ChatGPT: “The risk of mass manipulation will exacerbate the image battle between Web giants”

The conversational robot will not just shake up the Internet economy by threatening the advertising model of the digital giants, Gafam, explains economist Vincent Lorphelin, in an article in “Le Monde”.

We have become accustomed to using Google to direct us to relevant web pages. Why search these pages, and generate advertising revenue in the process, if artificial intelligence (AI) directly finds the answers to our questions? We use Amazon for a shopping experience documented by sponsored product listings. What will become of them when AI immediately offers the ideal product? We use Facebook or Instagram to be connected to our communities. What will become of this use if the connections and most of the content posted are produced by AI? The new generation of AI removes reasons to spend time or browse the web. However, this time and these clicks are monetized, which questions the very heart of the Internet: its advertising model.

In response, Gafam (Google, Apple, Facebook, Amazon, Microsoft) will probably not resist the temptation to replace advertising with manipulation. The Cambridge Analytica scandal showed, in 2014, that it was possible to influence voters' votes by presenting them individually with fabricated information. Since then, the game Candy Crush has demonstrated that it is possible to create enough addiction to charge for the opportunity to continue playing. Meta announced that it could modify its algorithms to increase the well-being of its users. The voluntary distribution of fake news has convinced 1 in 10 French people that the Earth is flat.

From now on, a similar use of new generation AI will make it possible to sell to the highest bidder not an audience but a service providing direct revenue, or even a quantity of favorable (or unfavorable) opinions! Collateral effect, this trade in manipulation will be at the expense of the capacity of judgment of consumers and voters. MEPs were alarmed by the danger revealed by Cambridge Analytica for democracy. It was the Stone Age of AI. Today, the industrialization of manipulation constitutes a systemic risk, in the same way that global warming is an ecosystem risk.

Furthermore, Gafam's commercial lands began to overlap. Amazon and Apple recently entered the advertising market dominated by the Meta-Google duopoly which, in return, is launching into e-commerce. Microsoft wants to extend the Xbox Store to mobile against Google Store and Apple Store. After video, music, cloud and payment services comes the turn of the metaverses, the challenges of which could exceed those of the smartphone, or even the Internet. While Apple has not yet announced a product in this segment, Meta and Microsoft have already joined forces against it.

This new competitive game revealed the importance of Gafam’s image. As early as 2019, Apple, by championing data confidentiality, showed that defending social issues could provide image benefits at the expense of its competitors. Since then, for image reasons, Gafam have in turn removed facial recognition, accepted a sanction for misleading commercial practice in order to avoid the label of planned obsolescence, waived commissions on turnover of applications for the benefit of small developers, while organizing campaigns to denigrate their competitors.

This importance of image is measured by the success of new products and the difference in salaries of engineers between companies. When it is bad, it can especially attract the attention of regulators, who use “business” to justify their action. The threat of dismantling Gafam still looms in the United States, but should decrease with the increase in competition between them. On the European side, on the other hand, the Digital Services Act is precisely designed to counter the risks of AI. But the risk of mass manipulation, which will undoubtedly trigger alerts that regulators will take, will exacerbate the battle for image.

Gafam will be attentive to AI errors causing these alerts. Amazon has already paid the price for a sexist recruitment algorithm, which mechanically reproduced ambient information and its biases. The new equivalents of Wikipedia, currently being built, are made by AI, but human contributors intervene to judge the result. The quality of information produced by AI will become as sensitive as respect for private data and will require human intervention to govern it. After several years during which they remained deaf to the demands of intellectual owners, first and foremost press publishers, Gafam will finally have to recognize that they need their stamps.

In all likelihood, the Internet's advertising model and intellectual property theft will decline. The Gafam, whose areas of action were historically separate, will compete and engage in a bitter battle for image. ChatGPT has introduced a new chapter in the digital economy.