Video summary
Dava Newman addresses the philosophical implications of artificial intelligence in space exploration, specifically referencing the famous scenario from *2001: A Space Odyssey* where HAL 9000 prioritizes mission objectives over astronaut safety. She argues that while this is a compelling ethical question regarding whether an AI made the right choice, future Mars missions will necessitate fully autonomous systems because real-time communication with Earth is impossible due to significant time delays of up to twenty minutes. Consequently, both human crews and robotic systems must operate independently without reliance on Mission Control on Earth, making autonomy a critical design challenge that requires rigorous testing before deployment in deep space environments where immediate intervention is not an option. The necessity for robust autonomy stems from the vast distances involved; whereas lunar missions can be tested with only three-second communication delays allowing for some level of oversight, Mars presents a scenario where systems must function entirely on their own once launched. Newman points to historical examples like Magellan and his crew, who departed Earth fully autonomous and had to rely on in-situ resource utilization upon landing, illustrating the mindset required for future exploration. These missions serve as proving grounds where technologies are validated under conditions of limited connectivity, ensuring that when humanity eventually reaches Mars, the infrastructure is capable of sustaining life and operations without constant terrestrial support. Regarding the extent of AI involvement in decision-making processes, Newman clarifies that humans will remain "in the loop" during system design, though future systems may evolve to handle complex mission decisions independently once deployed on other planets. Currently, however, space-based AI capabilities are heavily focused on Earth-centric applications rather than immediate Mars exploration needs. The technology is being trained using petabytes of data gathered from Earth to address urgent global issues such as climate change and environmental monitoring. By analyzing satellite imagery across ocean systems, land, and air, these advanced algorithms aim to provide actionable insights that can influence human behavior and help society live in balance with the planet. Ultimately, Newman envisions a future where AI acts as powerful tools for curating vast amounts of space data into accessible visualizations that empower individuals to understand Earth's changing climate systems. She emphasizes that before expanding our reach to Mars, there is an immediate imperative to care for "Spaceship Earth" by leveraging current supercomputing resources and artificial intelligence research. The goal is not just technological advancement but the acceleration of positive change through better data analysis and education regarding environmental challenges. By holding future satellite imagery in one's hands via these AI-driven tools, society can make informed decisions that foster sustainability and resilience on our home planet before turning its gaze to the stars.
Read the full video transcript
I love artificial intelligence and you
thought I've also saw that you've
enjoyed the movie space obviously 2001 a
Space Odyssey let me ask the question
about Hal 9000 that makes a few
decisions there that prioritizes the
mission over the the astronauts do you
think from a high philosophical question
do you think hell did the right thing
prioritizing the mission
I think our artificial intelligence will
be smarter in the future for a Mars
mission it's a great question that is
the the reality isn't for a Mars mission
you know we need fully autonomous
systems we will get humans but they have
to be fully autonomous and that's a
really important that the most important
concept because you know there's not
going to be a Mission Control on earth
you know I'd you know 20-minute time leg
there's just no way you're gonna control
so fully a ton so people have to be
fully autonomous as well but all of our
systems as well and so that's that's the
big design challenge so that's why we
test them out on the moon as well when
we have a okay a few seconds you know a
three-second time leg you can test him
out we have to really get autonomous
exploration down you asked me earlier
about Magellan and Magellan and his crew
they they left right they were
autonomous mm-hmm you know they were
autonomous they left and they were on
their own to figure out that mission
then when they hit land they have
resources as the in situ resource
utilization and everything else they
brought with them so we have to I think
have that mindset for expression again
back to the moon it's more the testing
ground the proving ground with
technologies but when we get to Mars
it's so far away that we need fully
autonomous systems so I think that's
that's where again AI and autonomy come
in really robust autonomy things that we
don't have today yet so they're on the
drawing boards but we really need to
test them out because that's that's what
we're up against
so fully autonomous meaning like
self-sufficient there's still a role for
the human in that picture do you think
there will be a time when AI systems
just beyond doing fully autonomous
flight control will also help or even
take mission decisions like how did
that's interesting it depends I mean
they're gonna be designed by humans as
you mentioned humans are always in the
loop I mean we might be on earth we
might be in orbit on Mars maybe the
systems the Landers down on the surface
of Mars but I think we're gonna get we
are right now just on earth-based
systems you know
I systems that are incredibly capable
and you know training them with all the
data that we have now you know petabytes
of data from earth what I care about for
the autonomy and AI right now how we're
applying it and research is to look at
earth and look at climate systems I mean
that's the it's not for Mars to me today
right now AI is two eyes on earth all of
our space data compiling that using
supercomputers because we have so much
information and knowledge and we need to
get that into people's hands we need
first there's the educational issue with
climate and our changing climate then we
need to change human behavior that's the
biggie so this next decade it's urgent
we take care of our own spaceship which
is spaceship earth so that's to me where
my focus has been for AI systems using
whatever is out there kind of imagining
also what the future situation is a
satellite imagery of Earth of the future
if you can hold that in your hands
that's gonna be really powerful will
that help people accelerate positive
change for Earth and for us to live in
balance with earth I hope so and kind of
start with the ocean systems so oceans
to land to air and kind of using all the
space data so it's a huge role for
artificial intelligence to help us
analyze I call it curating the data
using the data it has a lot to of
visualizations as well
you