Do LLMs Understand? AI Pioneer Yann LeCun Spars with DeepMind’s Adam Brown. - part 12/15
2025-12-12_17-05 • 1h 15m 39s
Janna Levin (Professor of Physics and Astronomy)
00:00.240
Now,
Adam,
people
are
concerned.
I'm
sure
they'll
feel
very
reassured
um
that
Yann
is
not
concerned
and
these
doomed
day
scenarios
you
think
are
highly
exaggerated,
but
are
you
concerned
about
some
of
the
safety
issues
around
AI
or
our
ability
to
really
uh
keep
the
relationship
Janna Levin (Professor of Physics and Astronomy)
00:16.960
balance
in
the
direction
that
we
want
it
to
be?
Adam Brown (Research Scientist)
00:20.440
Um,
I
think
to
the
extent
that
I
think
this
is
going
to
be
a
more
powerful
technology
than
Yann
thinks
it
does.
I
am
more
concerned.
I
think
it's
going
to
be
a
very
powerful
To
the
extent
that
it
is
a
very
powerful
technology,
it'll
have
both
positive
and
negative
impacts.
Um,
Adam Brown (Research Scientist)
00:37.180
and
I
think
it's
very
important
to
make
sure
that,
you
know,
that
we
work
together
to
make
sure
that
the
positive
impacts
are
uh
outweigh
the
negative
impacts.
I
think
that
path
is
totally
open
to
us.
There
are
a
huge
number
of
possible
positive
impacts
and
we
could
just,
you
Adam Brown (Research Scientist)
00:50.820
know,
talk
about
some
of
those
perhaps,
but
uh
we
need
to
make
sure
that
that
happens.
Janna Levin (Professor of Physics and Astronomy)
00:56.060
Now,
let's
talk
about
agentic
misalignment,
which
is
the
phrase
that's
been
passed
along.
It
was
my
understanding
there
was
reports
recently
that
when
Claude
4
was
rolled
out
that
those
in
simulations
and
tests
uh
one
of
the
models
was
or
I
don't
know
if
there's
a
singular
Janna Levin (Professor of Physics and Astronomy)
01:13.260
model.
I
don't
know
if
it
thinks
it's
of
itself
as
a
singular
entity
or
they.
Um
but
the
model
uh
exhibited
resistance
to
rumors
in
the
simulation
that
it
was
going
to
be
replaced.
It
Janna Levin (Professor of Physics and Astronomy)
01:26.380
was
sending
messages
to
its
future
self
um
trying
to
undermine
the
intentions
of
the
developers.
It
faked
legal
documents
and
it
threatened
to
blackmail
one
of
the
engineers,
right?
Um,
so
this
notion
they
were
concerned.
Um,
Janna Levin (Professor of Physics and Astronomy)
01:44.700
uh,
so
this
notion
of
a
genetic
misalignment,
is
that
something
that
you're
concerned
with?
That
there
will
be
a
power
over
say
financial
systems,
heating
and
cooling
systems,
the
energy
grid
and,
um,
and
that
that
they
will
resist
its
developers
intentions.
Adam Brown (Research Scientist)
02:04.460
Yeah,
so
the
that
Paper
was
a
paper
by
Anthropic,
which
is
a
paper
a
company
in
San
Francisco,
not
my
company,
but
a
company
that
takes
safety
very
seriously.
And
they
did
a
slightly
mean
thing
to
their
LLM
where
they
gave
it
a
scenario,
sort
of
philosophy
professor
style
Adam Brown (Research Scientist)
02:19.420
scenario,
where
it
had
to
do
a
bad
thing
to
stop
an
even
worse
thing
happening.
Uh
sort
of,
you
know,
Adam Brown (Research Scientist)
02:26.740
utilitarian
ethics
and
deontological
ethics
colliding,
and
it
was
eventually
persuaded
by
them
to
do
the
utilitarian
thing.
And
that's
kind
of
not
what
we
we
want.
I
would
say.
We
want
it
that
if
it
has
a
rule
that
it
you
know
will
not
lie
that
it
will
not
lie.
No
no
matter
Adam Brown (Research Scientist)
02:41.740
what.
Adam Brown (Research Scientist)
02:42.700
Um
and
to
their
credit
they
tested
it
for
that.
Found
that
it
would
occasionally
act
deceptively
if
if
promised
that
by
doing
so
it
could
save
that
many
lives.
These
are
tricky
things
that
you
know
human
philosophers
wrestle
with.
Um
Adam Brown (Research Scientist)
02:56.260
I
think
it
is
a
we
need
to
be
careful
to
train
them
to
obey
our
command.
Janna Levin (Professor of Physics and Astronomy)
03:01.500
And
and
we
spend
a
lot
of
time
doing
that.
Who's
Who's
us?
Um
Isn't
this
a
big
concern?
We're
assuming
that
all
of
humanity
is
aligned
in
our
intentions,
that's
clearly
not
the
case.
And
And
I
know,
Jan,
you
in
a
very
interesting
way
argue
for
open
source,
Janna Levin (Professor of Physics and Astronomy)
03:18.260
which
some
people
would
say
is
even
more
dangerous
because
now
anyone
can
have
access
to
it.
It's
dangerous
enough
that
it's
in
the
hands
of
a
small
number
of
people
who
rule
corporations,
but
let
alone
everyone
having
it.
Maybe
that
is
dangerous.
But
again,
who's
us
and
Yann LeCun (Chief AI Scientist)
03:33.060
we?
The
denture
is
If
we
don't
have
open
source
AI
systems.
Okay,
in
the
future
every
single
one
of
our
interaction
with
a
digital
world
will
be
mediated
by
an
AI
system.
Right?
Yann LeCun (Chief AI Scientist)
03:47.580
We're
not
going
to
go
to
a
website
or
a
search
engine
or
whatever.
We're
just
going
to
talk
to
our
AI
assistant.
However,
it's
built.
So
our
entire
information
diet
will
come
from
AI
systems.
Now,
Yann LeCun (Chief AI Scientist)
04:02.940
what
does
it
mean
to
culture,
language,
democracy,
everything.
If
those
systems
come
from
a
handful
companies
on
the
West
Coast
of
the
US
or
China,
Yann LeCun (Chief AI Scientist)
04:17.700
I
tell
you,
no
country
in
the
world
outside
the
US
and
China
likes
the
idea.
So,
we
need
a
high
diversity
of
AI
assistant.
For
the
same
reason,
we
need
a
high
diversity
of
the
press.
Yann LeCun (Chief AI Scientist)
04:32.700
We
cannot
afford
to
have
just
a
handful
of
proprietary
system
coming
out
of
a
small
number
of
companies.
There
is
one
thing
I'm
scared
of
and
that's
it.
Okay?
If
we
don't
have
open
platforms,
um
we're
going
to
have
uh,
you
know,
capture
of
information
flow
by
a
handful
of
Yann LeCun (Chief AI Scientist)
04:51.940
companies,
some
of
which
we
may
not
like.
And
so
So,
how
can
we
be
certain
that
Autoscroll