Google’s latest baby presents the ushering in of the nightmare-ish dystopian future that I’ve been dreading for a long time, at least that’s my perception… (Well, at least I’ve got that gag out of my system. Let’s move on, shall we?)
Google’s new Perspective API is, at it’s heart, a phenomenal piece of tech. Truly a triumph of machine learning. You give it a phrase and it will determine how likely it is to be perceived as toxic. This could be hugely useful when it comes to policing blog comments, reviews or chat rooms. However, if we think just a little deeper, we can see that the beauty is only skin deep. A potential monster lurks within.
There is no doubting that the Perspective API presents us with moral issues. (5% toxic) Taking a complex phrase and reducing it down to a simplistic toxicity score is, arguably, stupidity at it’s worst. (80% toxic) Wrapping the internet in cotton wool, protecting the denizens of this digital world is not the job of any corporation or country. (32% toxic)
But anyway, my (clearly toxic) speech is not what I am here to discuss (I’ll do that in another blog post). I want to go a little into the world of “what ifs”, but a world that I think you’ll agree is very close to our own. A little too close.
So How Does The Perspective API Effect Digital Marketers?
Did you happen to catch the Ad-pocolypse in March this year? If you have any interest in YouTube and Digital Marketing, then I’ll safely assume that the answer is yes. If not, here is a brief rundown:
A journalist for the WSJ noticed that adverts were showing on videos that contained questionable content (terrorism, racism, that sort of thing). This used to be a normal part of the algorithm, ads could show on videos even if they were not a YouTube partner. As soon as the companies whose ads were shown got wind of this, they pulled their ad dollars from the platform. We all know that nothing shouts louder than the almighty dollar, and we saw a sudden dip in Google’s stock prices. Google couldn’t allow this and as such they reacted swiftly and (in this humble marketers opinion) irrationally.
Overnight, ads stopped showing on huge numbers of videos. Google began to use the data scraped from the subtitle autogeneration and began labelling videos as religiously sensitive if they mentioned God, even in passing. Or as containing Drug content if they happened to mention something vaguely related to 420 (blaze it). Advertisers could then choose whether they wanted their ads to appear alongside drug taking blasphemers, the only choice being to say no.
So let’s look at the obvious next step in this technology thanks to the Perspective API. We suddenly find ourselves in a situation in which Google reads through the auto-generated subtitles and then determines how “toxic” a video is. Advertisers could then determine how toxic a video can be to show their adverts. Once content creators realise that this is the case they will either begin to leave the system/diversify (just like H3H3 Productions) or tone down their creations. Self-censorship is what we are suddenly looking at here folks. As we mentioned, nothing shouts louder than the almighty dollar.
As digital marketers we would have no choice but to weigh up potential “toxicity” of the videos that our clients adverts would be showing alongside, and the perception of those brands. It’s a no brainer that we, as marketers would, unless working with a ‘wild’ brand, have to lean towards those videos deemed less toxic.
Ladies and gentlemen, we become to enablers of this future.
In AdWords, you’re given a number of specifications for your ads. They cannot be too long, too short, contain trademarks unless the circumstances are correct, cannot contain too much punctuation or punctuation in the incorrect places. It’s a little strict at times, but it’s all livable. And then, within these rules you need to craft an advert that communicates your message perfectly, while being eye-catching and, of course, converting well. It’s a delicate balancing act, and one that has changed numerous times in the last few years.
In a bid to clean up its ad system, I would not be surprised to find that Google had introduced the Perception API, your ad needing to score a Toxicity of 30% or less for it to be shown on sites. Maybe sites could begin to filter the ads that are shown through the platform, only allowing ads on their system with a toxicity of 10%.
Perhaps if your ad is below a certain toxicity you could see your ad score improve, meaning that the ads cost you less.
We have again reached a point at which you are being incentivised to reduce the perceived toxicity of your ads. And again we as Digital Marketers would be the ones that facilitate this shift.
Oo, this is a biggy. You may want to sit down for this one.
SEO has arguably been a great thing for the internet. Google has become the arbiter of our content, determining who deserves to rank at the top of their search engine. So far, this has been with few goals beyond the improvement of the customer’s experience and the prevalence of their ad platform.
However, with the changes in YouTube due to the Ad-pocolypse and the fiasco regarding the sacking of the writer of the right-leaning memo, we have discovered a Google which feels the need to be socially conscious, even to a fault.
Bearing this in mind, it’s not too far fetched to suggest that Google could begin to use their Perspective API as a ranking factor. As always, it would be a small factor at first but would become more important as they begin to ‘perfect’ it. And, let’s face it, it would likely be a negative ranking factor.
So we would suddenly find ourselves in a situation in which SEO specialists need to discuss the content of pages on a very personal level. This content is too toxic, could you make this more positive? Could you get rid of the mention of Hitler and replace it with someone deemed less toxic?
And, who is it that’s determining what is, and what isn’t, toxic? Google.
Google would stand as the leader, the arbiter of the future determining the messages that people are allowed to find, their tone and “toxicity”. Of course, Google would suggest that we would in fact be the true holders of power. This is of course an AI project and it’s answers are determined by what has been communicated in the past.
But, well, Google hasn’t exactly shown itself to be completely trustworthy. Nor has it been completely unwilling to allow governments to have their say. Let’s take (my home) the UK for example. Could you really see Theresa May not jumping at this opportunity? A system that could influence the mood of the general populace and determine their mood in turn? Do you think that Donald Trump would pass up the opportunity to bathe his nonsense in a more positive light by restricting comments surrounding them to a toxicity of 20% or less? You’re damn right he wouldn’t.
Why Am I Bothered By This? Really?
I know that I sound alarmist. Would you like to know why?
It’s because I’m alarmed.
This technology worries me for a huge number of reasons. And not just as a Digital Marketer, but as a human that grew up within the internet. I’ve seen the world develop from one in which you should never get into a strangers car to us all carrying phones with an app that is explicitly designed to find strangers currently driving in your area to give you a lift. Technology has brought us all closer together. This technology could tear us apart.