Artificial Intelligence
-Ruchika Pareek
Two
young
innovators
from
Michigan
are
making
headlines
in
the
AI
sector.
They
refused
a
multimillion-dollar
offer
from
Elon
Musk,
focusing
instead
on
their
groundbreaking
project.
William
Chen
and
Guan
Wang,
the
brains
behind
Sapient
Intelligence,
met
in
high
school.
They
shared
ambitious
“metagoals.” For
Wang,
it
was
crafting
a
problem-solving
algorithm;
for
Chen,
it
was
enhancing
systems
across
various
applications.
“One
day,
we’re
going
to
have
an
AI
that’s
smarter
than
humans,” Chen
told
Fortune.
“Guan
and
I
always
say
it’s
like
Pandora’s
box.
If
we’re
not
going
to
make
it,
someone
else
will.
So
we
hope
that
we’re
going
to
be
the
first
one
to
make
that
happen.”
William
Chen
and
Guan
Wang,
co-founders
of
Sapient
Intelligence,
developed
the
Hierarchical
Reasoning
Model
(HRM),
an
AI
model
that
demonstrated
superior
performance
in
abstract
reasoning
tasks,
notably
outperforming
systems
from
OpenAI,
Anthropic,
and
DeepSeek
in
June
with
a
27-million-parameter
prototype.HRM’s
architecture,
differing
from
traditional
transformers,
uses
a
two-part
recurrent
structure,
mimicking
human
thinking,
and
the
company
plans
to
establish
a
US
office.
After
high
school,
Chen
and
Wang
attended
Tsinghua
University
in
Beijing.
Although
initially
challenged
by
rigorous
coursework,
they
soon
gained
faculty
support
for
their
ambitious
AI
project.
Their
efforts
led
to
the
creation
of
OpenChat,
a
small
large-language
model
(LLM)
known
for
its
ability
to
improve
through
reinforcement
learning.
This
innovation
earned
significant
recognition
in
academic
circles.
AI
Model
Innovation
and
Recognition
OpenChat’s
success
didn’t
go
unnoticed.
Elon
Musk’s
xAI
approached
the
students
with
a
lucrative
offer,
which
they
declined.
This
pivotal
decision
paved
the
way
for
Sapient
Intelligence
and
their
new
AI
model,
the
Hierarchical
Reasoning
Model
(HRM).
This
model
outshines
other
major
AI
systems
in
tasks
requiring
abstract
reasoning.
In
June,
their
HRM
prototype,
with
only
27
million
parameters,
outperformed
systems
from
OpenAI,
Anthropic,
and
DeepSeek.
It
excelled
in
complex
challenges,
such
as
advanced
Sudoku
puzzles,
maze-solving,
and
the
ARC-AGI
benchmark.
“It
was
crazy,”
Chen
said.
“Just
with
a
change
in
the
architecture,
it
gave
the
model
a
lot
of
what
we
call
reasoning
depth.”
Revolutionary
AI
Architecture
Unlike
traditional
transformers
that
rely
on
statistical
predictions,
HRM
uses
a
two-part
recurrent
structure.
This
design
mimics
human
thinking,
combining
deliberate
reasoning
with
quick
reflexive
responses.
“It’s
not
guessing,”
Chen
said.
“It’s
thinking.”
Their
model
also
experiences
fewer
hallucinations
than
typical
LLMs
and
matches
top
performance
in
fields
like
weather
prediction,
quantitative
trading,
and
medical
monitoring.
Sapient
Intelligence
is
set
to
expand
with
plans
to
establish
a
US
office
soon.
Their
journey
from
high
school
friends
to
AI
pioneers
highlights
a
commitment
to
innovation
and
a
drive
to
push
AI
boundaries
beyond
conventional
limits.
