Speech recognition racial bias

02apr20 26-24mar2020

On average, the systems misunderstood

     35 percent of the words spoken by blacks but only

     19 percent of those spoken by whites.

Automated speech recognition less accurate for blacks: study

All five speech recognition technologies had error rates that were almost twice as high for blacks as for whites—even

when the speakers were matched by gender and age and

when they spoke the same words.

Error rates were highest for African American men, and the disparity was higher among speakers who made heavier use of African American Vernacular English.

Hidden bias

The researchers speculate that the disparities common to all five technologies stem from a common flaw—the machine learning systems used to train speech recognition systems likely rely heavily on databases of English as spoken by white Americans. A more equitable approach would be to

include databases that reflect a greater diversity of the accents and dialects of other English speakers.

While the study focused exclusively on disparities between black and white Americans, similar problems could affect people who speak with

• regional and
• non-native-English accents,

the researchers concluded.

If not addressed, this translational imbalance could have serious consequences for people's careers and even lives.

• Many companies now screen job applicants with automated online interviews that employ speech recognition.
• Courts use the technology to help transcribe hearings.
• For people who can't use their hands, moreover, speech recognition is crucial for accessing computers.

https://techxplore.com/news/2020-03-automated-speech-recognition-accurate-blacks.html

Coronavirus: what engineers could do?

02apr2020  26mar2020

... put a virologist, a social scientist and an engineer in the same room (or videoconference), and new ideas to delay transmission and safely allow wider elements of day-to-day life continue could emerge.

No country, has managed to deploy engineering solutions in a way that also helps day-to-day life continue in some safe way (in coronatimes). 

When you look at the potential that engineering can bring to this in

• a public health (preventive) rather than
• a medical (restorative) setting,

it shows how much we're actually missing.

SAGE for coronavirus comprises two groups,

one which draws on the epidemiology of historic pandemic flu,

and one that focuses on the social science of public health. These groups feature experts across medicine, epidemiology and social sciences only.

But engineers? Not likely, it seems. There are no dams here to fix … or are there? Clearly the infection needs to be contained, and allowed to trickle out in a controlled flow in order for the whole system to be managed safely into a less critical state.

That sounds like an engineering challenge to me.

When we look at which countries have done relatively well in tackling the virus, there are signs of (computer) engineering at work. In South Korea it is reported that an app—Corona 100m—helped mobilise crowdsourced information on infections, both possible and actual. In China, part of the "effectiveness" in containing the deaths from the virus is reportedly due to the rapid construction of new hospitals. No country, however, has managed to deploy engineering solutions in a way that also helps day-to-day life continue in some safe way.

Ultimately, we need to design creative, workable and effective responses that better balance protection from the virus with disruption to daily life. Instead, we have scientists with fantastically deep knowledge about the biology and epidemiology of viruses, and maybe about the cognitive tricks that lean us towards doing one action over another.

What engineers could do

Let's pause for a moment and think about what could have been done to help slow the spread of COVID-19, without also shutting down the whole of society, with potentially huge, long-term economic and social consequences.

If virologists can give some insight into the main sources of transmission, could engineers design specific, deployable responses to that?

I'm not an engineer or a virologist, but I study how science, technology and engineering can be used in policy to change the world for the better. So while I don't have the answers, I can start the ball rolling.

What about focusing on the mass production and distribution of on-street hand sanitizer, or gloves treated with new, safe anti-viral coatings?

There may also be new ways of opening doors without grabbing the handle, or indeed pressing lift buttons.

Could we design better protective infrastructure for shop workers facing customers at tills?

What about bring-your-own trolley handles?

Or new kinds of easy-to-make and deploy protective face gear for the elderly and vulnerable?

Engineering solutions would have been especially effective early on during the outbreak, before measures like lockdown were introduced. But even during lockdown, they could help minimize the spread of the virus in the parts of society that are still open, such as banks and supermarkets.

Sat here on my own, I can't solve the problem, but put a virologist, a social scientist and an engineer in the same room (or videoconference), and new ideas to delay transmission and safely allow wider elements of day-to-day life continue could emerge.

When you look at the potential that engineering can bring to this in

• a public health (preventive) rather than
• a medical (restorative) setting,

it shows how much we're actually missing. It may be that these particular (disinfectant) solutions are not workable at scale, but the point is that engineers could probably come up with other design solutions that would work. It's their job.

The problem is, often social scientists just don't speak to or mix with engineers very much. It's a deep-rooted problem, like two parts of a family that fell out years ago over some obscure argument that nobody remembers, but everyone repeats.

I work at the Science, Technology, Engineering and Public Policy department at University College London. We see it as part of our mission to bring scientists and engineers together so everyone can benefit. The coronavirus case study shows now more than ever how much we need that kind of collaboration.

https://techxplore.com/news/2020-03-coronavirus-scientists-solutions.html

Inequality Regimes

02ap2020 26mar20

“Thomas Piketty Takes On the Ideology of Inequality”

‘Every human society must justify its inequalities,’ the book begins. 

What follows is a comprehensive investigation of how different societies have done precisely that, 

ranging through what the book terms various ‘inequality regimes.'”

The tendency in economics now—as well as in a great deal of public discussion—is to

view the economy as a natural force,

existing independently from our ideas about what it is and how it ought to work.

This book systematically demolishes that self-serving conceit by charting in extensive detail

how differently it has operated at different periods of time, and

how its operation is conditioned by the ideologies with which it co-develops.

‘The market and competition,

profits and wages,

capital and debt,

skilled and unskilled workers,

natives and aliens,

tax havens and competitiveness—

none of these things exist as such,’ Piketty insists.

‘All are social and historical constructs’ that ‘depend entirely’ on the ‘systems that people choose to adopt and the conceptual definitions they choose to work with.’…

An exhaustive assessment of Capital and Ideology would require more space and expertise than I have, but the basic contours of the book are easy enough to describe.

“Thomas Piketty Takes On the Ideology of Inequality” [Marshall Steinbaum, Boston Review].

Leituras pendentes 01apr2020

Desemprego estimulado

29/03/2020
24/08/2018

Inverter a espiral de desinvestimento e de desemprego avassalador: é necessário lembrar todo dia que esse desemprego foi programado.

Isso, ainda mais, depois de péssimas notícias, na área de economia, em relação ao desemprego recorde de:
- + de 11 milhões de desempregados;
- sobretudo chef@s de família;jovens;
Situação que só deve se agravar com a expansão da pandemia do Coronavirus.

Já dizia Manu d'Ávila, em 2018

''as medidas anticrise do desgoverno Temer exterminaram com o mercado de trabalho".

Eu digo, mais especificamente, que  "exterminam os trabalhadores e outros precarizados".

Já que os trabalhadores estão cada vez mais:
desalentados de procurar trabalhos.

E aqueles poucos que conseguem serviço,  encontram condições cada vez mais precarizadas com, entre outros aspectos:
- o avanço do trabalho informal;
-  a terceirização irrestrita, inclusive de atividades fins.
- uberização;

O "precariado" ( conceito de Jessé de Souza) precisa estar consciente de que seu  extermínio e degradação são  programados.

Nessa conjuntura devastadora de mais de 11% de desempregados no Brasil, como não me lembrar de que até  2013, o Brasil vivia com desemprego, pelo menos oficial, de menos de 5%.

Lembro-me ainda mais de economistas neoliberais decretarem todo dia  a necessidade de:

- aumentar o desemprego para diminuir o custo da mão de obra.

Você se lembra disso?

comente por favor:
o que pode ser feito para mudar essa desgraça?