June 8, 2026 • 21 mins
In this episode of "Full Tech Ahead," host Amanda Razani interviews Dr. Jason Corso, Toyota Professor of AI at the University of Michigan and Co-Founder of Voxel51. They discuss Voxel51’s role as a developer tool software company for physical and visual AI, which has achieved over 4 million downloads.
The core of the conversation focuses on Vigil, an innovative healthcare AI project led by Dr. Corso and funded by ARPA-H’s Paradigm program. Vigil tackles the critical shortage of specialists and brick-and-mortar hospitals in rural America by equipping mobile medical units (clinics on wheels) with physically grounded AI.
Instead of replacing clinicians, Vigil acts as an advanced co-pilot, using computer vision and on-the-fly micro-guidance to upskill generalist healthcare workers (like registered nurses or EMTs) to perform complex procedures, such as cardiac ultrasound diagnostics, directly in remote communities.
Key Quotes
- "Voxel51 is indeed a dev tool software company for AI that supports the developer... in the spaces of physical AI and visual AI."
- "I don't think AI is here to replace humans... I just believe that we are as technologists in AI, we are building tools that will augment humans."
- "We have this notion of a triangle of trust where the healthcare worker is trusting Vigil to help him or her, and the patient is trusting the healthcare worker, and then tacitly, the patient is trusting Vigil."
- "In the healthcare, in the visual domain, we can't hallucinate, first of all... We're really trying to get toward those guaranteeable guardrails."
Takeaways
- Upskilling the Generalist Workforce: AI can dramatically expand healthcare access without needing to "clone specialists." By equipping existing local nurses or EMTs with AI-guided tools, they can perform specialized tasks—like capturing precise cardiac ultrasound imagery—that normally require years of dedicated training.
- The "Triangle of Trust": Successful AI deployment in healthcare relies heavily on the bedside manner and human connection. The patient trusts the clinician, the clinician trusts the AI, and the patient tacitly trusts the AI. Maintaining this human-centered relationship is crucial.
- Guaranteeable Model Guardrails: Unlike conversational LLMs that are prone to hallucination and rely on post-hoc prompt filters, critical visual AI systems in healthcare require deeply grounded, mathematical, and theoretical guardrails that prevent errors before they happen to ensure patient safety.
- Augmentation over Replacement: The future of advanced technology, including robotics (like actuated robotic arms in mobile clinics), is to augment human capabilities. AI provides an extra set of un-blinded eyes and precise micron-level assistance, allowing human workers to perform their jobs faster, better, and more equitably.
Find Amanda Razani on LinkedIn. https://www.linkedin.com/in/amanda-razani-990a7233/
Follow the FTA LinkedIn Page: https://www.linkedin.com/company/full-tech-ahead/
Visit the FTA website: https://fulltechahead.com/
Check out the Substack Channel: https://fulltechahead.substack.com/
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_00 (00:19):
Hello and welcome to Full Tech Ahead.
I'm Amanda Rizzani, and with metoday I'm excited to have Jason
Corso.
He is a Toyota professor of AIat the University of Michigan
and co-founder of Voxel 51, acompany focused on visual AI.
He's also leading an exciting AIproject with ARPA H called
(00:42):
Vigil, which is exploring how AIcan help extend healthcare
access in remote communities bysupporting clinicians rather
than replacing them.
So, Jason, welcome to the show.
Can you share a little bit aboutVoxel 51, first of all?
SPEAKER_01 (00:57):
Absolutely.
Thanks, Amanda.
It's great to be here with you.
So Voxel 51 is indeed a dev toolsoftware company for AI that
supports the developer, like themachine learning scientist,
machine learning engineer, dataannotator, data curator in the
spaces of physical AI and visualAI.
Specifically, we make a softwaretool called 51, which supports
(01:21):
the full stack of the workflowsfrom managing, curating,
annotating data throughtraining, evaluating, and
deploying the models that aretrained from that data.
We work in various verticalslike automotive, healthcare,
physical security, productsupport, generally are supported
by a uh we're about 60 peoplenow, actually, which is great
(01:43):
for me.
You know, it's it's been massivegrowth in the last five, five or
so years.
Uh we have about four milliondownloads of the open source
software.
This is like a single userinstance of 51.
And we uh support both on-premand air gap solutions uh when
data security is reallyimportant to our customers.
SPEAKER_00 (02:01):
Awesome.
Well, I want to talk more aboutthis vigil.
What problem and role healthcarewere you trying to solve when
you created this?
SPEAKER_01 (02:11):
Awesome.
Yeah.
So right at the University ofMichigan, I have I had been
getting more into healthcare uhin the last three or four years
and was delighted when thisprogram from ARPH, the Paradigm
Program, came online.
So um not being a physician, youknow, this is I'm an engineer,
right, by training.
So having to learn aboutphysical physician or
medical-oriented problems,healthcare problems in the US
(02:33):
really is enlightening to me.
Um so Vigil really tackles a uma key problem in rural
healthcare in the U.S.
So generally, there's a lack ofspecialist, a lack of
expert-trained physicians, soMDs or DOs that operate in rural
America.
And to make that problem evenworse, there is an increasing
(02:56):
shortage of physical brick andmortar locations where rural
Americans can go nearby theirhomes to get healthcare, i.e.,
hospitals are closing, even justthe clinics are closing.
There's one study I'm aware ofthat was based in North Carolina
that measured the distance aNorth Carolina citizen had to
travel, the average distance, toget to their nearest brick and
(03:18):
mortar healthcare facilitybetween the years of 2012 and
2018 increased eightfold.
So that means someone wasdriving 15 minutes to get to
their nearest hospital.
Now they may have to drive twohours to get, or more, frankly,
because this is an average.
Um, so so ultimately, vigil uhis there to well, obviously, we
can't just, you know, clonephysicians, we can't build brick
(03:40):
and mortar hospitals, right?
That would have been done byother institutions already.
So what we are trying to do is,well, first of all, we make the
observation that although thereare few specialists, few
expert-trained MDs and DOs,there are actually quite a quite
a large number of registerednurses or physic physicians
assistants or even EMTs that arein those regions.
(04:02):
And although they may not bespecialists in any one
particular area, they aregenerally trained with medical
knowledge.
And so the idea is if we canequip them with kind of like a
hospital on wheels or like aclinic on wheels, um, and some
physically grounded AI, right?
So kind of like a Hay Sirimindset, right?
But uh but a Hay Siri, which weactually call Hay Vigil, right,
(04:25):
with with a vigil that actuallyhas cameras and eyeballs and
ears inside of this mobileclinic, so that when the
generalist is doing their workwith the patient, uh they're
actually able to be guided onthe fly by this AI system that
we call vigil.
Um clearly the this thisgenerally trained or a
generalist um healthcare workerhas the final say, right?
(04:48):
If they're ever not confidentwith what AI guidance they're
getting, they can always phonehome, you know, phone back to
the hospital, may elongate thevisit.
However, the idea is that we'llbe able to deliver significantly
better, uh richer, and morerobust healthcare in rural
America by building these mobilevans, staffing them with one or
two of these generalisthealthcare workers, and then
(05:11):
giving them a road, you know,affiliating with a with a rural
hospital, obviously, right?
Because we have to have, youknow, we're not gonna have
patients of our own, butaffiliating with a rural
hospital, and then they'll havea day in which they can treat
maybe 12, 16, 20 differentpatients who otherwise would
have had to take days off, fulldays off of work or what have
you.
SPEAKER_00 (05:29):
That is fantastic.
Well, I can say definitely, Ilive in the heart of Texas, and
there's a lot of ruralcommunities all over that are
all reliant on a hospital, likeyou said, uh hour, two hours
away.
So that is very helpful.
So why did you take an approachthat helps clinicians instead of
(05:50):
trying to just replace them withAI?
SPEAKER_01 (05:53):
Well, I think this goes back to my general
philosophy.
Um, you know, I I've been in thefield for about 25 years now.
I just I just believe that weare, as technologists in AI, we
are building tools that willaugment humans, right?
Humans are the center of ourexperience.
We are humans, right?
In some sense.
And I have this saying now thatI've been working with a
colleague of mine, for allhumans, for all life, right?
(06:14):
Like humans are the center ofour world.
And I don't think AI is here toreplace humans.
We probably could have anotherconversation about whether or
not like white-collar jobs aregoing to be lost to AIs.
I just don't believe so.
Frankly, I think it's gonna be aboon for white-collar jobs
because they'll have bettertooling around what can be done.
Uh, in in this particularhealthcare space, specifically,
(06:35):
uh, you know, we can't, we couldnot go out and do deliver vigil
in practice today because ofvarious state-level um like
laws, essentially, likeliability laws and you know,
very various safety laws thatare that are put in place for
great reasons to protectpatients.
Um and so we we just franklybelieve the right way today,
tomorrow, in the next five yearsto deliver this increased level
(06:56):
of rural healthcare is to justaugment the humans that are uh
in in the vehicle that will bein the vehicle to do the work
that they want to do, that theyknow how to do.
You know, it's one key reasonthere is that, right?
Like just tooling for thesehuman AI teams.
That's been the heart of mycareer in some sense.
Um, you know, another aspectthough is that, especially in
healthcare, there is asignificant value to the bond,
(07:19):
that the trust and the bondbetween the healthcare worker
and the patient.
And vigil is there not to do inany way uh violate or corrupt
that bond.
In fact, we we we have thisnotion of a triangle of trust
where the vigil worker, right,the healthcare worker is
trusting vigil to help him orher, and the patient is trusting
(07:40):
the healthcare worker, and thenkind of tacitly the patient is
hence trusting vigil.
And if that triangle doesn't getbuilt and get supported, then
vigil will not be a success.
SPEAKER_00 (07:49):
Yeah, I agree that personal connection and that
bedside manner are so critical.
I have actually been inpositions on a few occasions.
Um, my mom, for example, uhlives in a rural town where um
there's not as many eye doctorsavailable, and they roll up a
screen.
Um uh, and even that is a realperson on the screen, but it's
(08:13):
still a little odd when you'rejust talking to a screen, you
know, instead of going to, youknow, you go to the eye clinic
to see the eye doctor, but youjust get a screen rolled up.
So yeah, I understand.
SPEAKER_01 (08:27):
Uh I understand.
In fact, in fact, um I I believethere are on the order of 30,000
plus mobile health units inpractice today.
So your your kids are right.
So she's going somewhere andthen they're rolling up a screen
where where someone's beingvideo telecomed in.
But the notion of a mobileclinic is not novel to Vigil or
(08:48):
to this ARPA paradigm program.
I think what's what's actuallywhat's really novel, I believe,
is that there will be a humanoperating in the vehicle and
there'll be upskilled, right?
This generalist learning how todo things on the fly or not even
just learning, but like beingbeing reminded how to do things,
right?
So one for example, one one oneuh clinical service we we're
(09:09):
we're studying right now iscardiac diagnostic health
through ultrasound.
You know, typically anultrasound operator need must
take, I believe it's about twoto two years, maybe three years
of training to actually like beuh credentials in to perform an
ultrasound for cardiac health.
You know, it's it's great.
And I'm very grateful that wehave humans who go and do that
(09:31):
work.
And still we can't, but there'snot enough of them, frankly, not
only in rural places, but evenin in like urban, semi-urban and
urban places.
So um, in this case, instead,someone, you know, an RN who
might have touched an ultrasoundthree times can be upscaled on
the fly to go and do the samelevel of care, to get the same
quality images that either thecomputer will read automatically
(09:52):
or will just be stored, and thena radiologist would read post
hoc, um, which is not thatdifferent than going to a brick
and mortar uh facility.
You know, the the programmanager, uh Dr.
Bon Ku from RPAH, uh, theprogram manager of Paradigm, he
was visiting campus last weekactually for a site visit.
And uh in in his morningpresentation, uh made the made
(10:14):
the comment that he believes onthe order of 80% of all medical
care does not need to be done ina brick and mortar facility.
It could be done in thepatient's home or in in the
Walmart parking lot or in one ofthese mobile clinics.
And I think realizing thathealthcare of the future, or
like laying the ground, thestonework, the groundwork for
the healthcare of the future,you know, it's just it's frankly
(10:34):
exciting to be a part of that.
SPEAKER_00 (10:36):
Yeah, it is.
I mean, you think back to whenyou had those home office visits
from your doctor.
I mean, if you're really sickand ill, trying to get out and
go somewhere, it's difficult.
Not to mention the comfort ofbeing in your home.
So there's a lot of positivesfor sure.
Well, and as you said, kind ofbreaking down, you know, you
we're lacking in thesehealthcare areas.
(10:59):
So kind of breaking down thetime barriers, uh, making it
easier to get people out thereto help.
SPEAKER_01 (11:05):
Absolutely.
Um, indeed, I I remember, infact, when my spouse was about
to deliver our first child, thechiropractor came for a home
visit.
Uh, and that was uh that was abig help, in fact.
Yeah.
But these home visits can itjust changes the dynamic,
really, right?
So not not that there's uh well,I mean, the thing is it could
(11:25):
not not like levels of playingfield, but uh really just the
the notion that all humans arevalued and all humans are equal.
I I think it it becomes morerealizable when when we have
when we have the ability todeliver better better health
care.
SPEAKER_00 (11:39):
Absolutely.
Well, you know, AI can sometimesbe a hot button topic.
So when it comes to AI andhealthcare, what are some things
that people get wrong or somemisconceptions?
SPEAKER_01 (11:51):
Oh, it's a good it's a great question, right?
So um, well, first of all,indeed, there is a there's a
huge marketing machine behind AIthat broadly, you know, both in
healthcare and outside ofhealthcare, that unfortunately I
think is doing a little bit of adisservice to actual AI
technologists.
Uh, you know, in the sense thatuh there are increasing claims
(12:13):
about when AI will solve problemX or we'll take job Y, or we'll
have self-driving cars, youknow, 10 years ago, and we still
kind of don't have them, right?
So like there's just thischallenge where the message and
the hype around AI is notactually backed out by robust
engineering-based testing anddelivery.
Um, you know, in in thehealthcare domain, you know,
(12:36):
vigil is is um it treats theproblem a little differently
than we might think.
Um, so the the idea in Vigil isto first of all do no harm, kind
of like a you know traditionalhealthcare mindset, but that has
different layers of impact,right?
So some generalists maybethey're actually rather well
trained in in ultrasound, justas an example, right?
(12:58):
And some generalists are nottrained in ultrasound.
So if vigil gave the same levelof guidance to both of those
types of individuals, it wouldbe a problem, right?
They'd be distracting the moreexperienced individual.
And so what when we are modelingguidance, uh we're we're
actually using very contemporarystate-of-the-art artificial
intelligence methods.
You know, we have our ownhealthcare world model that sits
(13:21):
behind the technology, and thenan agent that is modeling not
only what is happening in theenvironment, but also what
should be happening in theenvironment in the context of
the knowledge about the skilllevel of the uh operator, of the
generalist.
And so if that agent decides,like that in some sense, the
goal of that agent is to decidewhether or not to provide
(13:44):
guidance or just let what isgood good, right?
Like, you know, we have thesemultiple screens happening in
the background that just kind ofwalk the generalist through the
steps of the procedure that theyshould that they should be
doing.
Those the generalists can lookat or not, right?
Those are not intrusive.
But at some point, for example,when you're holding an
ultrasound probe near the heart,and you need to get like what's
(14:04):
called an apical forechamberview to get an ejection
fracture, how healthy, how muchblood is getting pumped out of
the left ventricle.
Sometimes you have to make thesemicro adjustments, you know, it
a fraction of a degree.
And you really just build thissense over time.
Um, so Vigil can opt to givedifferent levels of training or
guidance at that point.
Uh, you know, it can render, forexample, through a projector,
(14:27):
yeah.
We actually create this kind ofshared perceptual space through
a projecting the interface ontothe patient's body.
And so it can render differentangles, different arrows.
Uh, it renders like a ghostprobe.
So you can, you know, map thereal probe to the ghost probe
just to really have differentways of interacting with the
patient.
And I think that that type of uhmodulated interaction through AI
(14:49):
uh is is really is franklyreally novel uh because we're
not you know, you as youprobably have heard, like these
general LLM type models, theythey're prone to hallucinate.
And in and right, and so and andif you tell them they're
hallucinating, they'll quicklybelieve you, right?
There's actually not much deepmodeling about the veracity of
(15:09):
the conversation.
Whereas in the healthcare, inthis in the vigil domain, we
can't hallucinate, first of all.
So we really have to be groundedinto what's allowable.
Um, but we also have to modelthe fallibility of the
generalist, the the likelihoodthat they may or may not make a
mistake.
Uh, and if they are making amistake, we need to both
(15:30):
anticipate it and then preventit, especially if there's any
any potential harm or whatever.
So modeling that like safety inthe interaction is super
critical.
Uh, and it's not something we'vereally seen enough of yet in the
general AI space.
I think that's one thing we'recollectively getting wrong,
right?
Like um, we we don't really haveappropriate uh in-model
(15:51):
guardrails.
From my understanding, most ofthe guardrails that are in the
systems that we are all many ofus are using on a regular basis,
are post or pre-hoc or post hoc,right?
They're they're guardrails onwhat could be set in the prompt
or what could be outputted bythe model to the to the user.
Um, and they're not sort ofguaranteeable in any notion of
like theoretical or mathematicalsense.
(16:13):
And we are in vigil, we'rereally trying to get toward
those guaranteeable guardrailsso that um uh we you know we we
we can really deploy this forfor the for the greater good.
Uh, you know, we haven't we havenot begun an FDA process yet.
We're only 18 months into afive-year project, right?
So so we are developing this,but but um we are thinking
(16:34):
already, you know, we've had acouple of conversations about
that, and like how do weactually go and guarantee visual
safety in the practice?
Yeah.
SPEAKER_00 (16:42):
Absolutely.
So it still has a ways to go.
And to that point of the hypearound AI and how fast it's it's
building, um, you know, you heara lot of things about, oh, soon
there will be AI robot, physicalrobot doctors and nurses.
What are your thoughts on that?
And would there be a time wherethat's even needed?
(17:02):
What are your thoughts?
SPEAKER_01 (17:04):
Yeah, I think it's a good, it's a good question
because we all, you know, we'vewatched the sci-fi movies, we we
listen to the media and so on,right?
So I mean, I think in order forthere to be any notion of an AI,
you know, robot doctor, we needto have uh we need a few pieces
in place.
We already need to have a like aplatform, like a physical
(17:25):
healthcare platform that isrobust and deployable and has
and has certain fail-safes in itthat could always um fit fall
back to the expert human to makea decision.
I think that ultimately to methat's going to be required in
healthcare.
Um, but we also need asignificant amount of um social
(17:45):
um education, social awareness,and kind of time, I think,
right?
Like time heals all wounds.
Like I think time is needed togo and work through these
changes.
These are revolutionary.
The potential for that type oftechnology is revolutionary.
And so we can't just throw itout the user, throw it at throw
it at all all Americans, throwit at all the world citizens,
and like expect them to adoptit.
(18:06):
And so I think when, you know,even even in the Vigil project,
right now all the all thegeneralists are intended to be
humans, and we expect that to bethe case.
That doesn't mean we aren'tenvisioning a future in which uh
we could have a um a robot armin the vehicle that's maybe
(18:26):
mounted on the ceiling ormounted underneath the table or
the chair where the the humanthe patient is sitting, that is
getting the best optimal view sothat vigil can make the best
decision.
Right?
That's not a robot doctor, butthat is a you know actuated
robotic arm in the vehicle uhthat will help the human make
the better decision.
In some sense, you know, youknow, like humans can only see
(18:48):
out of their eyes, we're onlylooking one direction, but with
a robotic uh helper, we can havea whole different angle, maybe
from the side view, for example,of you know, to round out and
build up a more rich perception.
And this is, I think this is thepathway toward a future in which
there's just generally betterhealthcare provided, whether or
not it's from a human or afuture robot.
(19:09):
You know, and I I mean I've Iexperienced, I was lucky enough
to do my graduate work at JohnsHopkins, uh, because this is you
know, this notion of a robotaugmenting or helping a human in
some way to do healthcare is notin any way new.
Uh, you know, at Johns Hopkins,in the they had an ERC for
computer integrated surgicalsystems and technology.
Um, and Dr.
(19:30):
Um Russ Taylor was the PI ofthat, of the NSF funded center,
and they had this home-builtrobot called the steady hand
robot.
You know, and it could takemovements of the operator, of
the human surgeon, and um reducethem down to like micron level.
Um, I'm I'm speculating a littlebit, I think it was micron
level, at least nano level, sovery, very tiny level motions,
(19:52):
so that you know, an eyesurgeon, to use the
ophthalmologicalophthalmological example, could
operate, you know, in a verysensitive part of the human
body.
Um, and and you know, this waslike, I mean, I'm getting old
now.
This was like 25 years ago,right?
To see this.
So we're on this pathway, Ithink, of technology to augment
humans doing good for you knowin healthcare with with the with
(20:14):
with new technology.
SPEAKER_00 (20:16):
Yeah, absolutely.
That's the exciting part.
Well, if there was one keytakeaway you could leave our
audience with today, what wouldthat be?
SPEAKER_01 (20:24):
Well, I think the the key takeaway for me would
always be that AI is here toaugment humans and to you know
to build these teams in which uhwhat we collectively can achieve
will be magnified through thehelp of new technology, once
delivered safely andresponsibly.
SPEAKER_00 (20:41):
All right.
Yes, indeed.
Well, thank you so much forcoming on the show and sharing
your insights and all the newtechnology that's out there in
the future.
SPEAKER_01 (20:51):
Thanks, Amanda.
This was fantastic.
Glad to talk about it.
SPEAKER_00 (20:54):
And thank you to our audience.
If you have any questions orcomments, leave those below and
I'll try to answer them as soonas possible.
And have a wonderful day.
Hello and welcome to Full Tech Ahead.
I'm Amanda Rizzani, and with metoday I'm excited to have Jason
Corso.
He is a Toyota professor of AIat the University of Michigan
and co-founder of Voxel 51, acompany focused on visual AI.
He's also leading an exciting AIproject with ARPA H called
(00:42):
Vigil, which is exploring how AIcan help extend healthcare
access in remote communities bysupporting clinicians rather
than replacing them.
So, Jason, welcome to the show.
Can you share a little bit aboutVoxel 51, first of all?
SPEAKER_01 (00:57):
Absolutely.
Thanks, Amanda.
It's great to be here with you.
So Voxel 51 is indeed a dev toolsoftware company for AI that
supports the developer, like themachine learning scientist,
machine learning engineer, dataannotator, data curator in the
spaces of physical AI and visualAI.
Specifically, we make a softwaretool called 51, which supports
(01:21):
the full stack of the workflowsfrom managing, curating,
annotating data throughtraining, evaluating, and
deploying the models that aretrained from that data.
We work in various verticalslike automotive, healthcare,
physical security, productsupport, generally are supported
by a uh we're about 60 peoplenow, actually, which is great
(01:43):
for me.
You know, it's it's been massivegrowth in the last five, five or
so years.
Uh we have about four milliondownloads of the open source
software.
This is like a single userinstance of 51.
And we uh support both on-premand air gap solutions uh when
data security is reallyimportant to our customers.
SPEAKER_00 (02:01):
Awesome.
Well, I want to talk more aboutthis vigil.
What problem and role healthcarewere you trying to solve when
you created this?
SPEAKER_01 (02:11):
Awesome.
Yeah.
So right at the University ofMichigan, I have I had been
getting more into healthcare uhin the last three or four years
and was delighted when thisprogram from ARPH, the Paradigm
Program, came online.
So um not being a physician, youknow, this is I'm an engineer,
right, by training.
So having to learn aboutphysical physician or
medical-oriented problems,healthcare problems in the US
(02:33):
really is enlightening to me.
Um so Vigil really tackles a uma key problem in rural
healthcare in the U.S.
So generally, there's a lack ofspecialist, a lack of
expert-trained physicians, soMDs or DOs that operate in rural
America.
And to make that problem evenworse, there is an increasing
(02:56):
shortage of physical brick andmortar locations where rural
Americans can go nearby theirhomes to get healthcare, i.e.,
hospitals are closing, even justthe clinics are closing.
There's one study I'm aware ofthat was based in North Carolina
that measured the distance aNorth Carolina citizen had to
travel, the average distance, toget to their nearest brick and
(03:18):
mortar healthcare facilitybetween the years of 2012 and
2018 increased eightfold.
So that means someone wasdriving 15 minutes to get to
their nearest hospital.
Now they may have to drive twohours to get, or more, frankly,
because this is an average.
Um, so so ultimately, vigil uhis there to well, obviously, we
can't just, you know, clonephysicians, we can't build brick
(03:40):
and mortar hospitals, right?
That would have been done byother institutions already.
So what we are trying to do is,well, first of all, we make the
observation that although thereare few specialists, few
expert-trained MDs and DOs,there are actually quite a quite
a large number of registerednurses or physic physicians
assistants or even EMTs that arein those regions.
(04:02):
And although they may not bespecialists in any one
particular area, they aregenerally trained with medical
knowledge.
And so the idea is if we canequip them with kind of like a
hospital on wheels or like aclinic on wheels, um, and some
physically grounded AI, right?
So kind of like a Hay Sirimindset, right?
But uh but a Hay Siri, which weactually call Hay Vigil, right,
(04:25):
with with a vigil that actuallyhas cameras and eyeballs and
ears inside of this mobileclinic, so that when the
generalist is doing their workwith the patient, uh they're
actually able to be guided onthe fly by this AI system that
we call vigil.
Um clearly the this thisgenerally trained or a
generalist um healthcare workerhas the final say, right?
(04:48):
If they're ever not confidentwith what AI guidance they're
getting, they can always phonehome, you know, phone back to
the hospital, may elongate thevisit.
However, the idea is that we'llbe able to deliver significantly
better, uh richer, and morerobust healthcare in rural
America by building these mobilevans, staffing them with one or
two of these generalisthealthcare workers, and then
(05:11):
giving them a road, you know,affiliating with a with a rural
hospital, obviously, right?
Because we have to have, youknow, we're not gonna have
patients of our own, butaffiliating with a rural
hospital, and then they'll havea day in which they can treat
maybe 12, 16, 20 differentpatients who otherwise would
have had to take days off, fulldays off of work or what have
you.
SPEAKER_00 (05:29):
That is fantastic.
Well, I can say definitely, Ilive in the heart of Texas, and
there's a lot of ruralcommunities all over that are
all reliant on a hospital, likeyou said, uh hour, two hours
away.
So that is very helpful.
So why did you take an approachthat helps clinicians instead of
(05:50):
trying to just replace them withAI?
SPEAKER_01 (05:53):
Well, I think this goes back to my general
philosophy.
Um, you know, I I've been in thefield for about 25 years now.
I just I just believe that weare, as technologists in AI, we
are building tools that willaugment humans, right?
Humans are the center of ourexperience.
We are humans, right?
In some sense.
And I have this saying now thatI've been working with a
colleague of mine, for allhumans, for all life, right?
(06:14):
Like humans are the center ofour world.
And I don't think AI is here toreplace humans.
We probably could have anotherconversation about whether or
not like white-collar jobs aregoing to be lost to AIs.
I just don't believe so.
Frankly, I think it's gonna be aboon for white-collar jobs
because they'll have bettertooling around what can be done.
Uh, in in this particularhealthcare space, specifically,
(06:35):
uh, you know, we can't, we couldnot go out and do deliver vigil
in practice today because ofvarious state-level um like
laws, essentially, likeliability laws and you know,
very various safety laws thatare that are put in place for
great reasons to protectpatients.
Um and so we we just franklybelieve the right way today,
tomorrow, in the next five yearsto deliver this increased level
(06:56):
of rural healthcare is to justaugment the humans that are uh
in in the vehicle that will bein the vehicle to do the work
that they want to do, that theyknow how to do.
You know, it's one key reasonthere is that, right?
Like just tooling for thesehuman AI teams.
That's been the heart of mycareer in some sense.
Um, you know, another aspectthough is that, especially in
healthcare, there is asignificant value to the bond,
(07:19):
that the trust and the bondbetween the healthcare worker
and the patient.
And vigil is there not to do inany way uh violate or corrupt
that bond.
In fact, we we we have thisnotion of a triangle of trust
where the vigil worker, right,the healthcare worker is
trusting vigil to help him orher, and the patient is trusting
(07:40):
the healthcare worker, and thenkind of tacitly the patient is
hence trusting vigil.
And if that triangle doesn't getbuilt and get supported, then
vigil will not be a success.
SPEAKER_00 (07:49):
Yeah, I agree that personal connection and that
bedside manner are so critical.
I have actually been inpositions on a few occasions.
Um, my mom, for example, uhlives in a rural town where um
there's not as many eye doctorsavailable, and they roll up a
screen.
Um uh, and even that is a realperson on the screen, but it's
(08:13):
still a little odd when you'rejust talking to a screen, you
know, instead of going to, youknow, you go to the eye clinic
to see the eye doctor, but youjust get a screen rolled up.
So yeah, I understand.
SPEAKER_01 (08:27):
Uh I understand.
In fact, in fact, um I I believethere are on the order of 30,000
plus mobile health units inpractice today.
So your your kids are right.
So she's going somewhere andthen they're rolling up a screen
where where someone's beingvideo telecomed in.
But the notion of a mobileclinic is not novel to Vigil or
(08:48):
to this ARPA paradigm program.
I think what's what's actuallywhat's really novel, I believe,
is that there will be a humanoperating in the vehicle and
there'll be upskilled, right?
This generalist learning how todo things on the fly or not even
just learning, but like beingbeing reminded how to do things,
right?
So one for example, one one oneuh clinical service we we're
(09:09):
we're studying right now iscardiac diagnostic health
through ultrasound.
You know, typically anultrasound operator need must
take, I believe it's about twoto two years, maybe three years
of training to actually like beuh credentials in to perform an
ultrasound for cardiac health.
You know, it's it's great.
And I'm very grateful that wehave humans who go and do that
(09:31):
work.
And still we can't, but there'snot enough of them, frankly, not
only in rural places, but evenin in like urban, semi-urban and
urban places.
So um, in this case, instead,someone, you know, an RN who
might have touched an ultrasoundthree times can be upscaled on
the fly to go and do the samelevel of care, to get the same
quality images that either thecomputer will read automatically
(09:52):
or will just be stored, and thena radiologist would read post
hoc, um, which is not thatdifferent than going to a brick
and mortar uh facility.
You know, the the programmanager, uh Dr.
Bon Ku from RPAH, uh, theprogram manager of Paradigm, he
was visiting campus last weekactually for a site visit.
And uh in in his morningpresentation, uh made the made
(10:14):
the comment that he believes onthe order of 80% of all medical
care does not need to be done ina brick and mortar facility.
It could be done in thepatient's home or in in the
Walmart parking lot or in one ofthese mobile clinics.
And I think realizing thathealthcare of the future, or
like laying the ground, thestonework, the groundwork for
the healthcare of the future,you know, it's just it's frankly
(10:34):
exciting to be a part of that.
SPEAKER_00 (10:36):
Yeah, it is.
I mean, you think back to whenyou had those home office visits
from your doctor.
I mean, if you're really sickand ill, trying to get out and
go somewhere, it's difficult.
Not to mention the comfort ofbeing in your home.
So there's a lot of positivesfor sure.
Well, and as you said, kind ofbreaking down, you know, you
we're lacking in thesehealthcare areas.
(10:59):
So kind of breaking down thetime barriers, uh, making it
easier to get people out thereto help.
SPEAKER_01 (11:05):
Absolutely.
Um, indeed, I I remember, infact, when my spouse was about
to deliver our first child, thechiropractor came for a home
visit.
Uh, and that was uh that was abig help, in fact.
Yeah.
But these home visits can itjust changes the dynamic,
really, right?
So not not that there's uh well,I mean, the thing is it could
(11:25):
not not like levels of playingfield, but uh really just the
the notion that all humans arevalued and all humans are equal.
I I think it it becomes morerealizable when when we have
when we have the ability todeliver better better health
care.
SPEAKER_00 (11:39):
Absolutely.
Well, you know, AI can sometimesbe a hot button topic.
So when it comes to AI andhealthcare, what are some things
that people get wrong or somemisconceptions?
SPEAKER_01 (11:51):
Oh, it's a good it's a great question, right?
So um, well, first of all,indeed, there is a there's a
huge marketing machine behind AIthat broadly, you know, both in
healthcare and outside ofhealthcare, that unfortunately I
think is doing a little bit of adisservice to actual AI
technologists.
Uh, you know, in the sense thatuh there are increasing claims
(12:13):
about when AI will solve problemX or we'll take job Y, or we'll
have self-driving cars, youknow, 10 years ago, and we still
kind of don't have them, right?
So like there's just thischallenge where the message and
the hype around AI is notactually backed out by robust
engineering-based testing anddelivery.
Um, you know, in in thehealthcare domain, you know,
(12:36):
vigil is is um it treats theproblem a little differently
than we might think.
Um, so the the idea in Vigil isto first of all do no harm, kind
of like a you know traditionalhealthcare mindset, but that has
different layers of impact,right?
So some generalists maybethey're actually rather well
trained in in ultrasound, justas an example, right?
(12:58):
And some generalists are nottrained in ultrasound.
So if vigil gave the same levelof guidance to both of those
types of individuals, it wouldbe a problem, right?
They'd be distracting the moreexperienced individual.
And so what when we are modelingguidance, uh we're we're
actually using very contemporarystate-of-the-art artificial
intelligence methods.
You know, we have our ownhealthcare world model that sits
(13:21):
behind the technology, and thenan agent that is modeling not
only what is happening in theenvironment, but also what
should be happening in theenvironment in the context of
the knowledge about the skilllevel of the uh operator, of the
generalist.
And so if that agent decides,like that in some sense, the
goal of that agent is to decidewhether or not to provide
(13:44):
guidance or just let what isgood good, right?
Like, you know, we have thesemultiple screens happening in
the background that just kind ofwalk the generalist through the
steps of the procedure that theyshould that they should be
doing.
Those the generalists can lookat or not, right?
Those are not intrusive.
But at some point, for example,when you're holding an
ultrasound probe near the heart,and you need to get like what's
(14:04):
called an apical forechamberview to get an ejection
fracture, how healthy, how muchblood is getting pumped out of
the left ventricle.
Sometimes you have to make thesemicro adjustments, you know, it
a fraction of a degree.
And you really just build thissense over time.
Um, so Vigil can opt to givedifferent levels of training or
guidance at that point.
Uh, you know, it can render, forexample, through a projector,
(14:27):
yeah.
We actually create this kind ofshared perceptual space through
a projecting the interface ontothe patient's body.
And so it can render differentangles, different arrows.
Uh, it renders like a ghostprobe.
So you can, you know, map thereal probe to the ghost probe
just to really have differentways of interacting with the
patient.
And I think that that type of uhmodulated interaction through AI
(14:49):
uh is is really is franklyreally novel uh because we're
not you know, you as youprobably have heard, like these
general LLM type models, theythey're prone to hallucinate.
And in and right, and so and andif you tell them they're
hallucinating, they'll quicklybelieve you, right?
There's actually not much deepmodeling about the veracity of
(15:09):
the conversation.
Whereas in the healthcare, inthis in the vigil domain, we
can't hallucinate, first of all.
So we really have to be groundedinto what's allowable.
Um, but we also have to modelthe fallibility of the
generalist, the the likelihoodthat they may or may not make a
mistake.
Uh, and if they are making amistake, we need to both
(15:30):
anticipate it and then preventit, especially if there's any
any potential harm or whatever.
So modeling that like safety inthe interaction is super
critical.
Uh, and it's not something we'vereally seen enough of yet in the
general AI space.
I think that's one thing we'recollectively getting wrong,
right?
Like um, we we don't really haveappropriate uh in-model
(15:51):
guardrails.
From my understanding, most ofthe guardrails that are in the
systems that we are all many ofus are using on a regular basis,
are post or pre-hoc or post hoc,right?
They're they're guardrails onwhat could be set in the prompt
or what could be outputted bythe model to the to the user.
Um, and they're not sort ofguaranteeable in any notion of
like theoretical or mathematicalsense.
(16:13):
And we are in vigil, we'rereally trying to get toward
those guaranteeable guardrailsso that um uh we you know we we
we can really deploy this forfor the for the greater good.
Uh, you know, we haven't we havenot begun an FDA process yet.
We're only 18 months into afive-year project, right?
So so we are developing this,but but um we are thinking
(16:34):
already, you know, we've had acouple of conversations about
that, and like how do weactually go and guarantee visual
safety in the practice?
Yeah.
SPEAKER_00 (16:42):
Absolutely.
So it still has a ways to go.
And to that point of the hypearound AI and how fast it's it's
building, um, you know, you heara lot of things about, oh, soon
there will be AI robot, physicalrobot doctors and nurses.
What are your thoughts on that?
And would there be a time wherethat's even needed?
(17:02):
What are your thoughts?
SPEAKER_01 (17:04):
Yeah, I think it's a good, it's a good question
because we all, you know, we'vewatched the sci-fi movies, we we
listen to the media and so on,right?
So I mean, I think in order forthere to be any notion of an AI,
you know, robot doctor, we needto have uh we need a few pieces
in place.
We already need to have a like aplatform, like a physical
(17:25):
healthcare platform that isrobust and deployable and has
and has certain fail-safes in itthat could always um fit fall
back to the expert human to makea decision.
I think that ultimately to methat's going to be required in
healthcare.
Um, but we also need asignificant amount of um social
(17:45):
um education, social awareness,and kind of time, I think,
right?
Like time heals all wounds.
Like I think time is needed togo and work through these
changes.
These are revolutionary.
The potential for that type oftechnology is revolutionary.
And so we can't just throw itout the user, throw it at throw
it at all all Americans, throwit at all the world citizens,
and like expect them to adoptit.
(18:06):
And so I think when, you know,even even in the Vigil project,
right now all the all thegeneralists are intended to be
humans, and we expect that to bethe case.
That doesn't mean we aren'tenvisioning a future in which uh
we could have a um a robot armin the vehicle that's maybe
(18:26):
mounted on the ceiling ormounted underneath the table or
the chair where the the humanthe patient is sitting, that is
getting the best optimal view sothat vigil can make the best
decision.
Right?
That's not a robot doctor, butthat is a you know actuated
robotic arm in the vehicle uhthat will help the human make
the better decision.
In some sense, you know, youknow, like humans can only see
(18:48):
out of their eyes, we're onlylooking one direction, but with
a robotic uh helper, we can havea whole different angle, maybe
from the side view, for example,of you know, to round out and
build up a more rich perception.
And this is, I think this is thepathway toward a future in which
there's just generally betterhealthcare provided, whether or
not it's from a human or afuture robot.
(19:09):
You know, and I I mean I've Iexperienced, I was lucky enough
to do my graduate work at JohnsHopkins, uh, because this is you
know, this notion of a robotaugmenting or helping a human in
some way to do healthcare is notin any way new.
Uh, you know, at Johns Hopkins,in the they had an ERC for
computer integrated surgicalsystems and technology.
Um, and Dr.
(19:30):
Um Russ Taylor was the PI ofthat, of the NSF funded center,
and they had this home-builtrobot called the steady hand
robot.
You know, and it could takemovements of the operator, of
the human surgeon, and um reducethem down to like micron level.
Um, I'm I'm speculating a littlebit, I think it was micron
level, at least nano level, sovery, very tiny level motions,
(19:52):
so that you know, an eyesurgeon, to use the
ophthalmologicalophthalmological example, could
operate, you know, in a verysensitive part of the human
body.
Um, and and you know, this waslike, I mean, I'm getting old
now.
This was like 25 years ago,right?
To see this.
So we're on this pathway, Ithink, of technology to augment
humans doing good for you knowin healthcare with with the with
(20:14):
with new technology.
SPEAKER_00 (20:16):
Yeah, absolutely.
That's the exciting part.
Well, if there was one keytakeaway you could leave our
audience with today, what wouldthat be?
SPEAKER_01 (20:24):
Well, I think the the key takeaway for me would
always be that AI is here toaugment humans and to you know
to build these teams in which uhwhat we collectively can achieve
will be magnified through thehelp of new technology, once
delivered safely andresponsibly.
SPEAKER_00 (20:41):
All right.
Yes, indeed.
Well, thank you so much forcoming on the show and sharing
your insights and all the newtechnology that's out there in
the future.
SPEAKER_01 (20:51):
Thanks, Amanda.
This was fantastic.
Glad to talk about it.
SPEAKER_00 (20:54):
And thank you to our audience.
If you have any questions orcomments, leave those below and
I'll try to answer them as soonas possible.
And have a wonderful day.
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.
Betrayal Weekly
Betrayal Weekly is back for a new season. Every Thursday, Betrayal Weekly shares first-hand accounts of broken trust, shocking deceptions, and the trail of destruction they leave behind. Hosted by Andrea Gunning, this weekly ongoing series digs into real-life stories of betrayal and the aftermath. From stories of double lives to dark discoveries, these are cautionary tales and accounts of resilience against all odds. From the producers of the critically acclaimed Betrayal series, Betrayal Weekly drops new episodes every Thursday. If you would like to share your story, you can reach out to the Betrayal Team by emailing them at betrayalpod@gmail.com and follow us on Instagram at @betrayalpod and @glasspodcasts. Please join our Substack for additional exclusive content, curated book recommendations, and community discussions. Sign up FREE by clicking this link Beyond Betrayal Substack. Join our community dedicated to truth, resilience, and healing. Your voice matters! Be a part of our Betrayal journey on Substack.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com