miércoles, 31 de octubre de 2018

Beyond digital skills

Many times in the past I have written about the need to promote digital skills. In particular, I think that it´s highly important to provide policy makers with the digital skills (most red post of this blog)  they lack of. However, I think that once more it´s needed to think out of the box and we need to go beyond the trending topics. Although digital skills are needed to work and joy in the digital era, they are not the only skills that are needed for the digital era.

Particularly, ethical skills should be in equal foot of importance as digital skills in any skills development strategy, wether in the national/international level or in an education centre. Digital technology applications are colliding with the principles and values we wre taught in many areas every day. The crash could only be avoided if "Don´t be evil" motto goes beyond being a catchy phrase, continuouslly forgotten even by the digital company who promoted it, and begin to be infused in engineers DNA.

Although many times is tried to be hidden, engineers and digital companies have a responsibility in the consequences of their invention. In the past, IBM played a central role as an enabler of the holocaust promoted by the nazi regime. Today, Amazon, Microsoft and others are providing the tools for the implentation of Trump´s anti-inmigration agenda

It´s not possible to stop the bad usage of a technology, but it´s possible not to develop this bad usage. Killer robots will be used for war it they are developed, but its development could be stopped. In the same manner, technology could be used for good purposes as warning about the existance of gender divide at work. But an ethical approach to tech development it´s only possible if ethics are included as a core matter in enginneer training.

Digital skills are needed for humans in the digital world. Ethic skills are critical for the existance of a digital world.

miércoles, 17 de octubre de 2018

No "data sovereignty" without "tech sovereignty"

A couple years ago, when the GDPR debate was in it last stretch, "data sovereignty" was one of those buzz words constantly repeated. Behind the mentioned concept is the idea that data sgould be subjected to the laws and governance structures within the nation it is collected. "Data sovereignty" was finally one of the pillars of the GDPR in spite of the opposition of US government and companies.

The idea of "data sovereignty" has taken roots in EU regulation and be extended beyond personal data. In the institutional agreement reached around the free-flow of non personal data regulation, the right of public authorities to access data for scrutiny and supervisory control wherever it is stored or processed in the EU was enshrined. Now, it looks that a new e-evidence regulation will provide EU authorities some kind of right to access to data for law enforcement purposes wherever is stored in the cloud.

Although EU "data sovereignty" is gaining momentum both for personal and non-personal data, it could loose the battle in the long term without the development of a deterrence force. This deterrence force is High Performance Computing (HPC) capabilities. It´s hard to imagine maintaining a strong EU authorities control on data processing and storage without a strong base of HPC. And this is not the situation now, while EU provides about 5% of supercomputing resources worldwide but consumes one-third of them. 

It is quite depressed that the EU looks behind in all the central pillars of the digital future, although as in many other cases in HPC it looks that now the EU is trying to join forces. Maybe the EU will recover some positions or maybe not, but we perhaps should start to forget about how to jump forward in a concrete ranking and readjust the complete framework that makes the EU lags behind in all the fields of the digital race. Otherwise we should forget about any kind of tech sovereignty.

miércoles, 10 de octubre de 2018

Foundations for an AI strategy

The technological hype now is called Artificial Intelligence. Whether you are part production or the consumer side of the digital economy, it is almost certain that you will read, listen or watch some kind of piece of new or report about AI in the next days, and in the following ones, and in days after. The references to AI look pervasive. However, It is curious that according with Google Trends, the term "Artificial Intelligence" is used and mentioned far less than in 2004.

Whether or not AI debate is more central now to the technological debate than before, many interesting studies are published continuously about its long and medium term impact. One of these studies was published some weeks ago by McKinsey Global Institute (MGI). The study concludes that AI has the potential to deliver additional global economic activity of around $13 trillion by 2030, or about 16 percent higher cumulative GDP compared with today.

However, AI adoption might be not equal across territories, companies and people. The study mentioned above warns about the potential emerge of a triple divide due to  this unequal adoption. On one hand, the countries leading AI development could capture an additional 20-25% in economic benefits compared with today, while emerging economies may accrue only an extra 5-15%. On the other hand, by 2030 the AI frontrunner companies will double their cash flow while the laggard ones might experience around a 20 percent decline in their cash flow. Finally, the job profiles characterized by repetitive activities could experience the largest decline as a share of total employment to around 30 percent by 2030, from some 40 percent.

What the MGI report does not analyse in deep are the causes for the triple divide. These causes will not reside in the future, the AI divides will be the corollary of the present breaches in the adoption of the digital technology. 

To begin with the geographical divide, AI cannot be adopted if you do not have access to it. Therefore, the precondition for an equal AI development in every country and region is having an homogeneous level of connectivity worldwide. Unfortunately, we are far from it. The UN Broadband Commission has been warning of the massive lack of broadband access in the developing countries, estimating that half of the world population does not have internet connectivity.

The case for the companies AI divide is slightly difference. Almost every company has some kind of Internet. But broadband networks are not the only required infrastructure on the digital world which economy is based on data. The adoption of AI is almost impossible for those companies that do not have access to storage and processing capabilities. And the adoption divide of the cloud services which provide these computing capabilities is already there. While almost 45% of large companies  in Europe has adopted cloud services, less than 20% of SMEs has done so.

Finally, the seeds for the future workers AI divide is the current difference in digital skills adoption. More than 40% of EU workers lack of the basic digital skills. There is few risk on a bet on these workers as the future losers of the AI revolution.

There is a frenzy among countries, particularly the economy leaders, in developing AI strategies to obtain its undeniable benefits. However, these strategies, even in these countries could be doomed. The lack of solid foundations ruins even the more splendid mansion and AI could not flourish without the right and previous deployment of connectivity, storage and processing infrastructures and digital skills.

miércoles, 3 de octubre de 2018

Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy - Cathy O'Neil

Weapons of Math Destruction: How Big Data Increases Inequality and Threatens DemocracyWeapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy by Cathy O'Neil
My rating: 4 of 5 stars

This is a must read book to understand our daily life and how it is mastered by algorithms and maths in a growing manner. But do not fear to meet with a complex text, the book is written in a readable style and you will almost read in a day or two if you are on holidays.

The author calls the algorithms used by companies "Weapons of Math Destruction (WMD)", characterized by its opaciity, scalability and damage. Obviously, not all the algorithms are WMDs, but there are many cases where they are. In the book, you can find many examples of cases where maths are weaponized through bias not only towards vulnerable minorities but towards almost every citizen, except the more privilegized.WMDs are now part of our all lifecycle, dominating education, hiring, evaluation of work performance, insurance and finance sector and, of course, our daily usage of the internet.

However, the book is not just another neoluddite piece. To begin with, the author is an specialist on data analysis who has worked many years on the matter. The author pointed to the bad usage of maths, hihlighting how algorithms could be a useful tools if we introduce human values in its application. The central perversion on our usage of math is the focus on efficiency, laying aside fairness and defining a paralel reality which perpetuates and reinforce the existing discriminations.

Nevertheless, the book is also a call for hope. In may passages of the book the author reminds us our own responsibility. Particularly, it points to those who work in the digital industry. Furthermore, it includes a kind of Hippocratic Oath for them in its final pages. In the end, reverting the situation is sill on our hands.

View all my reviews
palyginti kainas