Like other mass cultural phenomena, just like cinema, television or advertising, the Internet has influenced our way of speaking and has changed it over time: different generations of people use different languages and different words to express the same concept, even at the risk of not understanding each other. Indeed, machines are changing the basis of grammar, pushing us to use terms that at school they are indicated as incorrect (famous is the case of Google Docs and “what is”).
In the last couple of years another phenomenon has emerged, that of the so-called algospeak, of which The Washington Post also wrote in mid-April. The term comes from the union between the English words “algorithm” and “speak” and indicates a way of speaking (“to speak”, in English) conditioned by the algorithms of artificial intelligence which is entrusted with the moderation of content on social networks. Above all, a way of writing.
The analysis of OStili Words
Ukraine and social media: this is how the influencers (not) talked about the war
by Emanuele Capone
Dollars and eggplants to avoid censorship
Succede su Facebook, su Instagram, su Twitch and YouTube and a lot on TikTok, where the bulk of the public is made up of minors and where controls are even tighter. The idea is to avoid forbidden words and use others in their place, euphemisms or periphrases to express the same concept: in English, for example, you write “le dollar bean” or “le $ bian” instead of ” lesbian “(lesbian), “Nip nops” instead of “nipples” (nipples), “spicy eggplant” to mean a vibrator and those who are part of the Lgbtq community describe themselves as part of the “leg booty” community. In Italian it is more difficult, because our language lends itself less to these games, but it is not uncommon to find “se $$ o” instead of “sex”, truncated words because the whole one is forbidden (happens with “russia”, which on TikTok becomes “russ”) or emoji like that of the trumpet, the corn on the cob or the inevitable aubergine to indicate that we are talking about sex. Even if to a lesser extent, it is a problem that newspapers also have, which find themselves in difficulty when they have to tell a news story (or a war) and therefore use terms such as “death”, “murder” or “crime”.
It happens to prevent your content, whether it be posts, photos or videos, from being blocked or removed from platforms or prevented from reach a wider audience and be successful. It is a question of freedom of speech and defense of the possibility of expressing oneself as one wishes on the topics one wishes, but it is not only this. It is also a question of money: on social networks, especially on YouTube, the contents that contain prohibited words risk being demonetized (i.e. they cannot be accompanied by advertising banners) or see their CPM and RPM levels lowered, which is reflected in the earnings of those who created them. For TikTok, another argument is valid: how on Italian Tech we explained at the end of 2021, its section For youthe one where the potentially most successful videos are, is totally managed by a categorization algorithm, which decides what to show to whom and is based (also) on language parameters. And posting a video that contains prohibited phrases or words isn’t exactly a good way to go viral.
The analysis
From music to lyrics: how a song must be to go viral on TikTok
by Francesco Marino
What artificial intelligences have to do with it
Why is all this happening? Better: why do social networks use AI for content moderation, starting (to simplify) from a list of forbidden words that is enriched as the software learns the language of the people? Trivially, why now the platforms have become so big and host such a large amount of content that they couldn’t manage it without outside help. An artificial aid, that is.
Facebook has nearly 3 billion monthly active users, YouTube over 2.5 billion and TikTok over 1 billion. They are all people who write, photograph themselves, film themselves and then publish. They publish continuously: More than 500 hours of video are uploaded to YouTube every minute, that is 30 thousand hours of content every hour; in the last 3 months of 2021, TikTok removed over 91 million videos for the most diverse violations, which are a lot, but represent less than 1% of those uploaded in the period. Got the point? The point is that a moderation team entrusted only to people, no matter how many they may be, will never be able to keep up with this. daily invasion of posts, photos, videos, reels, stories, tweets, live shows and so on. And that’s why sites rely on AI to do much of this work.
Which then leads us humans to always look for new ways to cheat them and get around them: there is nothing wrong with this, and indeed perhaps it is a bit part of our nature. But it is worth making a clarification. It’s not that Facebook, TikTok and YouTube all want us to be modest and perfect, because they know very well that in this way they would end up not representing the real world: these rules are not there for us, they are for those who would end up exaggerating and going further, for those who would make TikTok a new version of OnlyFans. On the contrary: for those who made OnlyFans what it is nowand which is definitely not what its creators had in mind.