Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
[TEASER] RAFAH HOSN: What has changed is that in the old days, we had the luxury of creating something,
0:08
going and piloting for three months until we know whether it works or not, and then
0:14
taking one year to productize!
0:16
That … that, that doesn’t work anymore!
0:18
Because guess what? In three months, this innovation is, like, topped by four other innovations, be it at
0:25
Microsoft or elsewhere.
0:28
So that speed is really shifting the mindset and the spirit of people.
0:33
[TEASER ENDS] GRETCHEN HUIZINGA: You’re listening to Ideas, a Microsoft Research Podcast that dives deep
0:39
into the world of technology research and the profound questions behind the code.
0:45
I’m Dr. Gretchen Huizinga.
0:53
In this series, we’ll explore the technologies that are shaping our future and the big ideas
0:58
that propel them forward. [MUSIC FADES]
1:01
My guest today is Rafah Hosn. She’s a partner, group product manager for AI Frontiers at Microsoft Research.
1:08
I’d call Rafah a sort of organizational conductor, working both with leaders to drive
1:13
clarity around the mission as well as program managers to make sure they have solid operational
1:19
strategies to execute on it.
1:21
Rafah has mad skills in bringing research ideas from lab to life, and I’m thrilled
1:26
to talk to her today. Rafah Hosn, welcome to Ideas!
1:29
RAFAH HOSN: Thank you, Gretchen. Oh, my goodness, I have to live up to this introduction now!
1:33
[LAUGHTER] HUIZINGA: Well, before we talk about research ideas, let’s talk about you and your own
1:40
sort of “reason for being” in the research world.
1:44
How would you describe your motivation for working in research and—assuming there was
1:47
one—what was the “big idea” or animating “what if?”
1:51
behind what you’re doing today? HOSN: Yeah, you know, I don’t know.
1:55
There are so many big ideas, to be honest!
1:59
Every day, I wake up and I often tell my husband how lucky, like so totally lucky and humbled,
2:05
I am to be where I am right now in this moment, like right now when society as we know it
2:11
is being totally disrupted by this huge leap in AI.
2:15
And why research? Well, I’ve tried it all, Gretchen!
2:18
I’ve been in research, I went to product, I did engineering, and I did full circle and
2:24
came back to research. Because, you know, for me personally, there’s no other environment that I know of, for me,
2:32
that has this amount of creativity and just infinite curiosity and intellect.
2:37
So working with people that are asking “what next?” and trying to imagine the next world
2:44
beyond where AI is today is just … this is the big idea.
2:49
This is why I’m here. This is why I’m excited to come to work every day.
2:54
HUIZINGA: Yeah. Well … and I want to drill in a little bit just, sort of, personally because sometimes
2:58
there’s a story, an origin story, if you will, of some pivotal aha moment that you
3:07
say, oh, that’s fascinating, that’s cool, that’s what I want to do.
3:12
Anything that piqued your interest way back when you were a kid or, sort of, a pivotal
3:16
moment in your educational years?
3:19
HOSN: Yeah, you know, so many different things that inspire you along the journey, right.
3:27
It’s not just one thing, Gretchen.
3:30
My dad was a doctor. He was my biggest inspiration growing up.
3:35
And the reason is because he had a lot of depth of knowledge in his domain.
3:40
And I wanted that. I wanted to have depth of knowledge in a domain.
3:45
So I went engineering against his advice.
3:48
He really wanted me to be a doctor.
3:50
[LAUGHTER] So he was not too happy.
3:53
But, you know, throughout my education, you know, I was there when smartphones came about,
4:00
when the internet was a thing.
4:04
And now, like with generative AI, I feel like I’ve lived through so many disruptions,
4:10
and every one of those was, “Oh my gosh!
4:13
Like, I am exactly where I want to be!”
4:17
So multiple inspirations, and every day, I wake up and there’s new news and I’m saying
4:23
to myself, “OK, that’s great.”
4:25
I love it! HUIZINGA: What a time to be alive!
4:28
HOSN: It is amazing! HUIZINGA: Yeah.
4:31
Well, you recently took on this new role in AI Frontiers at Microsoft Research.
4:38
And that very word “frontiers” evokes images of unexplored, uncharted territories
4:42
like the Wild West or for Trekkies, maybe “space: the final frontier.”
4:47
So what does it mean to you to be working at the frontier of artificial intelligence,
4:52
and what’s the big idea behind AI Frontiers?
4:56
HOSN: You know, it’s my biggest and most exciting adventure so far!
5:01
Working under Ece Kamar’s leadership in this AI Frontiers is really trying to push
5:10
ourselves to think, what’s beyond what there is right now in artificial intelligence?
5:16
Where can we push more, from a scientific perspective?
5:21
How do we translate these scientific discoveries into capabilities that people can actually
5:26
use and derive value from?
5:29
It’s a big responsibility, as well, because we just don’t want to push the boundaries
5:34
of AI for the sake of pushing.
5:37
We want to push it in a safe and responsible way.
5:40
So it is a big responsibility.
5:42
HUIZINGA: Yeah … HOSN: And fundamentally, you know, the unifying big idea in this team is to explore, you know,
5:51
how far can we push intelligence further into models and encapsulations of those models
5:58
so that we can, you know, have not just sort of an assistant but really a personal assistant,
6:07
an agent that can, kind of, do tasks for us, with us, seamlessly across multiple domains?
6:15
So this is what we’re trying to push for.
6:18
HUIZINGA: Mmm. Rafah, do you feel like you’re at the frontier of artificial intelligence?
6:22
I mean, what are the emotions that crop up when you are dealing with these things—that
6:29
you and your teams basically know about but the rest of us don’t?
6:34
HOSN: For most days, it’s excitement.
6:38
Sometimes it’s [LAUGHTER] … it ranges, to be honest.
6:41
I would say there’s a spectrum of emotions.
6:46
The dominating one is really just excitement.
6:51
There’s so much that has happened with GenAI, but I feel like it has opened up so many different
6:58
paths, as well, for us to explore, and that’s the excitement.
7:04
And then every time the world accomplishes something, you’re like in astonishment.
7:09
You’re like, wow, wow. HUIZINGA: Yeah …
7:12
HOSN: And then, and then, oh my gosh, what’s next?
7:16
And so, it’s a range of emotions … HUIZINGA: Right …
7:18
HOSN: … but I would say the dominating one is enthusiasm. HUIZINGA: Yeah.
7:21
Well, I’ve heard other people on your teams use words like surprise, sometimes even shock
7:27
… HOSN: Yeah, yeah, there are a lot of “wow” factors.
7:31
Every day, every day, I wake up, I read like my three favorite AI tweets or things like
7:39
that, and I’m like, “Oh my gosh. I wouldn’t have imagined that this model could do this thing,” so [LAUGHS] … um,
7:46
but it’s exciting. HUIZINGA: We may have to get those accounts in the show notes so that we can follow along
7:53
with your surprise and amazement in the mornings!
7:55
HOSN: [LAUGHS] Yes! HUIZINGA: Well, listen, when we talk about measuring the success of an AI system, we
8:01
often use the common convention of what we call benchmarks.
8:05
But I want to zoom out from AI systems for a minute and ask how you might measure the
8:10
success of an AI lab, which is what you’re working in.
8:14
What are your benchmarks or key performance indicators—we call them KPIs—for the work
8:19
going on at AI Frontiers? HOSN: Yeah, so I’m going to start by something that may sound surprising maybe to some, but
8:27
I think it’s the culture first. It’s the culture of endless curiosity, of enthusiasm coupled with a bit of skepticism,
8:40
to be honest, to ask the questions, the right questions, and this drive to push further.
8:46
So I would say one KPI of success for me, personally, is, you know, can we maintain
8:54
this culture of enthusiasm coupled with skepticism so we can ask hard questions and an envelope
9:03
of enthusiasm and drive for everyone?
9:06
So that’s one. I would say the other three are … one is around how much can we push scientifically
9:15
as a community, right?
9:18
This is a team of people that are getting together with a mission to push the boundaries
9:24
of our understanding of artificial intelligence.
9:27
So are we pushing that scientific boundaries?
9:30
Are we creating insights, not just for the scientific community, but also for Microsoft
9:36
and the world, so that we know how to derive value from these discoveries, right?
9:40
At the end of the day, it is awesome to push scientifically.
9:44
It’s even more awesome if you take this and translate it into something a human being
9:51
can use … HUIZINGA: Yeah …
9:53
HOSN: … or an enterprise can use.
9:56
And I think … that’s kind of my KPIs of success.
10:02
Culture first, pushing on the scientific boundaries, creating insights for the scientific community
10:08
as well as for Microsoft so we can derive value for us as a society, right.
10:13
HUIZINGA: Yeah. Well, continuing on this idea of success, and you’ve alluded to this already in terms
10:19
of characteristics of curiosity and so on, part of your job, as you put it, was “enabling
10:27
brilliant minds to find success.” So talk a little bit about the personal qualities of these brilliant minds and how you help
10:35
them find success. HOSN: Yeah, you know, everybody I work with brings different aspects of brilliance to
10:45
the table—every day.
10:47
So in our community of engineers, PMs, researchers, everybody is present with their ideas and
10:57
their strengths. And they’re pulling together to push harder and faster on our key priorities.
11:03
And I find folks working in AI these days, you know, to have a renewed fire.
11:09
It’s really amazing to see.
11:13
And I talk a lot about curiosity, but, you know, I cannot emphasize how much this is
11:20
driving a lot of our community to explore new paths that they hadn’t thought about
11:28
prior to this GenAI coming along.
11:31
And so everybody is showing up, present, asking these questions and trying to solve new scenarios,
11:41
new problems that are emerging.
11:44
And from my perspective, you know, as you mentioned, I just try to unblock, basically.
11:51
My team and I are here to [LAUGHTER] … well, two things I would say.
11:56
First is bring the outside-in perspective.
11:58
That’s so important because science is amazing, but unless you can derive value from it, it
12:04
remains an awesome paper and an awesome equation, right.
12:08
So asking, who can use this?
12:11
What are the scenarios it could, you know, light up?
12:17
How can we derive value? So those are the questions that my team and I can contribute to, and we are trying to
12:24
participate from ideation all the way to basically delivering on key milestones.
12:30
And that last mile is so important.
12:32
Like, once you know what you want to do, how do you structure?
12:36
How do you have an operational strategy that is amenable to these times, which is fast,
12:43
fast, fast, and faster? So that’s, kind of, what we’re trying to do here.
12:48
HUIZINGA: Yeah, yeah. Well, two things came to my mind in terms of what kinds of people would end up working
12:55
in this area. And one would be agility, or agile.
13:00
And that would, to me, represent in a researcher that the person would be able to spin or pivot
13:06
if something didn’t work out.
13:09
And the other one is sort of a risk-reward mentality.
13:12
It’s like, where are you willing to push to get that reward versus what might keep
13:18
you from even trying? HOSN: Yeah, so definitely in this AI Frontiers community, I’m finding a lot of adaptability.
13:29
So people willing to try, failing fast when they fail, and pivoting.
13:35
And you have to, nowadays, in this atmosphere that we are living in.
13:43
And because we have the privilege of working in research—and it’s really an honor and
13:49
a privilege, and I’m not saying it just lightly—but it is the place where you can
13:53
take risks, Gretchen. It is the place where failing is totally fine because you’re learning and you’re pivoting
14:01
in a way that allows you to progress on the next thing you tackle.
14:06
So I feel like most of the people I work with in this community, AI Frontiers, we are risk
14:17
takers. We want to push, and it’s OK to fail, and it’s OK to adapt.
14:22
So, I think, as an aggregate, that’s kind of the spirit I’m seeing.
14:28
HUIZINGA: In the past, Rafah, you’ve stressed the importance of both teams and timing.
14:32
And so we’ve been talking about the teams and the minds and the kinds of qualities in
14:36
those people. But what about the “when” of research?
14:41
How does timing impact what gets done in your world?
14:45
HOSN: Well, in this new era, Gretchen, everything is yesterday!
14:50
[LAUGHS] I mean, it is true.
14:54
AI research is moving at such speeds that I feel like we need to get accustomed to a
15:00
timing of now.
15:02
And if it’s not now, it’s yesterday.
15:05
So the timing is important, but the leeway has shrunk so significantly that I feel like
15:16
we have to really just be present in the moment and just move as fast as we can because everybody
15:24
else is moving at the highest speed.
15:27
So timing is “now,” is what I would say.
15:30
HUIZINGA: On that note, with so many innovations in AI coming out every day, every minute,
15:36
what you’ve just expressed is that research horizons are shorter than ever.
15:40
But as one of your team members noted in a recent panel, it still takes a lot of time
15:46
to translate a research artifact, maybe a noteworthy finding or a published paper or
15:51
an equation, an algorithm, into a useful product for humans.
15:56
So how are you then dealing with these newly compressed timelines of “it needs to be
16:01
done yesterday to keep up,” and how has the traditional research-to-product pipeline
16:05
changed? HOSN: Yeah, it’s an awesome question.
16:09
It is so true that going from research to a production-quality algorithm or capability
16:18
takes time. But what I’m seeing is that the research-to-capabilities is accelerating, meaning if you look at the
16:27
world today in generative AI and its surrounding, folks even in research are creating assets
16:37
as they are creating their research.
16:40
And so they are thinking as well, how do I showcase this?
16:45
And of course, these assets are not production ready.
16:48
But here’s the kicker. I think that the product teams are also adapting to this generative AI era, and they are changing
16:59
to meet this disruptive moment.
17:02
They are changing the way they think, and they are accelerating the way they productize
17:09
and look at hardening and securing the assets so that they can put them in the hands of
17:16
even a limited set of users just to get a feel of what it means to have them in the
17:22
hands of end users and quickly iterating so that they can further harden and further improve
17:29
the design until it’s production ready.
17:33
And I feel like our product partners are meeting the moments, meaning they also are really
17:41
adapting their processes such that they can get these assets and put them in the hands
17:46
of users and test them out before they actually release them.
17:49
HUIZINGA: Right. Let’s drill in a little bit more on that and talk about the traditional research-to-product
17:56
pipeline, where you would have a researcher working on something and then an RSDE.
18:01
What does RSDE stand for? HOSN: A research software development engineer.
18:05
It’s a mouthful. HUIZINGA: Right.
18:08
And then to the PM, or program manager, and then to the engineer.
18:11
And you’ve said this provocative statement: now everyone is a PM!
18:15
HOSN: Everyone is a PM! [LAUGHTER]
18:18
HUIZINGA: What do you mean by that? HOSN: I just, I just feel like if we are to meet the moment, we need to be thinking outside-in,
18:27
inside-out simultaneously.
18:31
And I believe that the spirit of program management, which is looking at the design from a user-centric
18:39
perspective, is embedded as we are ideating, as we are trying to explore new methodologies,
18:48
new algorithms, new assets.
18:50
And so what has changed is that in the old days, we had the luxury of creating something,
18:59
going and piloting for three months until we know whether it works or not, and then
19:05
taking one year to productize! That … that, that doesn’t work anymore.
19:09
[LAUGHTER] HUIZINGA: Right.
19:11
HOSN: Because guess what? In three months, this innovation is, like, topped by four other innovations, be it at
19:17
Microsoft or elsewhere.
19:20
So that speed is really shifting the mindset and the, and the spirit of people.
19:27
I have colleagues and friends, researchers, that are asking me, oh, scenarios, users … I
19:34
mean it’s amazing to see.
19:36
So, yes, everybody has gotten a little PM in them now.
19:41
[LAUGHTER] HUIZINGA: Yeah, I did a podcast with Shamsi Iqbal and Jina Suh.
19:47
And Shamsi was talking about this concept, this old concept, of the researcher being
19:52
in their lab and saying, well, I’ve done this work; now go see what you want to do
19:55
with it. I don’t think you have that affordance anymore as a researcher.
20:00
HOSN: No … HUIZINGA: You’ve got to work much more tightly with other team members and think like a PM.
20:04
HOSN: Totally. HUIZINGA: So let’s talk about how the big general idea behind AI Frontiers is giving
20:11
birth to smaller, more specific ideas.
20:14
What are some of the research directions and projects that you could tell us about that
20:18
illustrate this vision here? HOSN: Yeah, and I’m sure you’ve heard some of it come from Ece Kamar as she spoke
20:27
on this community that we have.
20:30
In AI Frontiers, we’re exploring, I would say, three major areas of research.
20:38
And I want you to imagine a stack.
20:40
At the bottom of the stack, we’re asking ourselves questions around, what are some
20:46
new architectures we can be thinking about for these foundational models?
20:51
How do we create them? What kind of data we need to train them, to pre-train them.
20:57
And then on top of that stack, which starts with a foundation model, we’re asking ourselves,
21:03
OK great, you have a pretrained model.
21:06
In a lot of cases, when you’re creating especially small models, you need to fine-tune
21:11
them. So what is this methodology and data generation pipeline that we’re going to use to fine-tune
21:19
these models and specialize them for both across domains and across skill set?
21:28
And on top of that—so now we’re on the third layer—we have a final layer that encapsulates
21:34
these models and orchestrates among them to allow them the ability to do, you know, complex
21:43
tasks. And we don’t want to stop there because for us it’s … we don’t want to have
21:49
an agent that just does things and doesn’t learn.
21:53
So that learnability, that learning on the job, like we do as humans, is something we’re
21:59
asking ourselves, as well. Like, how do we encapsulate these models?
22:03
We orchestrate among them.
22:05
And we allow these encapsulated things, we call them agents, to learn on the job so that
22:12
they can accomplish more complex tasks.
22:14
So those are the three things. And then cutting across these three layers, imagine there’s a thing that cuts across
22:21
them, is doing everything in a way that allows us to rigorously evaluate and to ensure that
22:28
we’re doing things in a safe and responsible way.
22:32
So those are the main things that we’re working on.
22:35
Does that make sense? HUIZINGA: That’s … yes, it does.
22:39
And I imagine, you know, if you go to the website and you see those, kind of, three
22:44
main areas, I imagine that even under there, there are specific projects on, you know,
22:50
how then do we iterate?
22:52
How then do we explore? HOSN: That’s right.
22:54
That’s a good plug for people to visit the AI Frontiers website!
22:58
Thank you, Gretchen! [LAUGHS]
23:01
HUIZINGA: Well, I’ve been intrigued for a while by this idea of what you’ve called
23:05
bi-directional enrichment, which represents both how research informs product but also
23:11
how product informs research, but you’ve recently talked about how this idea has expanded
23:16
to embrace what you call multi-directional enrichment and co-innovation.
23:21
So what do you mean by that, and what does it look like for you?
23:24
HOSN: So we talked just moments ago how the time has shrunk tremendously in artificial
23:33
intelligence and the speed at which innovations are coming out.
23:38
So what does that mean when you are sitting in research and you’re trying to derive
23:44
value for Microsoft, for example?
23:47
It means that now, rather than going on a journey to try out you know different things,
23:57
what you want is for product to come on a co-innovation journey with you.
24:04
And not every team has the capability or the time or the resources to do it.
24:11
But sometimes product teams have applied scientists that are asking themselves very similar questions.
24:20
And so now we have this huge synergistic effect by which, you know, researchers can come and
24:26
explore their research but anchor them in a real-world scenario that the product team
24:35
is, you know, asking themselves about.
24:38
And that’s what I mean by co-innovation.
24:41
And we look for co-innovation, so these are product teams or applied scientists in product
24:46
teams that are not looking at something I can ship tomorrow.
24:49
Because that’s not … that’s not frontiers.
24:52
That’s feature-function that they can deliver right now to their customers.
24:55
When we co-innovate, we have to co-innovate on a bit of a longer timespan.
25:01
Now it’s no longer years, right? With generative AI, everything is months, but nonetheless, this is not next week.
25:09
This is in a few months.
25:11
And so … but this is really, really great because, again, I keep saying this and I have
25:16
maybe a huge bias, but I do believe that research, without it being anchored in real-world scenario,
25:23
just doesn’t have the same effect.
25:28
So I have a bias for that. It’s my PM hat, what can I say?
25:32
I love real-world scenarios! [LAUGHTER]
25:34
HUIZINGA: What you just referred to is an interesting flow.
25:38
I’ve noticed in my years doing this podcast that some people that started in research
25:45
ended up over in product—and we’ll call them embedded researchers, if you will—and
25:50
then some people that were in a product scenario come back over to research.
25:55
And so, there’s this flow, multi-directional, bi-directional, and also where they’re placed
26:03
within the company. How do you see that flow and the value of that flow between these organizations?
26:13
HOSN: Yeah, you know, like, I think that the flow is important because that’s how cross-pollination
26:23
happens. And you talked about brilliant minds.
26:26
In product teams, there are brilliant minds, as well, right.
26:31
And although their focus area is more around the product they live and breathe every day,
26:39
this is enriching to researchers and continues to be enriching because when you deploy research
26:47
capabilities in a real-world setting, there are surprising new research questions that
26:53
come up, not just engineering. A lot of times people think of research, OK, yeah, you scale it, you harden it, you secure
27:00
it, and it’s good to go. But that’s not always the case.
27:03
In a lot of cases, because of the interactivity that happens with real-world scenarios, it
27:11
opened up brand-new paths for research.
27:13
And so I think that flow continues to happen even now.
27:18
It’s just compressed. It’s just that researchers are no longer thinking six years.
27:25
Researchers are thinking three months. Like, what am I going to do in three months?
27:30
Because in three months, there will be a hundred other researchers that are coming up with
27:34
innovation on the same question.
27:37
So I think the flow still exists.
27:39
I think that time has shrunk.
27:42
And I think the mobility from researchers and research going to product and vice versa
27:50
is enriching for the people that do it because you gain different perspectives.
27:55
HUIZINGA: Well, and let’s push in even there a little bit.
28:00
Researchers like everyone else can get comfortable looking at things through a particular lens.
28:04
I would say that’s a human trait, not just a research trait …
28:07
HOSN: Absolutely. HUIZINGA: … until a disruption challenges their status quo.
28:11
So you’ve talked about LLMs, which we’ve called large language models, as being a good
28:16
forcing function for researchers to think differently, even about the questions they’re
28:22
asking. Can you elaborate on that a little bit?
28:24
HOSN: Yeah, yeah, so, you know, the large language models and this disruption that we
28:35
are living in at the moment is lighting fire underneath a lot of people’s intellect,
28:42
I’m going to say.
28:45
And so I think that people have to adapt quickly to change.
28:54
And this is key. Adaptability, I believe, is just a key ingredient in doing research nowadays.
29:03
Why? Because a lot of people are thinking directionally the same.
29:10
And so, you know, if you’re not the first, you’re going to have to adapt to what came
29:17
out. And then you have to think of, how do I differentiate?
29:21
So the second point I would say is differentiation.
29:25
And this mindset of, you know, how do I adapt to what just came out?
29:31
How do I differentiate?
29:33
And then—Rafah’s bias—how do I anchor in real-world scenario?
29:39
This is the home run.
29:41
And I would say you package all of this and focus, focus, focus … and you get a gold
29:47
mine. HUIZINGA: I’m hearing “yes, and …” in this response in the sense of not everyone’s
29:55
going to be first, but then, what else?
29:58
This is back to the frontiers. It’s like, how do I differentiate?
30:01
Yes, that’s awesome. And we’ve got this …
30:03
HOSN: Exactly. And how do I build on what has just been discovered and give it a little bit of an edge or push
30:11
it a little further or take it in a brand-new direction?
30:15
I mean, so many different possibilities, but it does take adaptability, like a flexibility
30:22
in the mindset, I would say. HUIZINGA: Yeah.
30:25
Well, let’s go back to what you alluded to earlier, this idea of responsible AI.
30:31
This is a big deal at Microsoft.
30:34
And researchers are very thoughtful about the question of what could possibly go wrong
30:37
if we got everything right. But how does that translate practically, and what concrete steps are you taking at what
30:45
I’ll call the “frontier of responsibility?”
30:48
HOSN: Yeah, and as I mentioned, you know, being at the frontiers is amazing.
30:55
It also holds a big responsibility.
30:57
We have so many different, I would say, checks and balances that we use, in model training
31:06
and fine-tuning, to ensure that we are on top of all the regulatory, the policymaker
31:15
suggestions, and we are abiding by Microsoft values first and foremost and responsibility
31:24
in creating these innovations.
31:27
So practically and tactically, what happens is that there are processes for how you actually
31:37
even release any type of model.
31:41
And this is just research.
31:43
And when it goes to product, they have their own compliance, you know, a stricter even
31:49
compliance, I would say, process that they go through.
31:53
So we try, and I try particularly, to partner with our privacy champions, with our legal
32:00
champions, with our people that are looking at this from a responsible AI perspective,
32:06
so that we bring them in early on, and we say, hey, we’re thinking of doing this.
32:12
And they tell us, well, you know, if you’re thinking about it this way, you might want
32:16
to consider this. So we’re trying to bring them in as early as possible so that also we don’t go all
32:22
the way and then we discover we did something wrong, so we have to backtrack.
32:26
So I would say, you know, having these partners and colleagues come in early in the game just
32:34
saves everybody a lot of time.
32:37
And all this responsible AI for us, it’s ingrained with how we work, meaning we bring
32:44
our champions early on and then we have them advise us as we move along the journey to
32:50
create these innovations. So by the time we’re done, we know we’re good, right.
32:55
And even by the time we’re done, we recheck everything, we run a lot of evaluation benchmarks,
33:00
and, you know, we do the right thing per policies at Microsoft.
33:05
So we take it very, very seriously.
33:07
HUIZINGA: Well, let’s go back to this idea of research horizons for a second and anchor
33:13
it in the way that we approach research.
33:16
So many ideas are basically iterative steps on existing work, and they make a lot of sense
33:21
… this is the next step … but then there are those out-of-the-box ideas that feel like
33:26
maybe bigger swings—some might even call them outrageous—and in organizations like
33:31
Microsoft Research, they might get the green light, too.
33:34
Where do you find this idea of the outrageous or maybe longer-term idea finding a home or
33:42
a place in an organization like Microsoft Research, and have you ever worked on something
33:46
that felt outrageous to you? HOSN: Umm, you know, we like outrageous!
33:52
That’s why we’re in research, right?
33:55
So outrageous is good.
34:00
I haven’t, to be honest, worked on an outrageous, but I am confident I will be.
34:07
So … [LAUGHTER] I just have this belief that in AI Frontiers, we are going to have
34:15
outrageous ideas, and we’re going to work on them, and we’re going to make bets that
34:22
basically are hard to make in other parts of the company because we have the privilege
34:29
of taking them and pursuing them.
34:33
And, yes, they may fail, but if we have a breakthrough, it will be a significant breakthrough.
34:38
So, so I think that outrageous is good.
34:41
We need to think big. We need to take big leaps, big ideas.
34:47
We also need to know how to fail gracefully and pivot fast!
34:51
HUIZINGA: Hmmm. Mmm.
34:54
You know, it strikes me, and I’m laughing to myself, it strikes me, even as we’re
34:59
talking, that the idea that you work in AI Frontiers, that’s outrageous to most people
35:06
and, and it’s normal to you.
35:08
So maybe this idea of, “I haven’t worked on anything outrageous” is like, no, you
35:13
live in outrageous, it just doesn’t seem like it!
35:16
[LAUGHTER] HOSN: Maybe.
35:18
It’s my day-to-day job, so, yes, I guess you’re right. HUIZINGA: Right.
35:21
I mean, yeah, you say, we love outrageous, and that’s where it is right now.
35:25
Every day that I follow, sort of, AI Twitter also and find myself going, seriously?
35:32
That happened yesterday?
35:34
What next? HOSN: Yeah, in two hours, there’ll be yet another thing.
35:37
So, yeah, I guess I am living in outrageous, and I love it!
35:41
It’s amazing! [LAUGHS]
35:44
HUIZINGA: Yeah, maybe the idea of outrageous is just changed.
35:49
HOSN: You know, you’re so right. I think that it’s become the norm.
35:54
And it is, once we anchor in generative AI, and we push further on this idea, maybe we
36:08
will go back in a cycle where outrageous is outrageous, but today it’s our life.
36:14
It’s where we live. It’s what we breathe every day.
36:17
So it’s become a norm.
36:20
HUIZINGA: Yeah. Well, as we close, Rafah, I want to ask a question anchored on the big idea behind AI
36:28
Frontiers.
36:30
What do you believe might be true in say 10 to 15 years, and what should we be doing about
36:35
it now? In other words, how does what we believe about the future influence how we conceptualize
36:40
and execute on ideas today? HOSN: Yeah, you know, it’s … I can’t even predict what I’m going to be doing
36:47
tomorrow! But … [LAUGHTER] here’s, here’s what I think.
36:51
I think that we are truly approaching a moment in human history where a lot of unsurmountable
37:01
problems, like very hard-to-tackle diseases that have been so hard, I think we are approaching
37:11
a moment, you know, soon, I hope it’s even sooner than 10 years, where generative AI
37:19
and innovations on top of it could lead to a lot of resolution for things that today
37:25
… that cause unsurmountable pain and suffering.
37:29
I’m very hopeful that with what we are creating that we can, you know, take inefficiencies
37:37
out of so many different things that we see today that take time so that we liberate ourselves
37:45
to think about the “what next” societally, right?
37:49
I think what we need to be doing right now, to be honest, to influence the future is think
37:58
about our curricula. What are we going to teach our kids?
38:01
What are they going to work in?
38:04
This is where I’m hoping that we pour some of our creativity, education system.
38:10
How are we preparing the next generation?
38:13
What are the paths that we are going to forge for them, knowing what we know today, knowing
38:18
what this technology can bring forth?
38:21
So my hope is that we put some brain power into that.
38:25
HUIZINGA: Rafah Hosn, it’s always a pleasure to talk to you.
38:29
A sincere pleasure, a delight.
38:32
Thanks for joining us today on Ideas.
38:34
[MUSIC PLAYS] HOSN: Thank you so much for having me, Gretchen.
38:37
[MUSIC FADES]
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More