- Hi, everyone, this is Chris Duffey
from the Think Tank by Adobe.
I'm joined by Kathryn Hume from Integrate AI.
We spent the last couple of days chatting about
all things AI, so we're thrilled to have you here
to continue the conversation.
- Yeah it was a wonderful day.
- First topic, potentially, is about your background,
you have a fascinating background.
It's really interesting intersection of the arts,
with your literature background,
but also mathematics, and I think that's a great
reflection of what Artificial Intelligence
is in some ways.
This marriage of art and science.
- Yeah, I got really lucky.
I must say, when I was doing my PhD in literature,
A, I never imagined that five or six years later
I'd end up in the position that I'm in.
And B, that my training and thinking critically about how
cultures struggle with adopting new technologies
would be as useful as it is for work in AI.
But, it's been great, and helps me think
about the ethical issues and so long et cetera.
- So how are you applying that background
and what you do to the future of AI in the enterprise space?
- Sure, I mean, we talked a lot yesterday about
this not just being about technology,
but always being about the magical mixture
between humans and machines.
And going beyond just optimizations of thinking about
ways to use technology to enhance our creativity,
or do creative things.
And so, for me, at Integrate we're focused
on everything that is customer experience, customer
journeys, helping enterprises apply AI
to deliver better experiences to customers.
And a lot of that starts with not,
all right, what's our metric, and how can we improve it?
That's important from a programmatic perspective,
but more, who are our customers, how do they feel,
where do they live?
And what does it mean to think about how we can use
technology to engage with them more impactfully.
- And we chatted a bit about magical, right?
The notion that the technology is not magical
in of itself, it's what it delivers in terms of the
experience, how are you approaching that?
- It's interesting.
I have two minds regarding thinking about AI like magic.
On the one hand, I think it's really important
to empower organizations to have them know, it's just math.
So, I think math's magical.
But, most people are kind of like, oh math,
it's kind of boring.
But, on the other hand, I think, just the ability
to create.
If you're thinking about being a consumer,
and suddenly you go from having to go fill out forms,
and being frustrated with an interface,
and it's like making it so hard to tap into the value
of business, to somewhere it feels like,
without being creepy, right?
The business knows you, empathizes you,
and creates something that can
make your life more meaningful.
For me, that's sort of where the magic lies.
- And so much about these magical moments
are driven by data.
Maybe we can chat a little bit about the responsibility
that that entails to deploy that.
- Yeah, absolutely, so let's say,
you, as a privileged white male,
who happens to be born the way that you were,
you might have a magical experience.
But, somebody who's an African American woman,
older woman, because the data the system's been trained on
data, there might be a lot of customers like you,
so they're really great for you.
But, they're really not great for her, right?
So when we think about creating magical experiences,
we have to think about magic for everyone,
and not just people that are well represented in the data.
And so it's just important that we don't replicate
some of the stuff that we've inherited from the past.
But, think about who our users are, across the board.
And, for me, here's where the magic comes in.
As opposed to seeing that as an obstacle,
and fearing bias, and fearing AI in terms of liabilities,
what if we think about it as new market development?
What if the realm of African American grandma's
is an untapped market for a business?
And can lead to huge returns on investments.
- Speaking of the future of AI,
and the enterprise, I think you mentioned a
bold statement that we're moving beyond rules based AI.
To some new forms, if you'd like to chat a little bit
about that?
- Yeah, for sure.
So I just gave a talk about building an enterprise AI
culture, and really love how Jeff Basos phrases it,
where he says we used to be able to automate things
where we could clearly say, if this happens,
and that happens, and that's great when we're
doing analytics and counting, and want to know how
a twitter campaign is performed.
When we go to the role of AI,
we can basically make an educated guess,
that something might be the case.
And the cool thing is, there's a lot of things
that are more fuzzy like that.
And that's where creativity comes from.
Because its not that this needs to be the case,
but if we tweak that, and at this variation and
parameters, we get this sort of beautiful new thing.
The problem is, it can be really hard for businesses
to manage probabilities.
Because it's hard to say how much what's the ROI gonna be,
what does this product look like?
What are the risks?
They have to design experiments,
and get a lot of feedback, so it's a different mode
of operating, but that can lead to unexpected,
but really amazing returns.
- I think one thing that we also collectively agreed on,
is the future is going to be changing.
Not only drastically, but even faster than ever before.
And there was this notion of creative elasticity
meaning that, the one thing that we can predict
with certainty is that we have to go into it with
agility and knowing that the land, marketplace
is going to be changing.
Thoughts on that?
- I think for me, this goes back to your first question.
So, my background in math and comparative literature,
I believe that having a sound, liberal arts education,
and being a critical thinker who is always
questioning the status quo, always sitting there and asking
what really matters, how things might change,
actually is a way to build an agile and adaptive mindset.
And its very different than learning one skill,
and going and specializing and knowing how to do one
thing really well.
I think there still will be room for really deep
specialists in the future, but I think the role
of the generalist who's able to adapt.
Might not be the best at one thing, but is good enough
at many things, which is kind of the opposite of AI.
AI's are idiot savants, they're not Renaissance men.
So I actually think it's pretty cool
that we humans, get to explore curiosity.
And just dabble in different domains,
and let the machines take care of the stuff
that's all sort of one track minded.
- And how do you define artificial intelligence?
We chatted a little bit about that yesterday, as well.
- Oh yeah, for sure.
In a way that's playful and provocative.
So, I think it's whatever computers can't do
until they can.
And this was my former colleague, Hillary Mason,
first introduced me to this phrase,
I think it dates back to a science fiction writer.
And I love it because the tech is changing so quickly,
it actually goes through agility.
We can't define it precisely, because tomorrow,
it's gonna be able to do stuff that it couldn't yesterday.
So we just have to bake all of that progress
into the definition, and recognize that
we're always in this limbo, related to our own
superiority or specialness as humans.
It's kind of adds humility.
- And in closing, if there's one place
to start, where would you recommend?
An enterprise start.
- So what's interesting, I was talking with your colleague
about this, and I think the real value that AI can provide
is it actually empowers the customer to shape
their experience of the business,
as opposed to just sort of receiving what's been
given to them.
So then you say, where do you start to get there?
I think it starts with picking existing business metrics,
finding one that where if you were to modify
it a little bit, it could lead to huge returns.
So you'd only have to get little wins,
to have big results.
And that helps get executive buy in,
that helps people go from oh there's this thing in
the research lab, to like whoa, this is really valuable
to my business.
And then from that you can get into some of the more
complex things around personalization,
and lifetime value optimization.
But it starts with a tweak on what's happening today
in a meaningful way using data.
- Wonderful, great way to end it.
Thanks everyone for joining,
please check us out at hashtag AdobeTT.
Thanks again.
Không có nhận xét nào:
Đăng nhận xét