Propaganda!

THE COMING AUTOMATION OF PROPAGANDA
https://warontherocks.com/2019/08/the-coming-automation-of-propaganda/

[...]

What the American public has called AI, for lack of a better term, is better thought of as a cluster of emerging technologies capable of constructing convincing false realities. In line with the terms policymakers use, we will refer to the falsified media (pictures, audio, and video) these technologies generate as “deepfakes,” though we also suggest a new term, “machine persona,” to refer to AI that mimics the behavior of live users in the service of driving narratives.


Improvements in AI bots, up to this point, have mostly manifested in relatively harmless areas like customer service. But these thus far modest improvements build upon breakthroughs in speech recognition and generation that are nothing short of profound.

OpenAI, a project Elon Musk founded, made headlines this year for its GPT-2, a text generation language model the organization deemed “too dangerous to release.” This framing was perhaps an exaggeration, but OpenAI’s work was impressive nonetheless. Testers gave the algorithm 40GB of seed text from links aggregated across the Internet, which it studied with the aid of a supercomputer, producing a lightweight output that a regular desktop could run. OpenAI released a toned-down version of the algorithm to the public, but the products the organization revealed of the full version were remarkable. Though OpenAI admits to taking a few tries to get a good sample, given the first line of Orwell’s 1984, “It was a bright cold day in April, and the clocks were striking thirteen,” it eventually produced a coherent opening to a near-future novel set in Seattle. With an opening line about the discovery of unicorns in the Andes, an article GPT-2 produced wouldn’t look at all out of place in a pop-science website. That is, apart from the subject. The “fake news” applications require little imagination. One study explored this exact scenario, showing that GPT-2 was able to generate foreign policy news that subjects rated on average only marginally less credible than the New York Times seed text.


These developments aren’t mere science projects either, but beneficiaries of market forces. Companies have used natural language processing (NLP) and generalized text generation to automate a growing share of the customer service and information technology workforce, cutting labor costs and freeing skilled labor from menial tasks. Advances in text generation have greatly benefited journalism in particular, driving media companies to invest in generating ever more believable content. However, NLP is just one facet of the AI revolution.


Advancements in image recognition and generation can now produce faces that are almost entirely indistinguishable from those of real humans. Intelligence organizations have already used this technology to solicit unwitting contacts through social media. The same underlying technologies have also led to a recent spike in deepfake videos, now letting anyone with at-home software  blend real footage almost seamlessly with generated content. And you don’t have to take our word for it, trust former president Barack Obama.

AI technology also has less flashy, but no less substantial applications in influencing what users see online. Social media platforms work by identifying trending content and boosting it into the feeds of other users. While the case varies from platform to platform, these trend algorithms tend to be a function of ‘likes,’ ‘retweets,’ or ‘upvotes’ over time, but they weight early interaction most strongly. This means that a small, concentrated burst of interaction at the birth of new content is often all that is necessary to send it trending, pushing it into the feeds of thousands of legitimate users.
[...]