(music)
- I couldn't, truly couldn't imagine
a more interesting and better guest than Galen Buckwalter.
Galen was the inventor of the matching algorithm
at eHarmony, the dating site,
which was the very first dating site in the world
using an algorithm to match people.
He's today the Chief Science Officer at psyML,
and we'll talk a whole bunch about psyML today,
where he's using psychological assessments
to bring machine learning and robotics
in through truly their human realm.
Galen, I'm so stoked to have you here.
- Ah, pleasure
- You've spent like literally your whole career
on trying to figure out how people tick, right?
And you've done this at eHarmony,
and we'll talk a little bit about
the dating space a little bit later,
but what are, there's this concept of psychometrics
which is really like your life's work, right?
- Absolutely - What are psychometrics
and how do we use them in a technology sense?
- Psychometrics, literally, is
the process of applying numbers to human behaviors.
So when we know that as people we have traits,
like intelligence, like openness, like we have emotions,
all of these exist, but if we wanna study them
we have to be able to measure them.
The process of quantifying these latent traits
is a very, kind of arduous and systematic process.
But basically it boils down to,
you hypothesize something that exists,
then you collect evidence of how it exists internally,
with the measurements, the items that you're using
and then how it relates to other
constructs that you hypothesize.
So ultimately you get to a position where,
you know you say it looks like a duck, it quacks like a duck
you know you have anger, you have openness
or some personality trait.
So that's the process of psychometrics.
Applying it online has, to me, always seemed
like a gift to psychologists.
Because online we're able to, for the first time,
apply psychometrics at scale.
The biggest problem in academic research
around human traits, has been simply
that you know you look at studies, of even 20 years ago
you were surprised if you saw an N of 200 subjects.
And that probably took the poor dissertation student,
you know a couple of years to collect.
But now, with the internet, with crowd sourcing sites
like Mechanical Turk, we're able to get thousands of people
to take questionnaires overnight.
Which has sped up the pace
of personality and emotion research just exponentially.
So it's really an exciting time to uh--
- Interesting, let me ask a question,
so when we measure, particularly a concept like personality,
in your view, and I think this is at the age old question,
how much of personality is stuck,
is like, you know it is who we are,
and it will pretty much always be who we are,
and how much of it is a fluid concept
which changes pretty dynamically
in terms of like, you know my evolution as a being?
- Yeah, yeah, I mean it's a great question
and it's one we're still figuring out.
There are clearly dimensions that change
minimally across the lifespan, but there's always change.
Particularly around periods of
great kind of relationship and social change,
so you know we see major personality shifts
like around the time when young people go to college.
You know a great increase in openness,
you know kind of exploration behaviors.
You also see changes around the time
that people get into long term relationships.
So you probably see a common theme there,
that personality changes most in response to relationships.
Which tells you a lot about our brain, I think.
You know that, our brain, our personality,
I think it forms fairly early on, by seven or eight.
It also forms around six dimensions
that we've only recently really nailed down.
- That's the HEXACO model, right?
- That's the HEXACO model.
- Can you explain this a little bit?
- Sure, you know over the generations,
there's been, psychologists have always
had theories about personality.
You know Freud had the theory of subconscious,
and the unconscious, you know Young had the archetype,
and then there was also a very empirical aspect
of psychology that looked at personality,
and using advanced analytics, primarily factor analysis.
Interestingly, in the late 80s, early 90s
both of those converged into
what was initially called the OCEAN Model
or the Five Factor Model,
which was openness, conscientiousness,
extraversion, agreeableness, and neuroticism.
And really the past five to ten years,
a sixth dimension has been added, honesty.
And additionally, we've shifted from using
the somewhat pejorative term, neuroticism,
to calling that dimension emotionality.
So now we have these six dimensions,
which really, for the first time,
give us a rock solid understanding of personality.
I mean, this exists everywhere.
Every culture, both genders, it's really solid,
it gives us something to start to look at
how much we change over the course of lifespan.
Again, we see subtle movements in these,
but by and large, you know, the general structure
of our personalities stays the same.
So you know in terms of online applications,
that's where at psyML, we're really excited
to not just measure people for academic research purposes,
but to use that in applied settings.
My partner, Dave Herman and I
had worked at a financial services company,
called Payoff before we started psyML,
and there we were able to understand
how much credit risk was related to personality factors.
That you can just use the HEXACO.
- It's also a little scary, right?
In the shoes of like, the person
applying for credit, for example.
- If handled in the wrong way, I think it could be.
What our approach is, is that we only use it
in the context of self-insight.
So we're not going to be using this
in a way that people don't understand.
And we also think that in the context of self-insight,
people are able to then start
taking control of their personality.
They understand it, they can better understand
why they have problems with credit.
And then they can work, kind of in conjunction
with their personality, to do what it takes for them.
That's a very critical point about applying this research,
is for psyML, we envision our AI
to ultimately be empty engines,
that you know we know a lot of what's good about humanity.
We know that gratitude is hugely powerful,
and hugely beneficial to our very brain.
So if we can start using this information
and helping, using machines to help us understand
and advance our own empathy abilities,
our own ability to be grateful,
that's when it seems like it's a very positive thing.
- But you're applying this also to robotics, right?
The plan is to like make robots more human?
- Very much, you know we think,
like right now we're using it to just to,
one instance that we're using this technology
is with people who have abused drugs
and are trying to stay off of them.
So we have a bot that our brilliant developers,
Grace Gee and Eugene Wang, shout out,
that they have developed these AI systems
that are capable of interacting with people
that are experiencing desires to use drugs
in a very empathetic, understanding way.
In addition to understanding personality,
we're putting a lot of effort into understanding emotion,
which is our short term mood state.
So no matter what our personality is,
you know if something really bad happens,
we're going to be sad, or we're going to be angry.
And that's going to influence everyone, how they react.
So we wanna be able to understand
someone's long term characteristics, their personality,
but then we react in the moment also,
based on their emotions.
So with the drug abuse prevention program,
we check in every morning with the people
to get a read on whether they're feeling angry or happy,
what their energy level is,
because sometimes with drug abuse
if you're feeling really high and good, that's a risk.
So based on all that information,
then we react, we give people exercises to do,
basic cognitive behavioral therapy approaches.
But throughout all of this, the machine is learning.
So we think that giving the machine this understanding of us
this understanding of people
as growing through self understanding,
through empathy towards others,
that puts a very different light on AI.
I don't think, if Elon Musk was thinking about AI
in this context, would he be afraid of it?
- Well, I also can totally see the applications
if we're thinking about particular machines
in the form of physical machines, robots,
becoming much more embedded in our lives
we want them to be, likely we want them, not all of them,
like the cooking robot doesn't need to be empathetic I guess
probably he needs to, because if I feel down
he should cook me, it should cook me comfort food, right?
But I can totally see the implication,
I'm really excited about your work.
Before I have my last question for you for this interview,
where can people learn more about your work, psychML,
how can they probably get involved?
- Our site is at psyML.co, dot C-O.
We're developing that quite rapidly now,
so you'll start to see that.
I have a Medium page, at Galen Buckwalter
you can look up what we're writing there.
- Okay, so psyML.co, and then check out Medium.
My very last question, and I'm sorry we have to go there,
because you brought the world eHarmony.
With online dating has gone through so many iterations,
we had eHarmony, and then we had Match,
and then we had like Plenty of Fish,
and now we've got Tinder.
Do you think it's today,
the way to actually find your spouse,
and full disclosure, I found my spouse the old way
and I'm very, very happily married
to like the most amazing woman on the planet,
but do you think on average, the machines are better
to find our perfect match than,
you know like the old fashioned way?
- Yeah, yeah, well there is some data starting to show up.
There was a large scale study published in
the Proceedings of the National Academy of Science
just about two years ago I think,
that looked at eHarmony, they looked at Match,
they looked at you know your method of
dating in the wild, as we call it,
but what they found was that eHarmony,
the divorce rate was significantly lower than Match,
but Match was also significantly lower than the wild.
So it seems like just the process of
looking at your potential partners
in a more kind of systematic way,
I mean marriage is not, I mean marriage requires
the passion and the kind of the chemistry,
but if you rely on that,
that is not a good start for a marriage.
And I think in the wild, that's a temptation.
So yeah, I think there can be advantages
to a more systematic exploration
that online dating kind of enhances.
- Okay, and I have to ask you a very last, cheeky question.
Where did you meet your wife?
- The old fashioned way, in a class I was teaching.
- Fantastic, I love it.
So all of you guys, get out there, find your match.
If you don't find your match out there,
use apps, and definitely, definitely keep an eye on psyML
and Galen's incredible work.
I truly believe your work will influence and change
the way we think about machines interacting with us,
making them truly more human.
(light electronic music)
For more infomation >> Spanish ASMR. Liquid massage, ear to ear (Guy) - Duration: 20:12. 
For more infomation >> 🔴LIVE Minecraft Story Mode Season 2 Episode 3 | Jailhouse Block Sub goal 1000 - Duration: 1:26:30. 
For more infomation >> Ico HD | Part 1 - No commentary [Full Eng] - Duration: 43:23. 
Không có nhận xét nào:
Đăng nhận xét