The use of deep learning for social media and online content generation has gained equal parts attraction and concern in recent years. With a reputation as a cog in the fake news machine, deep learning has been a driving force behind political propaganda, false information and, to the untrained eye, almost undetectable impersonations of real, influential people. It’s not hard to see why this could be problematic, and even dangerous, in many cases.
However, before we condemn deep learning technology and the organizations propagating it, let’s also consider the benefits it can provide. For example, natural language-powered automated translation has the power to bring communities closer together and democratise communication. In countries such as India, where there are 22 official languages, deep learning and natural language technology can break down barriers, for example in healthcare where patients may present with regional dialects that even the most multilingual team cannot contend with.
For example, Bonaventure Dossou, a Canadian student researcher, was keen to improve phone conversations with his mother, who would often send him voice messages in Fon, a Beninese language. Bonaventure often couldn’t understand some of the phrases his mother used, which made conversation difficult. As a rarely documented language (one of hundreds in Africa) translation was difficult. To overcome this, he created an artificial intelligence language translation model, powered by deep learning, often used by modern NLP algorithms. The model is still a work in progress, but it shows how real-time translation can allow people who speak different languages to communicate with each other.
Dark side of fake news
It’s not just translation where deep learning-powered NLP can be used effectively. It can also ease the workload of people in industries like journalism, where fact-based, ‘template’ articles such as financial reports (which few journalists enjoy writing), can and, in many cases, are being written automatically using deep learning technology. While ‘bots’ can by-line these time-draining articles, journalists can spend more time on important research and investigative work, writing the more creative, in-depth articles that require the real skills of journalism.
Of course, we can’t talk about journalism and deep learning without considering the dark side of fake news. In March 2021, a landmark vote was scheduled to decide whether the first-ever labor union should be created at a US-based Amazon warehouse. Leading up to the vote, multiple deepfake Twitter accounts were created, using convincing profile pictures and posts to defend Amazon’s working practices.
These fake accounts were unlikely created by Amazon itself; but regardless of the motivation or perpetrator, they are an example of ‘deepfake’ language technology at work. Alongside the written content these accounts included profile pictures that had many of the characteristic flaws of deepfake images. In this instance, the tweets had little impact other than to cause confusion over who was responsible. However, it was a worrying reminder of how easily a false narrative can be created and perpetuated using deep learning technology.
These technologies are becoming more accessible than ever. Creating 8.5 million 20-word comments on Twitter would only cost around £6,000. In many cases, you wouldn’t even need half that many tweets to spread a convincing and powerful narrative. The impact a well-funded campaign could have on elections, public health communication, and market trends is real. So, how do we combat this and ensure that deep learning technology is being used for good rather than bad?
Very real dangers
Work has gone into counteracting the negative implications of deep learning and its role in the spread of fake news. For example, one algorithm can detect fake tweets with about 90 percent accuracy. Another solution relies on whether a text model can predict the next word in an article or post. The more predictable the content is, the more likely it was written by a machine.
It’s certainly encouraging that these solutions are being developed. However, there’s going to be a constant game of cat and mouse between those generating fake news and the people looking to put a stop to it. The deep learning technologies used to create the content will be updated to avoid detection, putting the ball firmly back into the court of social media providers and regulators to find new ways to detect and remove it.
But are the social media companies really invested in stopping the use of these technologies to spread fake news? Evidence would suggest not – former Facebook data scientist turned whistle-blower Frances Haugen told the US congress that the social media network was putting profit before powers to curb misinformation. In a complaint to federal authorities, she alleged that Facebook’s own research shows that it amplifies hate, misinformation and political unrest.
Deep learning technologies are undoubtedly playing a significant role in this amplification, yet it seems that even the world’s most successful and powerful platform is far from invested in countering its ill-effects. Facebook even disbanded its own counter-misinformation team that was formed at the time of the 2020 US election with the aim of ensuring content likely spawned by Russia did not ‘infect’ the news feeds of voters.
While misinformation on Facebook is still mostly human generated, it’s perpetuated by fake accounts and bots. Deep fakes are still largely in the future, but we’re starting to see the very real dangers of what can happen with little to no regulation in this area.
As deep learning technology becomes even more accessible, the onus is on social media providers to find a solution to the spread of fake news and invest in research and tools to counter it. At the same time, investments in solutions that power positive applications, such as NLP that breaks down language barriers, shouldn’t be side-lined.
Mike Loukides, Vice President of Emerging Technology, O’Reilly