►
Description
Date: 11/06/20
Presenter: Dr. Elieu Huerta
Institution: Illinois University
Title: "Training the Next Generation of AI Practitioners"
http://sbdh-prod.ideas.gatech.edu/resources/newsblog/education-and-workforce-working-group
A
So
the
center
has
three
main
components,
and
that
is
why
there
are
three
gentlemen
working
on
this
project
and
for
you
that
have
been
working
on
data
science.
You
know
very
well
that
we
must
not.
We
should
not
disentangle
academia
from
industry
and
innovation,
harnessing
what
is
happening
in
ai,
not
only
in
hardware
but
also
in
software,
and
those
three
components
are
represented
by
these
individuals.
A
A
A
A
A
As
you
may
know,
we
are
also
interested
in
harnessing
large
scale
infrastructure
like
supercomputers,
that
I
will
show
in
the
next
slide
now
in
the
center
bottom
center.
You
see
an
image
of
industrial
applications.
A
We
are
not
doing
one
without
the
other
and
we
also
want
to
do
or
establish
a
virtue
cycle
in
which
academic
experts
interact
with
industry
leads.
We
learn
about
their
challenges
and
then
we
incorporate
those
needs
into
our
curriculum
so
that
the
students
go
and
have
internships
with
these
companies.
A
They
apply
their
skills
that
are
top-notch,
but
then
they
realize
that
they
are
not
that
good,
that
they
are
missing
some
skills,
and
so
when
they
go
back,
they
tell
us.
You
know
I
had
some
fun,
but
I
really
don't
know
how
to
do
this.
So
let's
incorporate
that
in
the
curriculum
and
then
we
also
have
a
number
of
initiatives
to
train
students
and
what
we
have
encountered
when
we
select
the
students
to
participate
in
those
research
teams
is
that
they
know
the
theory
of
ai
and
machine
learning
very
well.
A
A
We
have
hands-on
ai
tutorials
every
week
in
virtual
mode
now,
due
to
covets,
and
we
also
organized
through
these
semester,
loan
programs,
we
organized
hackathons
that
are
funded
by
nvidia
and
so
over
the
weekend
we
pose
some
problems
to
the
students
suggested
by
faculty
members,
and
then
we
allow
them
to
go
and
show
off
their
skills.
And
surprisingly,
these
kids
come
up
with
some
really
innovative
solutions
and
they
also
walk
away
with
a
very
fancy
gpu
in
their
hands.
A
So
it
is,
it
is
a
beautiful
program
and
it
is
a
win-win
situation,
because,
as
a
faculty
member,
when
you
want
to
recruit
a
student,
you
want
an
independent
student
and
it
is
through
these
hackathons
that
we
see
a
reality
check.
I
don't
care
about
a
beautiful
cv.
I
want
to
see
your
skills,
and
here
is
how
we
we
get
to
select
students.
A
This
program
has
been
quite
impactful,
and
here
you
have
some
eye
candy
for
recent
accomplishments,
but
let
me
just
go
back
to
reality
check
okay,
when
we
attend
these
conferences
and
we
talk
about
these
nice
accomplishments,
harnessing
ai
and
extreme
scale
computing
for
big
data
experiments.
A
For
the
most
part,
people
do
not
ask
the
right
questions
right.
The
questions
are
not.
Oh,
that's
really
interesting.
How
did
you
train
your
algorithm?
How
do
you
interpret
the
results
of
your
algorithm?
Usually
the
first
reaction
you
get
is,
I
don't
believe
you
right,
and
why
am
I
saying
this?
Well,
it
seems
that
now,
in
addition
to
training
your
students
with
ai
skills,
you
also
need
to
give
them
some
emotional
and
sociological
support
and
there
are
different
ways
to
interpret
this.
A
A
A
So
really
I
don't
see
the
need
of
having
this
struggle.
You
know
I
prefer
to
stay
in
industry
as
opposed
to
going
back
to
academia,
and
I
am
telling
you
this
because
sometimes
in
forums
like
this
there
is
this
common
question.
We
are
losing
students
to
industry.
There
is
a
depletion
of
ai
talents.
How
can
we
solve
that?
A
So
that's
something
that
you
really
need
to
discuss
in
your
department
when
you're
going
to
talk
to
people
who
want
to
incorporate
data
science
in
whatever
field
they
want.
As
the
very
nice
talk
that
we
just
listened
to
or
in
astronomy
and
physics,
whatever
discipline
are
we
really
to
start
incorporating
data
science?
Are
we
socially
sociologically
prepared
for
this
transition
and
that
is
related
to
this
nice
figure?
That
shows
the
amount
of
expertise
available
in
the
planet
to
do
ai
and
there
you
see
the
other
problem.
A
A
Is
this
the
right
solution
for
what
I
am
doing,
or
I
can
solve
that
with
a
traditional
techniques
I
mean
ai
will
not
solve
all
the
problems
and
it
doesn't
have
to.
It
has
to
solve
some
specific
problems
very
well
and
that's
a
win
right.
We
don't
have
to
replace
everything.
We
need
to
be
smart
about
that
and
so
to
try
to
reach
out
to
a
broad
community,
at
least
here
at
university
of
illinois.
As
I
said,
we
provide
a
weekly,
hands-on
training
programs.
A
We
teach
students
how
to
use
hardware
that
is
tailored
for
artificial
intelligence.
We
teach
them
how
to
apply
now
well-known
techniques
that
computer
scientists
use
and
how
we
can
apply
those
into
more
complicated
scenarios.
How
to
process
experimental
data
sets
how
to
interpret
the
results
of
neural
nets.
How
to
incorporate
interpretability
uncertainty
quantification
in
these
models,
and
we
are
also
trying
to
reach
out
to
other
communities.
A
We
had
some
modest
expectations
about
the
reaction
of
this
sector
to
ai
and
they
ended
up
providing
an
overflow
room,
because
you
know
the
appetite
for
this
was
really
big
and
you
know
following
good
practices.
We
provide
all
this
information,
the
the
software
and
the
instructions
to
use
the
models
open
source
so
that
people
don't
have
to
recreate
the
wheel,
and
in
this
upcoming
was
meeting,
we
are
also
doing
this,
providing
access
to
ai
or
in
different
worlds.
We
want
to
democratize
access
to
this
knowledge,
and
this
is
open
to
everyone.
A
If
you
have
time
every
monday
at
11am
central
time,
we
also
encourage
everyone
to
attend
this
ai
seminar
series.
Here.
We
have
experts
across
domains
that
talk
about
how
they
are
using
ai
to
advance
their
research
domain,
and
we
have
experts
not
only
from
the
us
next
semester.
We
will
have
speakers
from
the
alan
turing
institute
in
the
uk
from
the
helmholtz
ai
cooperation
unit
in
germany.
Industry
leads
from
nvidia,
ibm
and
other
companies
that
we
invite.
A
B
Thank
you
so
much.
This
is
definitely
something
that
I
would
co-sign
and
also
heed.
This
is
something
that
we
put
in
the
workbook
back
page
six.
Is
you
need
to
start
meeting
with
people
right
up
front,
because
if
you
wait
to
the
end
you're
you're
sunk
in
these
problems,
because
no
one
will
hear
you
out
unless
they've
been
a
part
of
it
from
the
beginning?
B
So
really
you
know
taking
in
this
idea
that
data
science,
as
well
as
ai,
is
a
is
partially
a
teaching
process,
even
if
you're
in
the
technology,
business
you're
in
the
process
of
teaching
others
no
matter,
you
know
just
by
default,
because
of
so
it's
so
new.
So
it
is
12
exactly
if
people
have
to
go,
they
you
can.
If
you
want
to
stay
or
ask
questions
of
the
speakers,
we
have
a
few
more
minutes.
B
If
you
know
you
know
where
that
is,
if
people
ask
questions
in
the
ether
pad,
if
you're
willing
to
take
a
look
and
put
those
things
in
there,
that'd
be
great
but
I'll
open
it
up
now,
if
anyone
has
any
questions,
but
those
that
have
to
go,
you
know
feel
free
as
well,
and
if
one
of
the
speakers
has
to
go,
thank
you
as
well.
So
are
there
any
questions
for
for
this?
I
I
have
one.
A
B
So
did
you
see
the
link
to
the
etherpad?
It's
in
the
chat?
So
if
you
click
on
that
link,
I
think
people
have
put
in
their
information
where
they
are
so
you
can
reach
out.
If
people
are,
if
you're
comfortable,
putting
your
email
in
there
as
well
for
others,
you
can
do
so.
We
take
those
ether
pads
down
after
the
end
of
today
and
then
we
consolidate
the
information
and
links
into
what
would
be
a
resource
guide.
So
you
can
come
back
to
it
anytime
today
and
take
links
out
or
you
know.
B
We
also
will
link
them
and
the
video
of
the
presentations
on
the
south
big
data
hub
website.
So
we
can
put
that
website
on
the
ether
pad
as
well.
So
it's
just
south
bigdatahub.org
and
every
month
we
will
post
the
presentation,
videos
and
a
link
to
a
cleaned,
or
you
know
taken
taken
down
that
live
ether.
B
B
All
right
well,
thank
you
everyone.
I
did
have
a
question
for
at
least
you
know
when
you're
doing
this,
do
you
have
a
process
like
a
formal
process
that
you
take
people
through?
Have
you
like
a
formal
framework
for
how
to
scope
and
go
through
these
problems
with
ais?
Specifically,
so
I
you
know,
for
instance,
we
talked
about
doing
stakeholder
meetings,
but
do
you
actually
have
them
talk
to
the
students
talk
to
the
stakeholders
and
you
just
try
to
help
them
scope
the
problem
like
walking
them
through
it
yourself.
A
They
start
participating,
and
you
know
this
is
very
effective,
because
these
companies
are
very
excited
about
the
skills
of
these
students
and
they
go
to
the
point
of
saying
you
know:
I'm
really
excited
about
what
you
do,
I'm
going
to
pay
your
gra
for
the
following
year,
and
so
that
way
students
have
more
time
to
participate
on
these
projects
and
they,
you
know
this
is
the
catch.
They
overlap
very
well
with
their
research,
so
they
are
not
distracted
by
working
with
these
companies.
B
Okay,
all
right,
so
you
have
students
that
are
working
in
similar
topics.
Well,
that's
interesting
so
and
we're
I'm
just
interested
in
how,
but
do
you
have
faculty
that
are
participating
in
the
problem?
Scoping
part.
You
know
in
general
to
make
sure
that
they're.
A
That's
right,
yeah,
so
that
is
one
of
the
main
conditions
that
faculty
that
have
teams
are
in
the
in
the
center
and
they
provide
their
expertise,
their
interests,
and
when
there
is
a
topic
that
attracts
their
attention,
then
they
can
participate
and
also
also.
A
Another
important
angle
here
is
when
researchers
are
interested
in
data
science,
but
they
don't
have
the
students,
that
is
when
they
suggest
topics
for
hackathons
right
and
then
the
students
are
performing
really
well,
then
now
they
are
absorbed
by
the
faculty
leads
and
that's
how
they
start
building
up
their
teams.
B
No,
that
makes
sense
you
were
talking
about.
You
know
the
independent
student,
and
so
that's
one
thing
we've
found
is
you
know,
students
are
independent,
but
getting
that
intuition
about
how
what
problems
to
do
and
when
to
do
them
is
the
part
that
takes
that
takes
some
process.
So
that's
that's
where
we've
been
focusing
as
well
to
get
them
to
that
stage
of
independence
after
doing
projects
all
right.
Well,
thank
you.
Thank
you
again
to
the
speakers.
Thank
you
for
giving
your
you're
talking
for
everyone
now
that
we're
able
to
attend.
B
Yes,
they
appreciate
you
for
coming.
I
hope
to
see
you
in
december
and
we
will
look
forward
to
your
feedback.
If
you
have
any
questions
or
suggestions
for
topics,
we
will
be
thinking
about
next
year
and
sometime
in
the
spring.
So
if
there
are
other
types
of
topics,
we're
interested
in
hearing
those
to
think
about
what
speakers
they
want
to
recruit
past
2021,
all
right,
all
right,
next,
we're
actually
out
until
june
of
next
year.