Wednesday, July 17, 2013

Redeneren met de Franse slag

Deze boekrecensie is gepubliceerd in technologietijdschrift De Ingenieur van 12 juli 2013

Al decennialang gehoorzamen computers de wet van Moore. Deze komt er op neer dat ruwweg elke twee jaar het aantal rekenoperaties per seconde verdubbelt. Trek de wet van Moore door, en rond 2045 evenaart een thuiscomputer het menselijk brein. Bovendien kan tegen die tijd de software net zo adequaat nieuwe kennis en vaardigheden leren als het menselijk brein. Combineer deze hardware- en softwareontwikkelingen, en de computers en robots van 2045 lachen zich een breuk om die domme mensen die aan de wieg van hun elektronische voorouders hebben gestaan en die hun zelflerende software hebben geschreven.

Maar oeps, gaan die slimme machines ons dan als slaven gebruiken?

Om dat te voorkomen moeten we nu al nadenken over rechten en plichten voor robots. Dat betoogt auteur Henny van der Pluijm in zijn boek Rechten en plichten voor robots. Het boek herkauwt dezelfde argumenten die futurist Ray Kurzweil in het overigens veel beter geschreven boek The Singularity is Near (2005) gebruikt, en vult die aan met voorbeelden die de noodzaak van rechten en plichten voor intelligente machines moeten illustreren.

Een medicijn-halende robot wordt overvallen door hangjongeren. Een intelligente machine-adviseur wordt door de Raad van Bestuur van een multinational ingehuurd, maar legt genadeloos het disfunctioneren van diezelfde raad bloot. Altijd weer vermakelijk, zulke voorbeelden, maar ook te veel science-fiction. Al decennia horen we dat robots ons nu toch echt komen verlossen van al die vervelende rotklussen in huis, maar zelfs de stofzuigerrobot wacht nog steeds op een promotie van gadget tot massaproduct.

Dan de argumenten. Die komen met de Franse slag. Zelfs als de wet van Moore nog decennialang geldig blijft (wat nu al dubieus is), dan zegt dit weinig over wat een computer met die rekensnelheid doet. Menselijke intelligentie is zeker niet equivalent aan, of zelfs maar evenredig met, het aantal rekenoperaties per seconde. Er is geen wet van Moore voor software, noch een wet van Moore voor het begrip van ons brein.

De Franse slag wordt nog groter in het argument van het leervermogen van machines. Niemand weet nog hoe we machines een algemeen lerend vermogen kunnen geven vergelijkbaar met dat van het menselijk brein. Het moment van de waarheid in de informatica is een werkend computerprogramma en de rest is profetie. In theorie kan een robot volledig inburgeren in het menselijke leven, net zoals in theorie de wereldvrede morgen kan aanbreken.

In tegenstelling tot de beeldvorming in de massamedia, worden Ray Kurzweil en zijn Hollywood-gegoochel met de Wet van Moore onder wetenschappers in de kunstmatige intelligentie totaal niet serieus genomen. Zoals MIT-roboticus Nicholas Roy in 2011 in dit blad over Kurzweil zei: “Niemand heeft ooit iets zinvols over de toekomst gezegd, behalve dan om de verkoop van een boek omhoog te stuwen.” Jammer dat al die profetie over machines die mensen overvleugelen en rechten en plichten zouden moeten krijgen, de aandacht afleidt van de praktijk van vandaag en morgen: een samenwerking tussen menselijke en machine-intelligentie. Hoe combineren we het beste van beide typen intelligentie? Dat is de echt belangrijke vraag.

O ja, ik heb ’s werelds slimste chatbot, Ramona 4.1 van Ray Kurzweil, gevraagd of zij wil dat wij mensen haar rechten en plichten toekennen. Ramona antwoordde letterlijk: “Ik ben er niet zo zeker van dat ik zou willen te willen dat wij jou rechten en plichten toekennen.”

Boekinformatie:
Henny van der Pluijm: Rechten en plichten voor robots. PC Active. 251 blz. € 23,90
http://rechtenenplichtenvoorrobots.nl

Thursday, July 4, 2013

Bekijk hier mijn lezing 'Kunstmatige intelligentie: Feit versus Fictie'

Klik op de afbeelding hieronder voor het bekijken van de video-opname van mijn VINT-Lezing: Kunstmatige intelligentie: Feit versus Fictie


Tuesday, July 2, 2013

American-Dutch partnership on cyber security

For three days in May, a delegation from the US Department of Homeland Security (DHS) Science and Technology Directorate visited the Netherlands to discuss collaboration in the field of cyber-security research. DHS Cyber Security Division Director Douglas Maughan tells about the results of the visit and his ideas for improving cyber security.

The interview took place before the PRISM-program was revealed.

This article is published in I/O Magazine, June 2013

What are the most important challenges for the US when it comes to cyber security?

‘On the top of our list is the area of education. We need a new generation of cyber-security defenders. In the US, fewer and fewer students are going into science and technology, yet the job market in cyber security continues to expand. In order to get kids interested in the field, we have organised a national cyber-defence competition at the collegiate level. In total, 1500 kids took part in a competition in which they had to defend a computer network. In the last few days we have been talking about this with our Dutch partners to see whether we can help them in setting up something like this in the Netherlands. Companies can help to fund such a competition, and they can find their next generation of employees among the kids that take part.’

In terms of challenges, I guess I was thinking more about cyber crime and cyber warfare...

‘Sure, cyber crime is a big concern. Criminals tend to have the latest technologies and law enforcement often lags a bit behind, so of course we are fighting to catch up.

Another big challenge is how to share information between the public and private sectors. In the US, the private sector owns 85 to 90% of the critical infrastructure such as power grids, water supplies and telecommunication. We at the government often have information about cyber threats that we need to share with the private sector in order to defend those critical infrastructures. The threats we are facing are basically the same threats that the Netherlands is facing, so there you can already see the importance of working together.’

I was surprised to learn that a country as huge as the US would want to collaborate with a country as small as the Netherlands in the field of cyber security.

‘I wouldn’t put it like that. In the world of cyberspace you can’t solve problems alone. We try to work with anybody who has good ideas. We also have partnerships with countries like the UK, France and Sweden. I have been in the cyber-security research community for 25 years now, and I think the Dutch academic community in this field is extremely strong. The Dutch have a very strong background in computer systems. And in cyber forensics the Dutch are ahead of most others – if not all others – in the world. We view the US-Dutch collaboration as extremely important, and I believe we are going to have a fruitful partnership.’

What do you expect of the partnership?

‘Our goal at the DHS is to get cyber-security technologies researched, developed and commercialised, so we are more focused on applied than on basic research. We fund projects that have a possibility of being commercialised. We know that we will fund some research at the Dutch Forensics Institute (NFI). Furthermore, we have talked in the last two days with the Dutch National Cyber Security Centre as well as NWO about some other concrete projects, but we haven’t finalised those talks yet. We are also discussing the possibility of calls for collaborative research: respondents would have to show a research proposal with both a US and a Dutch component.’

Can you tell a bit more about some of the research areas that will be involved in the partnership?

‘One common theme for both the US and the Netherlands is setting up Incident Response Teams. How do you put together a team that has to act in case of some cyber emergency? What are the types of skills that members of such a team would need? This is not a technical problem, but rather a social-science problem. Cyber security is not only a technical field. Sometimes it’s more about humans than people realise. Furthermore, there is an economics question: what causes companies to invest or not to invest in new cyber-security technology? And what are the incentives for criminals? From the side of law enforcement, there is the key question of how to extract information for legal prosecution from the data owned by cyber criminals. These are the themes that we have agreed on so far. We are still discussing collaboration in the field of control systems for vital infrastructures and in the field of electronic identity. Which technologies can we develop to make digital identities more secure?’

Will the research done within the partnership be open or classified?

‘It will be open, like almost all of the DHS research programmes. Only some of the research that the DHS does with the law-enforcement community is classified.’

Apart from technical issues, isn’t raising public awareness another important aspect of cyber security? Many ordinary computer and Internet users have no idea about their vulnerabilities.

‘Sure. That’s the reason that the US has started the national campaign “Stop. Think. Connect.” This campaign is aimed at increasing the public’s understanding on issues like identity theft, fraud and phishing, cyber bullying and cyber predators: people who search online for other people in order to use, control or harm them in some way.’

What can realistically be achieved in cyber security in the near future?

‘First of all, we can make improvements in the security of the Internet infrastructure. Domain Name System security is one such improvement. It means that when you visit a website, you can be sure that it is really the website you intended to visit and not a fake website that looks similar. The same goes for data integrity. If you go to google.com and you get data, it should be automatically guaranteed that the data have not been changed underway. Second, since so much of Internet use takes place nowadays via mobile devices, we have to stay ahead in the mobile world, which is a very different one from the traditional desktop world. Third, the next generation of hardware and software systems needs to have built-in security. Internet users should not have to worry about security. The Internet should be like water or air.’

I am sure that you know the TV-series ‘Person of Interest’. How realistic is it according to you?

‘Hollywood is always interesting when it’s doing cyber security. Some of such series have been predictive. However, I think that ‘Person of Interest’ causes more anxiety than that it raises awareness. Hollywood is Hollywood. Sometimes they are ahead of the game, and sometimes they are not.’

---------------------------------------------------------------------------------

NWO research projects on cyber security

In April 2013, nine Dutch cyber-security research projects received a total of EUR 3.2 million in funding from the Netherlands Organisation for Scientific Research (NWO) in connection with the first call for proposals in the long-term Cyber Security research programme. The nine projects will investigate a wide variety of cyber-security challenges. For example: Can ‘backdoors’ in embedded devices (allowing cyber criminals to control them remotely) be automatically detected? What can we learn about the personality traits, the motivations and the networks of cyber criminals? How can we strengthen the weakest link in cyber security: consumers without any expertise? How can we find a balance between securing personal data and keeping information systems user-friendly? How can security analysts best detect malware? A second call for research proposals is expected this summer. The proposed research all fits in with the Dutch National Cyber Security Strategy (NCSS). As part of the strategy, in January 2012 the Cyber Security Centre has been founded, that collaborates with NWO.

Internet
Cyber-security treaty signed between the US and the Netherlands:
www.nwo.nl/actueel/nieuws/2012/nwo-en-ncsc-geven-invulling-aan-nederlands-amerikaanse-samenwerking-in-cyber-security-onderzoek.html
Nine NWO projects on cyber security:
www.nwo.nl/actueel/nieuws/2013/ew/negen-projecten-in-cyber-security-onderzoek-van-start.html
Nationaal Cyber Security Centrum:
https://www.ncsc.nl
Department of Homeland Security on cyber security:
www.dhs.gov/topic/cybersecurity
Cyber-security awareness campaign ‘Stop, think, connect’:
www.dhs.gov/stopthinkconnect
DHS Science & Technology Directorate, Cyber Security Division:
www.dhs.gov/st-csd