Listening to Tim Berners-Lee and Vinton Cerf debate AI at The Web Conference

Listening to Tim Berners-Lee and Vinton Cerf debate AI at The Web Conference

What happens when internet giants and brilliant minds sit down together and talk? The result is a discussion about the future, but one that does without irrational fantasies and fears but concentrates on the real challenges. That’s exactly what we got to see on April 26 in Lyon in the debate on artificial intelligence, organized as part of “The Web Conference”, the large annual international academic conference. Among others, the panel was made up of Vinton Cerf, one of the founding fathers of the Internet and VP at Google, as well as Tim Berners-Lee, principal inventor of the World Wide Web... I would call that quite an illustrious circle!

 

In the debate, which was enthusiastically moderated by – equally renowned – Professor Wendy Hall, Director of the Chair of Computer Science at the University of Southampton, three ideologies around data and artificial intelligence (AI) became apparent: the need for ethics around the use of data, the common interest in sharing data (as we can trust in the companies’ ethical conduct) and finally the need for education campaigns to avoid excessive and counter-productive regulation.

"Data is a right“. This is the main characteristic Tim Berners-Leer attributes to AI. This man, who founded the web, basically gave it to humanity as a gift but has taken note of how it can be used for good deeds and bad; now he promotes a “human-centered” approach towards AI. While Tim Berners-Lee welcomes the advancement of AI in light of the great new ideas it can spark and the positive impacts it can have, he is well aware of how the future will depend on the exact way the technology will be employed. What worries him most is its use by the companies, their approach being purely based on capitalist thinking. Tim Berners-Lee’s wish is for the people to be the primary beneficiaries of the resulting data, but he fears that it “may already be too late”. Indeed, AI having already exceeded human capacity in certain tasks such as trading, leading to optimized profits, the technology could strongly contribute to exacerbating inequality. Current events have been confirming this view, thinking of the Facebook scandal around Cambridge Analytica. When cautiously asked about it, Antoine Bordes, research lead at Facebook's research facility in Paris, acknowledged the need to be more vigilant about data usage. He also emphasized that AI researchers had been taken by surprise "by the sudden notoriety of their subject" and that they were now beginning to integrate the aspect of ethics and privacy into their research. And while this is a first step towards making algorithms more "ethical", what is needed is that companies’ business models follow suit.

Data, an obligation? Another ideology, marked by a strong liberalism, was promoted by Kira Radinsky, director of data science at eBay. She defended the business logic of companies by reminding people that we "are all the trainers of machine learning algorithms", due to the data we produce. Her notion is that what we are doing is fighting the wrong enemy and thus wasting time and energy; besides, "worrying about singularity or a strong AI is like worrying about overpopulation on Mars.” As to granting more control to individuals, limiting the distribution of data is considered by her something that stands in contrast to the common good. According to Kira Radinsky, data may in fact become the equivalent of a currency and it would even be dangerous not to share them – as they could always benefit someone somewhere in the world, especially when it’s data related to health. Her focus is thus clearly on the aspects of the common good and corporate responsibility. Ruhi Sarikaya, scientific director behind Amazon's digital assistant, Alexa, thinks that people should learn to accept that in certain situations, e.g. health issues, sharing their private data is important and could also improve their lives. He deplores the fact that buzz phrases such as "if it's free, then you're the product" are destroying matter-of-fact debates and tarnish the image of AI in general, including its potential benefits. Ruhi Sarikaya strongly believes in the desire of companies to improve people's daily lives; of course, you can jeer his naivety, but it was in fact hard to not believe his good intentions when listening to his passionate speech about Alexa at the beginning of the conference, a product that is on its way to become “friction-free”. He also announced that Alexa would be equipped with a new “memory skill” for more efficiency.

It's all a matter of balance. Vinton Cerf presented himself as a worthy representative of some sort of ideology of “wisdom”. The 74-year-old engineer, who has advised several governments and probably has one of the longest CVs on earth, calls for the right balance between regulation and education. If regulations such as the GDPR can be a blessing, they can also be a curse. For him, it is a question of where to begin, and governments should not act too fast, simply because at this point nobody has any concrete idea of the exact impacts of AI. He added that “conscience is the most powerful tool” and that we would need to "make AI creators and users aware" of all the issues around the technology; thus, a clear focus on education and spreading information. Cerf did not promote blind trust in the businesses, though, saying that "more transparency from the business side" was necessary – which is quite surprising coming from a Google evangelist. It is not known whether Mounir Mahjoubi, the French Secretary of State in charge of digital affairs, will follow his advice; in any case, he concluded the debate by reminding everyone of the need for diversity and education in this matter. He also took the same line as some of his panel members and called for companies to share their data in the sectors of health, agriculture or energy.

 

In France, where the topic of AI is ubiquitous, the debate we saw was definitely one of the most high-level ones. It must be said that, as the technology matures, the number of self-proclaimed AI evangelists at conferences is decreasing and also that the Web Conference was targeted at a rather informed audience, or one could even say experts in the matter. The event also excelled due to its truly international character, with the best American, Asian and European experts present. This diversity brought together this exclusive panel of web giants (including the inventor of WWW) and made it possible to identify the various ideological standpoints on AI that are translating into the policies we see being pursued around the world. Europe seems to be following Tim Berners-Lee’s approach, while North America acts along the thinking of Kira Radinsky. China, by contrast, is going down another – and frankly speaking, pretty unsettling – path in terms of civil liberties.