A display shows a vehicle and person recognition system for law enforcement

نهج احترازي في التعامل مع الذكاء الاصطناعي

May 1, 2018 Maciej Kuziemski

فلورنسا ــ من منظور صانعي السياسات في أي مكان، تتلخص أفضل طريقة لاتخاذ القرارات في تأسيسها على الأدلة، حتى وإن كانت البيانات المتاحة منقوصة. ولكن كيف يتصرف القادة عندما تكون الحقائق نادرة أو منعدمة؟ هذا هو المأزق الذي يواجه أولئك الذين يتعين عليهم أن يتعاملوا مع التداعيات المترتبة على "الخوارزميات التنبؤية المتقدمة" ــ أحجار البناء الثنائية للتعلم الآلي والذكاء الاصطناعي.

في الأوساط الأكاديمية، سنجد أن العلماء الباحثين في مجال الذكاء الاصطناعي إما "فردانيون" أو "حاضريون". يزعم الفردانيون" في عموم الأمر أنه في حين تشكل تكنولوجيات الذكاء الاصطناعي تهديدا وجوديا للبشرية، فإن الفوائد تفوق التكاليف. ولكن برغم أن هذه المجموعة تضم العديد من النجوم البارزين في مجال التكنولوجيا وتجتذب قدرا كبيرا من التمويل، فإن ناتجها الأكاديمي كان فاشلا حتى الآن في إثبات حساباتها بشكل مقنع.

على الجانب الآخر، يميل الحاضريون إلى التركيز على العدالة والمساءلة والشفافية في التعامل مع التكنولوجيات الجديدة. فهم مهتمون على سبيل المثال بدراسة الكيفية التي ستؤثر بها الأتمتة على سوق العمل. ولكن هنا أيضا كانت الأعمال البحثية غير مقنعة. على سبيل المثال، عقدت مؤخرا مجلة تكنولوجي ريفيو التي يصدرها معهد ماساتشوستس للتكنولوجيا مقارنة بين 19 دراسة رئيسية بحثت في خسائر الوظائف المتوقعة، ووجدت أن التوقعات لعدد الوظائف "المدمرة" على مستوى العالَم تتراوح بين 1.8 مليون إلى ملياري وظيفة.

To continue reading, register now.

As a registered user, you can enjoy more PS content every month – for free.

Subscribe now for unlimited access to everything PS has to offer.

Already have an account? Log in

Subscribe to Project Syndicate

Enjoy unlimited access to the ideas and opinions of the world’s leading thinkers, including long reads, book reviews, topical collections, short-form analysis and predictions, and exclusive interviews; every new issue of the PS Quarterly magazine (print and digital); the complete PS archive; and more. Subscribe now to PS Premium.

new comment has been posted. new comments have been posted.

0 Comments on this paragraph, 6 Comments on this article

Prashant Kotak May 6, 2018

"...A requirement of explainability of algorithmic decision-making can change incentives..."

Well, now.
A new religion of the EU, and with it, a new priesthood arises. And, a new debate about the number of angels that can dance on the head of a pin.

But I am clearly in the presence of an ayatollah of the religion, so I must take this opportunity to make confession, and also take some theological instruction.

Forgive me father, for I have sinned. I have in four decades of programming coded innumerable algorithms, with not a thought for the adverse consequences for others. In my late teens I coded AI techniques like Minimax, and also (gasp!) Alpha-Beta-Pruning. My professors warned me too much of that kind of thing can lead to blindness, but my only excuse is, I was young you know.

But now to the present - I have coded and used all of Bucket sort, Bubble sort,, Insertion sort, Selection sort, Heapsort, Mergesort, Radix Sort, Shellsort and Quicksort.
Can you advise which is the theologically correct, EU approved sort for me to use henceforth please? I'm guessing it can't be Quicksort, because the EU doesn't do quick, but they are good with anything between glacial and turning supertanker. Anyway, I eagerly await your judgement.
A new reply to this comment has been posted.
Jacob Hughes May 4, 2018

Europe is voiceless on the global stage and lacks fundamental knowledge of geopolitics, using its own forms of independently-developed machine-learning technologies to integrate into global networks. There are European space (non-global) conditions that require urgent addressing, which is perhaps not as clearly evident in the article and related books on the subject, despite its intention. As the continent exists as an insular, unconnected entity on the international stage and separate from geopolitics, it deploys inchoate technologies to access broader international knowledge in order to evaluate urgent space conditions, as well as colonization efforts elsewhere. "Exhibit 1" is Europe's marginality.
- Jacob Hughes May 4, 2018
  
  addendum--the function of AI in this context serves a humanitarian, globally-minded end stressing positive environmentalism. Global AI infrastructures facilitate the transfer of perspectives from all populations and countries, which includes but is not limited to global powers (China, Russia, and the USA). Granted, the article's articulation of this core function is imperfect in its construction and depiction of "weaker voices," for instance, which is certainly worthy of critique.The aim is best expressed in the "preventing blackboxing" category, and allowing public and private sectors to connect. Precautionary approaches are also useful in that "Decision-making would...be more inclusive and deliberative, and produce solutions that more closely reflected societal needs" and "consider multiple options," particularly to articulate visions for the future of AI, science, and technology.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Peter Schaeffer May 3, 2018

Europe should not hesitate to apply the "precautionary principle" to AI. Europe is already a Tech backwater. With a few more rules and regulations, Europe can make it self completely irrelevant (of course, that may already be true). China doesn't care about Europe's silliness and will develop AI without hesitation. The major military powers (China, the USA, and Russia) will develop and use AI weapons without giving the issue a second thought. The absence of Europe from the AI field, will make it more profitable for everyone else.

A cliche is that Europe combines utter arrogance with total weakness. This article could be used as exhibit 1.
A new reply to this comment has been posted.
Michael Public May 1, 2018

The author seems ignorant of the fact that AI is essentially software, a lot of which is open source rendering it unsuitable to regulation. Regulating it will probably just drive it underground like alcohol during the prohibition, creating AI barons who are just modern day Al (ha ha) Capones. Worse, it is inevitable just like the space race and the arms race - China, Russia and of course the US will throw everything they have into developing the most powerful AI algorithms possible - the militaries will get preference. Anything dangerous or outside the rules will simply go ahead as part of a black ops project.
- Prashant Kotak May 6, 2018
  
  I can sympathise with the AI Capone joke, having personally had some dealings with Bugsy Malware and Bootsie Rom.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.