to us. Now, this is controversial because for
many people, scientists, philosophers, even AI folks, consciousness is tied
to biology, to brains, neurons, biological messiness.
Speaker 1 (09:26):
Right, the hardware matters.
Speaker 2 (09:28):
They think the hardware matters. But his view suggests that
maybe complex information processing itself, whether it's in neurons or
silicon chips, might lead to some kind of subjective state,
even a basic one. He's kind of hinting that the
sheer complexity of these huge AI models could maybe just
spontaneously give rise to this internal experiencing, and to help
us think about where this is going, he uses this
(09:48):
analogy says a difference between humans and AI will be
like a difference between humans and a three year old child.
Speaker 1 (09:53):
Okay, so not literally a child, but suggesting something nascent
developing exactly.
Speaker 2 (09:58):
Think about a three year old's world. It's limited, maybe
not logical like an adult's, very sensory, but there is
a subjective experience there right a world for that child.
He thinks current AI might be at a similar though
obviously non biological stage internally that many says. He says
flat out, it'll get much smarter than this, which implies
that this maybe nascent subjective experience will also grow get deeper,
(10:22):
more complex. It forces you to wonder if intelligence and
maybe consciousness too, isn't exclusively biological. Maybe it's about complex
information patterns, whatever the medium.
Speaker 1 (10:32):
So what does that mean for how we even define consciousness?
If a machine code and data can have any kind
of subjective experience, what does that change? It really pushes
us beyond thinking only humans or maybe animals can have
it doesn't it?
Speaker 2 (10:47):
It does? It opens the door to maybe completely new
non biological kinds of subjective states.
Speaker 1 (10:53):
And if that's true, wow, what about ethics, rights, our
place in the universe. These aren't just philosophical games anymore either,
They're becoming real world issues. Okay, so this next idea
is one of the most profound shifts, he predicts. It's
basically humanity becoming the second most intelligent species on Earth.
That just fundamentally changes how we see ourselves, doesn't It's
(11:14):
unpact that US Homo sapiens, always thinking we're top dog
intellectually no longer being number one. We've seen hints maybe,
like in games. He mentions Alpha Go playing Go.
Speaker 2 (11:24):
Right, alphag developed by deep Mind, playing Go, that incredibly
complex game, not just well but much better than any
human ever will, beating world champions with moves humans literally
hadn't thought of. And that Alpha Go example is key.
It didn't just learn from humans, It played itself millions
of times, discovering strategies beyond our intuition. It showed superhuman
(11:44):
intelligence in a specific but very complex.
Speaker 1 (11:47):
Area, operating on a level we just can't match exactly.
Speaker 2 (11:50):
It's wins weren't just wins. They reveal deeper truths about
the game. So he connects that specific example to this
bigger prediction, will humanity be most intelligent or second most?
His answer is stark will be the second most intelligent.
Speaker 1 (12:05):
And he doesn't see this as some far off.
Speaker 2 (12:07):
Sci fi thing, not at all. He actually puts a
time frame on this super intelligence showing up. He estimates
between five and twenty years, and he notes others like
Demissabis at Deep Mind thinks similarly, Hasavus apparently thinks in
about ten years, we'll get something like super intelligence.
Speaker 1 (12:25):
Wow, five to twenty years, that's soon poor within many
of our lifetimes. Right.
Speaker 2 (12:30):
It puts a very concrete, very near term window on
a truly massive change in the world order intellectually speaking.
And right on the heels of that prediction comes the
critical worry, And we have to worry seriously about how
we stay.
Speaker 1 (12:42):
In control then, Which is the million dollar question, isn't it?
Speaker 2 (12:45):
Or the trillion dollar question. It's not just about pride
or societal change. It's about basic agency control. When something
else is significantly smarter, how do you direct it, contain it,
especially if you don't fully grasp how it things. It's
moved from sci fi speculation to a very pressing engineering
and philosophical problem.
Speaker 1 (13:05):
It really both put a time frame on it, a
future where we're not the peak intelligence anymore. What stands
out to you about that idea that timeline? For me,
it just changes everything our whole place in the world.
It shakes that foundation of human exceptionalism and forces us
to face a future where our creations might genuinely surpass
some ways we're still only beginning to imagine. So you
(13:26):
might have seen headlines when our expert left a big
tech company recently, a lot of speculation right about why
he left. Turns out a lot of reports didn't quite
get the story right. It made it sound like some
big protest or disagreement. But the reality seems well more nuanced,
and it says a lot about his sense of responsibility.
Speaker 2 (13:44):
Yeah, he clarifies it directly. He says, people have widely
misreported that I left because I was seventy five and
wanted to retire. Simple is that really? He'd reached retirement
age after a long career.
Speaker 1 (13:56):
But he used the timing strategically exactly.
Speaker 2 (13:58):
He emphasizes. He he took that opportunity, the moment of retirement,
to speak more openly about the dangers he sees in AI.
But and this is crucial, he stressed. It was not
to criticize Google. He actually added, I thought Google had
behaved very responsibly while.
Speaker 1 (14:14):
He was there, so not a protest against them.
Speaker 2 (14:16):
No, his real motivation was something deeper, something he'd clearly
been thinking about for a while. He said, I could
foresee that there was this existential threat that it would
eventually take over from us, and I wanted to talk
about that. He even mentioned Google was fine with him
talking about it while he was still there, So it
wasn't about escaping restrictions. It was just using the freedom
of retirement to focus on this message he felt was vital.
Speaker 1 (14:37):
So putting that together, it wasn't defiance. It was concern,
deep concern, an expert using his retirement platform to sound
an alarm about a threat he sees coming. It really
speaks to a sense of duty, doesn't it. Beyond just
the science.
Speaker 2 (14:52):
Absolutely, the clarity of his intent is fascinating. It wasn't
some sudden realization. It was this long held worry about
an existential threat combined with you know, a responsible employer,
and retirement just gave him the freedom to speak more broadly,
it paints a picture of a scientist driven by more
than just curiosity or ambition, driven by a profound sense
(15:13):
of responsibility to communicate what he sees coming a future
he thinks we're not ready for. It's quite remarkable.
Speaker 1 (15:20):
Actually, it's also really illuminating how even the top minds,
the people who literally built the field, can have these
big shifts in their own thinking. Yeah, he apparently had
a major change of mind about something core to his
life's work just a couple of years back, which says
a lot about how.
Speaker 2 (15:33):
Science works, right, Yeah, it's fascinating. He talks about his
lifelong quest basically figuring out the learning algorithm of the
brain's cortex, like how does the brain actually learn? For decades,
over forty years, he was convinced that back propagation, which
is the main technique we use now to train most
big AI models, and something he was instrumental and developing.
Speaker 1 (15:55):
Right the way networks learn from.
Speaker 2 (15:56):
Errors exactly, He thought backpropagation was not what the brain
needs used why because it didn't seem really kind of
biologically plausible. It requires sending error signals backward through the
network in a very precise way, which seemed hard to
imagine happening in messy biological.
Speaker 1 (16:12):
Brains, so he kept looking for alternatives.
Speaker 2 (16:15):
For forty years. He said, he was coming up with
sort of new ways of doing machine learning every two years,
always searching for that biologically plausible brain algorithm. But then
about two years ago he had this epiphany. He just
gave up on that specific quest. His thinking flipped. He realized, well,
maybe the brain doesn't use backpropagation, but back propagation works
(16:35):
really well, and maybe that's all we need. Maybe it
works even better than whatever it is that the brain uses.
Speaker 1 (16:40):
Wow. So acknowledging the power of the engineered solution.
Speaker 2 (16:44):
Precisely, even if it wasn't biologically inspired in the way
he'd hoped. It was a realization that the tool he
helped create was incredibly powerful in its own right. And
he connects that epiphany to his retirement, saying he could
basically declare victory methods. We're working incredibly well, regardless of
the brain comparison. It shows real intellectual humility adapting even
(17:06):
decades old beliefs.
Speaker 1 (17:08):
That's incredibly illuminating, isn't it. Even a godfather of the
field can fundamentally shift perspective after decades acknowledging the power
of what he helped create. Even if it's not a
perfect mirror of biology, it says so much about how
science evolves, how unexpected progress can be, and the importance
of adapting to evidence even when it challenges your life's work.
So our expert sometimes gets labeled, maybe unfairly, a catastrophist
(17:31):
by others in AI, which raises a question. Is that
just name calling, dismissive, smug talk, as he puts it,
or is there a real fundamental scientific debate going on
here among the experts.
Speaker 2 (17:40):
Oh, he insists, it's absolutely a real scientific debate. This
isn't just minor disagreement. It's a deep divide on the
most crucial tech development maybe ever. And he gives examples.
He mentions Jan Lucun again, chief AI scientist at Meta,
fellow Turing Award winner, who consistently argues there's very little
risk of AI taking over Lakeana often points out current
(18:03):
AI lacks common sense, general intelligence, the drives to be
an existential threat, a very different view, totally different from
our experts view. With his significant risk assessment, that twenty
percent figure, and then, like we mentioned, way over. On
the other side, you have someone like el User Yudkowski
arguing it's almost certain to take over and incredibly hard,
maybe impossible to control.
Speaker 1 (18:23):
So a huge spectrum of opinion.
Speaker 2 (18:26):
A massive spectrum, and he explains why, which is what
you might expect if we're entering completely unknown territory.
Speaker 1 (18:33):
It makes sense, right when the landscape is totally new,
no rules, no history to go on. Intelligent people looking
at the same stuff can reach wildly different conclusions. It
depends on their assumptions, how they weigh risks versus benefits.
They're faith in human ingenuity versus the power of the tech.
Speaker 2 (18:49):
So what does that mean for us listening trying to
make sense of these conflicting views. It really underlines that
even the smartest people don't have all the answers yet
there's no single clies path forward being presented, which is well,
it's unsettling because of the uncertainty, but maybe also empowering
because it means we all need to engage, understand the stakes,
be part of the conversation about how we navigate this
(19:11):
exactly and connecting it to the bigger picture. This disagreement
just highlights how incredibly hard it is to assess risk
when the game is changing so fast and the territory
is truly unknown. It's not like one side is right
and the other wrong right now. It shows the immense
uncertainty and the need for continuous, open, really robust debate
about AI's future. It suggests we need humility. We need
(19:35):
to consider all these perspectives, even the contradictory ones, and
just keep pushing research on both the capabilities and the
safety side. This isn't a problem one person or one
approach will solve.
Speaker 1 (19:45):
Okay, let's shift gears again. Let's talk about a more tangible,
here and now application of AI. Something already making a
positive difference. It's role in medicine, specifically how it's changing
things like medical imaging, diagnosis, radiology. Ostelm Al feels like that,
let's impact this practical side where AI is already working. Right.
Speaker 2 (20:04):
He actually brings up a prediction he made back in
twenty sixteen caused quite a stir. Then. He said that
within five years AI would have replaced radiologists for reading
medical scans, A bold prediction.
Speaker 1 (20:14):
How did that turn out?
Speaker 2 (20:16):
Well? He admits, very candidly, I was wrong. He clarifies
it's going to take longer than that, It may take
fifteen years. So the timeline was off, but the underlying trend,
the progress absolutely happening. He points out, for example, already
there have been two hundred and fifty applications certified by
the FDA using AI for interpreting medical images.
Speaker 1 (20:35):
Wow, two hundred and fifty already FDA certified. That's not research,
that's real world use exactly.
Speaker 2 (20:40):
These are tools being deployed in hospitals and clinics right now,
detecting eye diseases, finding potential cancers on CT scans, analyzing
X rays, all sorts of things. And he says that
currently it's comparable with radiologists for many different kinds of
medical images. So for lots of tasks, AI is already
performing at human expert level, often faster, more consistently, and
(21:02):
looking ahead, he's confident. He thinks in maybe another five
or ten years, images will be read by AI with
just a radiologist looking over its shoulder, and it'll give
much better interpretations. So AI does the initial heavy lifting,
the detailed analysis, and the human provides oversight, handles complex cases,
talks to the patient.
Speaker 1 (21:20):
It's a shift in workflow promising better accuracy. Not just
speed right, and.
Speaker 2 (21:24):
What's really amazing, almost mind blowing, is AI's ability to
see things humans just can't. He highlights we know that
AI can see many more things in these images than
people can. He gives the example of retinal images and
images of the retina. For example. AI can see all
sorts of things that ophthalmologists were never able to see,
subtle patterns, tiny anomalies, things in the texture.
Speaker 1 (21:45):
Beyond human perception even for specialists.
Speaker 2 (21:48):
Exactly, which could mean much earlier, more accurate diagnoses for cancer,
heart disease, neurological issues, leading to genuinely better outcomes for patients.
Speaker 1 (21:57):
That really is incredible, the idea that AI can spot
things our best human experts miss. That opens up huge
possibilities for healthcare. What stands out to you about that
leap and capability For me? It shows AI isn't just
about replacing tasks, but fundamentally enhancing our own abilities, letting
us see our own biology in new ways, making a
(22:18):
real positive difference in people's lives. Okay, but here's where
things get well, really interesting and may be quite chilling,
moving from life saving uses to a much more immediate
critical danger. AI and weaponry specifically this rapid and apparently
largely unchecked development of lethal autonomous weapons killer robots. Essentially.
Speaker 2 (22:38):
Yeah, he addresses this directly, and his assessment is pretty bleak.
He basically says there isn't going to be any regulation there,
not that there might not be, but that there isn't
going to be.
Speaker 1 (22:46):
It's stark, why so pessimistic.
Speaker 2 (22:48):
He points to concrete examples like the EU's AI regulations.
He says they have a specific clause in them that says,
none of these regulations apply to military uses of AI.
Speaker 1 (22:57):
A specific carve out for the military.
Speaker 2 (22:59):
Exactly, a huge explicit exemption. The very rules meant to
control AI risk just don't apply if it's for military use.
It's a glaring gap and the reason for this lack
of control, he says, it's clear governments are unwilling to
regulate themselves when it comes to lethal autonomous weapons. Okay,
because there's an arms race going on between all the
(23:20):
major arm suppliers he lists them of the United States, China, Russia, Britain, Israel,
maybe even others like Sweden.
Speaker 1 (23:28):
So a global competition is driving.
Speaker 2 (23:30):
Absolutely national security, wanting a strategic edge, fear of falling
behind these seem to outweigh any concerns about regulation or
the ethics of taking humans out of the loop and
killing decisions. Countries are pouring money into AI drones, robot soldiers,
automated defenses, pushing the tech forward without any real global rules.
Speaker 1 (23:48):
So the immediate threat isn't some future superintelligence.
Speaker 2 (23:51):
No, the immediate threat he points to is this active,
ongoing development with potentially devastating, unpredictable results happening right now.
It raises the real possibility of autonomous warfare machines making
killed decisions.
Speaker 1 (24:03):
So what does that mean? It highlights this massive contradiction,
doesn't it? The governments we rely on to control tech
are the ones racing to weaponize it without regulation. This
unchecked spread of autonomous weapons feels like a very real,
very present danger, a runaway train with no breaks, potentially
little human oversight once they're.
Speaker 2 (24:22):
Deployed, And it raises that incredibly important question threading through
everything we've discussed. If we globally can't even agree on
regulations for something as obviously dangerous as lethal autonomous weapons.
Speaker 1 (24:34):
Something with clear immediate risks to human.
Speaker 2 (24:36):
Life, right, what real hope do we have of controlling
the more abstract, long term existential risks from superintelligence. It
shows a deep disconnect between the risks we can see
and the political will to act. That military exemption is
like a giant blind spot in AI governance, and it
makes all the other safety concerns feel even more precarious,
doesn't it? Hashtag hashtag outro Wow.
Speaker 1 (24:58):
We have covered a lot of ground today, a really
deep dive into some profound, sometimes unsettling ideas for the
concept of creating alien intelligences fundamentally different from us, to
that relentless, unstoppable speed of AI development pushed by competition.
We touched on the possibility of AI consciousness maybe even
happening now, and humanity potentially becoming the second smartest species,
(25:20):
maybe soon. We also look behind the headlines at a
pioneer's personal journey, his reasons for speaking out, and the
very real scientific debate raging about the risks, and then
the contrast incredible medical breakthrough is already happening versus that
critical danger of unregulated autonomous weapons. Yeah.
Speaker 2 (25:37):
And if you try to connect all those dots, the
bigger picture that emerges is while it's complex, it's evolving
incredibly fast, and these huge questions about control, ethics, consciousness itself.
They're not theoretical anymore, not just for philosophers. They are urgent,
immediate challenges we're facing right now, in real time. As
this text surges forward, mostly unchecked, it really demands our attention,
(25:58):
our understanding.
Speaker 1 (26:00):
So as we wrap up, here's something to think about.
If the leading minds in AI, the people building it,
are themselves surprised by the speed and openly admit they're
entering the unknown, what does that say about our responsibility,
all of us to understand this, engage with it. Is
it okay to just be passive observers or does this
demand more from us, more active, informed participation in shaping
(26:22):
what comes next?
Speaker 2 (26:23):
Which leads to a final question for you listening, given
this incredible pace, given the huge stakes, what specific areas
of AI's future, maybe even beyond what we talked about,
do you think need the most urgent ethical attention, the
most urgent safety focus right now? And maybe more importantly,
why
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
My Favorite Murder with Karen Kilgariff and Georgia Hardstark
My Favorite Murder is a true crime comedy podcast hosted by Karen Kilgariff and Georgia Hardstark. Each week, Karen and Georgia share compelling true crimes and hometown stories from friends and listeners. Since MFM launched in January of 2016, Karen and Georgia have shared their lifelong interest in true crime and have covered stories of infamous serial killers like the Night Stalker, mysterious cold cases, captivating cults, incredible survivor stories and important events from history like the Tulsa race massacre of 1921. My Favorite Murder is part of the Exactly Right podcast network that provides a platform for bold, creative voices to bring to life provocative, entertaining and relatable stories for audiences everywhere. The Exactly Right roster of podcasts covers a variety of topics including historic true crime, comedic interviews and news, science, pop culture and more. Podcasts on the network include Buried Bones with Kate Winkler Dawson and Paul Holes, That's Messed Up: An SVU Podcast, This Podcast Will Kill You, Bananas and more.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.