Cathy O'Neil: The era of blind faith in big data must end
Cathy O'Neil: La era de la fe ciega en los datos masivos ha de terminar
Data skeptic Cathy O’Neil uncovers the dark secrets of big data, showing how our "objective" algorithms could in fact reinforce human bias. Full bio
Double-click the English transcript below to play the video.
the winners from the losers.
de los perdedores.
se les invita a una entrevista
that we don't understand
secretas que no entendemos
and often hoping for.
by looking, figuring out.
mirando, descubriendo.
what is associated with success.
se asocia el éxito,
in written code.
mediante un código escrito.
to make a meal for my family.
para preparar la comida en casa.
of ramen noodles as food.
de fideos como comida.
if my kids eat vegetables.
si mis hijos comen verdura.
from if my youngest son were in charge.
si mi hijito tuviera el control.
he gets to eat lots of Nutella.
mucha Nutella.
most people think of algorithms.
se imagina los algoritmos.
and true and scientific.
objetivos, verdaderos y científicos.
y temer los algoritmos
las matemáticas.
blind faith in big data.
confiamos a ciegas en datos masivos.
She's a high school principal in Brooklyn.
de una escuela de Brooklyn.
her teachers were being scored
sus maestros se clasificaban
what the formula is, show it to me.
cuál es la fórmula, muéstremela.
to get the formula,
"Trate de conseguir la fórmula,
told me it was math
de Educación me dijo
que no la entendería".
a Freedom of Information Act request,
Ley de Libertad a la Información.
and all their scores
y su puntuación
as an act of teacher-shaming.
avergonzar a los maestros.
the source code, through the same means,
código base, usando el mismo mecanismo,
had access to that formula.
a la fórmula en Nueva York.
got involved, Gary Rubenstein.
inteligente, Gary Rubenstein.
from that New York Post data
los datos del New York Post
a un maestro.
for individual assessment.
para evaluar a una persona.
with 205 other teachers,
recommendations from her principal
de la directora
of you guys are thinking,
datos, los expertos en IA
the AI experts here.
an algorithm that inconsistent."
un algoritmo tan inconsistente."
with good intentions.
sin querer.
that's designed badly
silently wreaking havoc.
provocando un desastre silenciosamente.
about sexual harassment.
acoso sexual.
to succeed at Fox News.
tener éxito en Fox News.
but we've seen recently
pero hemos visto que hace poco
to turn over another leaf?
their hiring process
de contratación
aprendizaje automatizado?
21 years of applications to Fox News.
21 solicitudes recibidas por Fox News
stayed there for four years
hubiera estado alli unos 4 años
to learn what led to success,
que logra el éxito.
historically led to success
llegaron al éxito
to a current pool of applicants.
who were successful in the past.
hayan tenido éxito en el pasado.
blindly apply algorithms.
if we had a perfect world,
don't have embarrassing lawsuits,
no estan involucradas en litigios,
de esas empresas
it means they could be codifying sexism
y así podríamos codificar sexismo
all neighborhoods
todas las ciudades y los barrios
only to the minority neighborhoods
solo a barrios minoritarios
we found the data scientists
cientificos de datos
where the next crime would occur?
dónde ocurrirán los próximos delitos.
criminal would be?
el próximo criminal.
about how great and how accurate
de su grandeza y de la precisión
but we do have severe segregations
pero tenemos grandes segregaciones
and justice system data.
legislativos sesgados.
the individual criminality,
recently looked into
lo estudió hace poco.
during sentencing by judges.
al hacer sentencias judiciales.
was scored a 10 out of 10.
sacó una puntuación de 10 de 10.
3 out of 10, low risk.
3 de 10, bajo riesgo.
for drug possession.
the higher score you are,
a mayor puntuación
a longer sentence.
una sentencia más larga.
technologists hide ugly truths
ocultar verdades feas
de algoritmos
important and destructive,
importantes y destructivos
and it's not a mistake.
building private algorithms
que construyen algoritmos privados
for teachers and the public police,
de los maestros y la policía pública
instituciones gubernamentales.
the authority of the inscrutable.
autoridad inescrutable.
since all this stuff is private
ya que todo esto es privado
will solve this problem.
podrá solucionarlo
to be made in unfairness.
con la injusticia.
económicos racionales.
in ways that we wish we weren't,
de una forma que no quisiéramos,
have consistently demonstrated this
lo han demostrado consistentemente
de empleo
of applications to jobs out,
pero algunas con apellidos blancos
have white-sounding names
the results -- always.
decepcionan, siempre.
into the algorithms
a nuestros algoritmos
about ramen noodles --
en los fideos--
picking up on past practices
basados en prácticas pasadas
to emerge unscathed?
algoritmos emerjan intactos?
we can check them for fairness.
interrogados,
the truth every time.
We can make them better.
Y mejorarlos.
del algoritmo,
integridad de datos.
algorithm I talked about,
we'd have to come to terms with the fact
implicaría una conciliación
smoke pot at the same rate
negros fuman marihuana
to be arrested --
los arresten
depending on the area.
veces más dependiendo de la zona.
in other crime categories,
otras categorías criminales,
the definition of success,
en la definición del éxito,
algorithm? We talked about it.
de la contratación?
and is promoted once?
y asciende de cargo una vez?
that is supported by their culture.
apoyado por la cultura.
the blind orchestra audition
orquesta de ciegos
are behind a sheet.
detrás de la partitura.
have decided what's important
decide lo que es importante
distracted by that.
auditions started,
de orquesta de ciegos
went up by a factor of five.
un factor de cinco veces.
for teachers would fail immediately.
del valor añadido fallaría.
the errors of every algorithm.
errores de cada algoritmo.
and for whom does this model fail?
y con quiénes falla?
de los algoritmos,
que engendran.
had considered that
de Facebook lo hubieran considerado
only things that our friends had posted.
publicadas por nuestros amigos.
one for the data scientists out there.
uno para los científicos de datos.
not be the arbiters of truth.
ser los árbitros de la verdad.
of ethical discussions that happen
discusiones éticas que ocurren
for our algorithmic overlords.
a los lores de los algoritmos.
in big data must end.
datos masivos debe terminar.
ABOUT THE SPEAKER
Cathy O'Neil - Mathematician, data scientistData skeptic Cathy O’Neil uncovers the dark secrets of big data, showing how our "objective" algorithms could in fact reinforce human bias.
Why you should listen
In 2008, as a hedge-fund quant, mathematician Cathy O’Neil saw firsthand how really really bad math could lead to financial disaster. Disillusioned, O’Neil became a data scientist and eventually joined Occupy Wall Street’s Alternative Banking Group.
With her popular blog mathbabe.org, O’Neil emerged as an investigative journalist. Her acclaimed book Weapons of Math Destruction details how opaque, black-box algorithms rely on biased historical data to do everything from sentence defendants to hire workers. In 2017, O’Neil founded consulting firm ORCAA to audit algorithms for racial, gender and economic inequality.
Cathy O'Neil | Speaker | TED.com