Your AI Friday – 10001

Welcome to the weekly selection of links compiled by PAC's AI team. This week, the articles highlight how delicate it is to properly program AI to avoid false positives. Enjoy reading!



#eatingTheSeedCorn #GAFAvsTheFuturOfKnowledge #LetThemGrow

Due to the hunt for skilled workers by GAFA (Google, Apple, Facebook, Amazon) – but also others – universities have to struggle with a decreasing number of lecturers. Even if these – now privatized – researchers declare to still invest 20% of their time in teaching, 100% of their research time is now dedicated to a single entity under private law. A fair modus operandi will thus have to be put in place, and quickly, if we don’t want to risk losing one of the most promising parts of an (already stricken) education system.

“As it builds these labs, Facebook is adding to pressure on universities and nonprofit A.I. research operations, which are already struggling to retain professors and other employees.” […] ““If we lose all our faculty, it will be hard to keep preparing the next generation of researchers.”​

Facebook Adds A.I. Labs in Seattle and Pittsburgh, Pressuring Local Universities - Cade METZ



#AIDoWhatProgramSays #ItDependOfTheContext #WeStillNeedHumans

"The car has seen the victim but decided to ignore it”. Apart from this attention-grabbing headline, this unfortunate event illustrates two fundamental aspects in the interaction of "autonomous entities" with the real world: i.e. that complementary human skills are essential, so AI-based solutions can be implemented well. And that the context for which the solution has been developed has to be considered to be able to define the rules to be applied and analyze the consequences (if need be).

"Given Uber’s relentless dedication to cutting corners and driving growth in any way possible, it’s easy to assume this is just another case of Uber being Uber. At the same time, the accident caused Arizona to suspend Uber from conducting any further self-driving car tests in the state, and the company proceeded to shut down tests in all cities. It’s not at all clear if Uber was taking unnecessary risks to hit its internal goals, but it’s certainly paying a huge price right now.“

Uber's Self-Driving Car Sensors Ignored Cyclist In Fatal Accident - Rhett Jones



#WhatYouSeeIsNotWhatYouGet #PersonOfInterestOrNot #WeStillNeedHumansBis

Of course, there is an interest in using an artificial intelligence solution for facial recognition in combination with an urban video surveillance network. And of course, there are pros and cons. At the moment, these are above all the many technical improvements that can be achieved, monitoring and systematic human decision-making. However, what shouldn’t be forgotten is the need to formulate a clear framework of when and for what purpose such highly-intrusive methods may be used.

"But experts have warned that the systems used by South Wales Police and other forces, have little regulatory oversight and lack transparency about how they work. There are also questions over their accuracy. On the other hand, police say the correct matches have led to arrests and that the system is improving. […] The force says each time the system issues an alert a police officer reviews whether it is a match and if it may be can send an 'intervention team'. These teams are only sent when it is believed the match is correct."

Facial recognition tech used by UK police is making a ton of mistakes - Matt Burgess



#JasonBournevsAI  #MassSurveillance #CyberColdWar

As any lover of spy novels knows pretty well, intelligence services always have the most technically advanced, state-of-the-art gadgets up their sleeves. So, why should anyone be shocked about the news that the CIA is planning to replace its spies with AI…? On the contrary, aren’t we more inclined to think that such programs must have been in use for a long time already (e.g. Palantir algorithms and methods). Widespread video surveillance and automated processing of images, sounds, activities and other events through big data are already an old story and have been discussed in the form of films/series many times over, just to name Jason Bourne or Person of Interest. What would be the actually interesting thing to know – but this is where the article does not go – would be the AI’s degree of autonomy and how vulnerable it is to false positives (since spy agencies are also quite well-known for their “black ops” culture).

“America’s oldest spy agency is transforming from the kind of outfit that sends people around the globe to gather information, to the type that uses computers to accomplish the same task more efficiently. This transition from humans to computers is something the CIA has spent more than 30 years preparing for.”

CIA plans to replace spies with AI - Tristan Greene




AI in all its facets: ads, videos, partnerships, computer graphics... to read, see and listen to this week.

Nokia acquires SpaceTime Insight, adding AI to its Internet of Things business -

The foundational tension in machine learning is between optimization and generalization - Twitter

Artificial Intelligence Is Cracking Open the Vatican's Secret Archive -

Particle physicists turn to AI to cope with CERN’s collision deluge -

OK Google, book me a reservation for… Twitter

ServiceNow acquires AI startup Parlo for its NLP techv -


Follow the A(I) Team on Twitter:

    'Hannibal' @OlivierRafal 

    'Looping' @dagautier 

    'Futé' @VicCrete 

    'Barracuda' @StephaneLinise