Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Microsoft at
0:02
middle age. From
0:04
American public media,
0:07
this is Marketplace
0:10
Tech. I'm
0:12
Megan McCarty
0:15
Carino. Microsoft
0:17
celebrates its 50th
0:19
anniversary this month. The company
0:21
started as a small software
0:24
startup co-founded by Bill Gates
0:26
and Paul Allen in a
0:29
garage in Albuquerque, New Mexico.
0:31
And it went on to
0:33
revolutionize personal computing. business productivity,
0:36
and now, it hopes, artificial
0:38
intelligence. With big investments in
0:40
open AI, the maker of
0:43
chat-GPT, Microsoft has set about
0:45
integrating the technology across its
0:48
products, and it recently unveiled a
0:50
slew of new upgrades to its
0:52
co-pilot AI assistant, like memory. So
0:54
it retains personal details like what
0:57
food you like or when your
0:59
kids' birthdays are, and can use
1:01
that information to make dinner reservations
1:03
for you or pick out a
1:06
gift. It's also added a vision
1:08
upgrade that allows the AI to
1:10
analyze photo and video and give
1:12
tips on, say, redecorating your kitchen.
1:15
We recently spoke with Yusuf Meti,
1:17
consumer chief marketing officer at Microsoft,
1:19
to learn more about the new
1:21
features. I'm really excited about the
1:23
idea that Copilot can now
1:25
anticipate your needs and, you know, notify
1:27
you of things that are upcoming. It can
1:29
start to give you better answers.
1:32
So, for example, if you're always
1:34
trying to solve a particular problem,
1:36
you don't have to go back and ask
1:38
it every single time. Now you can say,
1:40
hey, remember that time I was trying to
1:42
change the oil on my car or I
1:45
was trying to configure something on my
1:47
PC. My PC. Copilot can say, oh yeah, this
1:49
is what we did last time. One of the
1:51
things it's beautiful is as you're walking around the world
1:53
if you have a question about something you're looking
1:55
at like a certain building and you want to
1:58
know the origins of it or you're looking
2:00
at a I was doing this
2:02
yesterday with some friends, you look
2:04
at a particular flower, you can
2:06
just, you know, pull up your
2:08
phone and say, tell me about
2:10
this flower, and Copala can answer
2:12
that for you as well. With
2:14
this greater personalization and memory of
2:16
personal details, I think always come
2:18
concerns about privacy, how is Microsoft
2:20
handling the security of, you know,
2:22
this intimate personal data? For sure,
2:24
privacy and security are top of
2:26
mind for us. I would say
2:28
we start with that as the
2:30
key feature. Because as we've learned,
2:32
it's more important that people trust
2:34
the technology than it is any
2:36
particular neat thing that you can
2:38
do with it. So on things
2:40
like memory, you're completely in control.
2:42
If you don't want to have
2:44
copilot remember things, you can just
2:46
turn it off. If you want
2:48
to selectively delete things that you've
2:50
done, you can do that as
2:52
well. With things like vision. those
2:54
are not on by default. You
2:56
have to manually go turn those
2:58
on and then we'll have a
3:00
very prominent notification that co-pilot vision
3:02
is on so that you know
3:04
when to turn it off when
3:06
you're done using it. And where
3:08
are these kind of requests being
3:10
processed? You know, all of the
3:12
various tasks, are they going to
3:14
the cloud? Are they on device?
3:16
Yeah, so the way it works
3:18
is when you initiate a co-pilot
3:20
action. For example, You know, my
3:22
wife's birthday. Essentially, what happens is
3:24
Copilot uses the browser, your browser,
3:27
to then essentially go to the
3:29
website and then Copilot tries to
3:31
navigate and make reservation. So it'll
3:33
go to the website, it'll go
3:35
to the website, it'll find the
3:37
links for reservation, and it'll use
3:39
AI to figure out what's there.
3:41
And then as appropriate, it'll check
3:43
back in with you. So when
3:45
the time comes to, for example,
3:47
fill out a capture. And when
3:49
it comes times to, you know,
3:51
actually book the thing, if you,
3:53
for example, need to enter your
3:55
credit card, Copot will tell you,
3:57
hey, you need to take over
3:59
and enter your credit card information.
4:01
So this kind of agentic AI
4:03
personal. AI, it is all the
4:05
rage now, and there are kind
4:07
of a lot of similar features
4:09
in products out there, including even
4:11
from, I think, Microsoft's partner, Open
4:13
AI. What do you think differentiates
4:15
co-pilot from the others out there?
4:17
Well, one of the things we're
4:19
doing on Microsoft is we actually
4:21
have two co-pilots. We have co-pilot
4:23
for your consumer life, and we
4:25
have Microsoft 365 co-pilot for your
4:27
work life. For the Microsoft 365,
4:29
we can provide enterprise security security.
4:31
We have compliance with the data
4:33
in your enterprise. It can access
4:35
your corporate graph data so that
4:37
you get very unique capabilities there.
4:39
On the consumer side, we lean
4:41
more to providing a little bit
4:43
more personality, a little bit more
4:45
what we think of as IQ
4:47
support, but EQ support, meaning it
4:49
can support you with the environment
4:51
as a whole, even beyond just
4:54
answering questions. So the warmth, the
4:56
personality, the support for you. It's
4:58
a much richer experience and when
5:00
you use it you really can
5:02
tell the difference between it and
5:04
tools that are one-size-fits-all. We'll be
5:06
right back. You're listening to Marketplace
5:08
Tech. I'm Megan McCarty Carino. We're
5:10
back with Yusuf Meti, consumer chief
5:12
marketing officer at Microsoft. Microsoft would
5:14
seem to have kind of a
5:16
big scaling advantage here with the
5:18
office suite already being so widely
5:20
adopted. But given that scale, I
5:22
mean, what is the cost of
5:24
adding all these AI features and
5:26
sort of training users to be
5:28
more and more reliant on advanced
5:30
AI to accomplish tasks like search
5:32
that were much cheaper before you
5:34
add an LLLM in? Yeah, that's
5:36
a good question. There's two parts
5:38
to it. First, I would say
5:40
on the proactive side, you were
5:42
just so much more efficient using
5:44
copilot than say search. because you
5:46
can get right to the answer
5:48
you want. You don't have to
5:50
go through a bunch of links
5:52
and eventually find what you want
5:54
and frankly the accuracy is just
5:56
that much better. So the time
5:58
savings for you as an individual
6:00
is significant. From a cost of
6:02
building it, we use a variety
6:04
of different tools with Copilot and
6:06
a variety of different AI models
6:08
underneath the covers. So we're able
6:10
to optimize the cost to serve
6:12
and provide that functionality. And the
6:14
way that technology is just moving
6:16
at today's pace, we're ever increasing
6:18
the ability to reduce the costs
6:20
for offering these great capabilities for
6:23
people. Let's talk about another big
6:25
cost, which is environmental. Microsoft famously
6:27
has a goal to get to
6:29
carbon negative by 2030, and all
6:31
this AI uses an incredible amount
6:33
of energy. How do you square
6:35
that? Well, we're very focused on
6:37
our goals and we're making great
6:39
progress. So we feel good about
6:41
making our goals for 2030. Part
6:43
of how we do that is...
6:45
We really, as I said before,
6:47
take advantage of a number of
6:49
technologies to reduce the cost, to
6:51
reduce the power consumption, to answer
6:53
those. Some of that is more
6:55
advanced, sophisticated models. Some of it
6:57
is we're able to run some
6:59
of the AI locally on your
7:01
device on your computer. And so
7:03
you don't actually have to use
7:05
all of that power in the
7:07
cloud. You can use your local
7:09
PC and that's what we have
7:11
with these co-pilot plus PCs that
7:13
can run AI locally on your
7:15
device. the first 50 years of
7:17
Microsoft. How do you define its
7:19
legacy? And how are these new
7:21
tools kind of bringing the company
7:23
into the next 50 years of
7:25
that legacy? Well, one of the,
7:27
you know, the visions and dreams
7:29
Bill Gates had was really to
7:31
build Microsoft as essentially a software
7:33
factory so that it could produce
7:35
all sorts of software at increasingly...
7:37
more accessible, more inexpensive ways, essentially
7:39
building the platform for the software
7:41
industry. And now as we look
7:43
forward, we're kind of back to
7:45
the same roots. We want to
7:47
build the platform for AI development
7:50
for the world. We're building the
7:52
AI tools that will allow anyone
7:54
to be a great creator, have
7:56
anyone be someone that can build
7:58
the next big AI software. application
8:00
or computer experience. And so we're
8:02
excited that in the next 50
8:04
years, at this place in time
8:06
and this place with the technology
8:08
development, that is going to happen
8:10
and will be an empowering company
8:12
in that world. That was use
8:14
of Meti at Microsoft. Microsoft
8:22
held its 50th anniversary festivities last
8:24
Friday at its headquarters in Redmond
8:26
Washington, but the event was interrupted
8:28
by a couple of employees protesting
8:31
Microsoft's connection to the war in
8:33
Gaza. An associated press investigation earlier
8:35
this year found AI models from
8:38
Microsoft and Open AI had been
8:40
used as part of an Israeli
8:42
military program to select bombing targets.
8:45
Some employees also rallied outside the
8:47
event. Microsoft said in a statement,
8:49
quote, we provide many avenues for
8:52
all voices to be heard. Importantly,
8:54
we ask that this be done
8:56
in a way that does not
8:59
cause a business disruption. If that
9:01
happens, we ask participants to relocate.
9:03
We are committed to ensuring our
9:06
business practices uphold the highest standards.
9:08
The AP reported that two employees
9:10
who interrupted the event had lost
9:12
access to their work accounts, though
9:15
it wasn't clear if they were
9:17
fired. Daniel Shin produced this episode.
9:19
I'm Megan McCarty Carino, and that's
9:22
Marketplace Tech. This is APM. If
9:24
there's one thing we know about
9:26
social media, it's that misinformation is
9:29
everywhere, especially when it comes to
9:31
personal finance. Financially inclined from marketplace
9:33
is a podcast you can trust
9:36
to help you get serious about
9:38
your money so you can build
9:40
a life you've always dreamed of.
9:43
I'm the host Janelius Pinal and
9:45
each week I ask experts important
9:47
money questions like how to negotiate
9:50
job offers how to a
9:52
college that you
9:54
can afford, and
9:56
how to talk
9:59
about money with
10:01
friends and families. Listen
10:03
Listen to financially Encline
10:06
you get your
10:08
podcasts.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More