This is a sample blog from ThePendits, for the Artificial Intelligence industry. It is a blog that would refer you with the format and style that is maintained in our blogs for the specified industry.
What if tomorrow we wake up to a post-COVID-19 world where the species called writers no longer exists. The promotional emails you receive, the blogs you enjoy, the special news feature on a particularly overwhelming “human” story – all would lack a human brain, or rather a human heart, behind the words.
Tomorrow has already been sold out to Artificial Intelligence, where programmed intelligence takes over spontaneous thoughts; bots counsel you out of emotional clogs, and human androids meet you with business proposals.
All that you read and receive online will be cleverly constructed by software that plays to the tunes of SEO algorithms to win the game rather than to appease your minds.
And, how would you like that?
THE AI BLOG THAT WENT VIRAL
Recently, there was an interesting piece of news about how a self-help blog by a University of California student ranked #1 on the social media news website Hacker News. The blog, aptly titled “Feeling unproductive? Maybe you should stop overthinking” was understandably lapped up by the readers amid the plummeting virus-infected spirits.
It wasn’t the popularity of the blog that created news. But the fact that it was an AI blog, entirely made-up by an AI tool, and nobody seems to notice!
For Liam Porr, a student of Computer Programming, Berkeley, it was an experiment to test the convincing capabilities of GPT-3, the language-generating AI tool from OpenAI.
The post got upvoted 200 times on Hacker News, brought in 71 responses from the site’s commenters and even quite a few subscriptions for Porr’s account.
As the borders between creativity and programmed intelligence get wafer-thin, and AI invades almost every section of human life, the question remains:
How would you like to have software dictate your text?
What Liam Porr found disadvantageous about the AI blogs
As promising as it may sound, Porr’s experimental AI blog does address a few perturbing truths in his own words. Though there is a pretty wordplay, the AI tool, according to Porr, “is not very good at being rational and logical.”
This puts the spotlight on NLP’s (Natural Language Processing) involvement of AI in augmenting researchers by producing multi-page summarized articles. Summarizing content through NLP can be either “extractive” or “abstractive”. In extractive summarizing, only the most relevant text is filtered in, as in Google searches. Abstractive, which is more MI-based, involves AI to come up with its own list of words for a given text.
In Porr’s trial with GPT-3, the tool works out well only when the topic that is fed to it isn’t demandingly logical. So Porr chose trending topics (like self-help and productivity) from the “less logical” categories and created suitable headlines for the tool to generate text. The formula worked out when he stayed within those “safe” parameters.
Though many applauded the blog, “unaware” of its origin, there were a few who seemed to take notice of the lack of originality. One reader commented, “This is either written by GPT-3 or a human equivalent. Zero substantive content, pure regurgitation.”
Porr also points out the “other side” of his AI content experiment. As convincing as it could look, AI content can be manipulated as pure clickbait, fooling millions of readers or consumers. He very foresightfully alerts about the “mediocre blog content” that is going to flood the internet as the hurdles become so easy with AI-backed content generation. “I think the value of online content is going to be reduced a lot.”
WHERE DOES AI FAIL TO SUBSTITUTE FOR AUTHENTIC “HUMAN” CONTENT?
A similar experiment like that of Porr was conducted by Kristin Tynski, VP of the digital marketing firm Fractl, to test AI’s possibilities in generating online content. Tynski created a website with AI tools (that were publicly available) complete with 30 well-written blog posts and an AI-generated profile of the non-existent author.
The experiment according to Tynski, throws light upon a scary side of the digital future where fiction duplicates fact so well that it will be hard to filter one from another.
Like how a few of the Hacker News’ readers got sceptical about Porr’s ‘artificial’ content, the AI-driven generation will soon see everything through a digital eye, dismissing, at times, even original ‘human’ content as SEO-oriented AI clones.
As Porr says, online content will lose its value.
The scary shadows of AI Content
A predecessor to Porr’s AI blog content was shared on ClickZ by Mike O’Brien two years back. Amusingly, he had asked an AI Content creation company to generate a piece on “AI writing articles”. The writer then went on to share the AI-written, unedited content on the website.
The content is fine when you first go through it. Good and relevant words. But as you keep reading you might bite a few chunks that taste odd!
To quote Mike O’ Brian’s words, “Some of the language is a little off, reminiscent of those literal Google translations that result in sentences that are technically correct, but awkwardly worded.”
Though a couple of years behind Liam Porr and Kristin Trynski, O’Brian also reflects the same fear shared by them about the future of AI content. Yes, there could be a day when AI is irreversibly woven into our social fabric or perhaps takes over lives for good. But with power, comes abuse.
As Porr remarked about the AI content generation,”it was super easy”, and that is “the scary part.”
Even as GPT-3 is introduced in succession to the earlier GPT-2 version, OpenAI’s own blog posts voice anxiety over the tools’ potential to be misused for fake information. The AI tools are easily vulnerable to be used for mass generation of spam-posts stuffed with keywords to tease Google’s bots.
Will anyone know the difference?
FUTURE OF AI INSTIGATED CONTENT GENERATION
Ironically, a post-COVID-19 business scenario would require empathetic marketing with a more human-centric approach. According to B2B content marketing strategist Megan Thudium, three ‘natural’ ways that companies can engage in a humanistic approach to consumers are by “thinking like people”. As she puts it, connecting to people “starts with listening, continues with empathy and ends with being meaningful”.
While businesses and economies grapple back on to recovery roads, those companies will sustain and win who reach out to the masses in their language. Brands will need to listen first, and then talk to their audience in a language they can emotionally connect with.
Liam Porr’s final post in his AI content experiment blows the whistle on the not-so-pleasant realities of AI-rooted content. After a two-week series of AI-generated blogs, Porr concluded with a self-written ‘human’ message that explained his experiment as hypothetical. It was titled, “What I would do with GPT-3 if I had no ethics.” And that perfectly echoes the uncertain terrains of a warped future in AI-based content writing.
Amid the mass content by AI tools that would soon over-stuff the web, what will stand out as the best? As most emotionally appealing to customers? Or as most “human”?
While “intelligent” content programmed to please SEO manifestos are churned out every second, how will the human consumer connect and choose? Will there be further additions to filter the machine-content and the human one? Or will you be left clueless in a sea of online content to sort out your own realities?
Conclusion
Yes, AI will generate content. More and more AI tools and companies will crop up in this decade. More content in a fraction of a second will rewrite the nature of content marketing. But you will need a human hand, and a human heart - if not a human brain – to leave the final touches; to dab it with the necessary human element that your consumers can identify and connect with.
You will need a voice so distinct that it would stand out among the clones. Just as we know the original as the divine hand leaves its signature flaw within every creation, so would the consumers know, in the long run, what is human and what is not.