Of course digital marketers still have jobs. But it won’t be the exact same job, as the AI begins its “second act.”
Yes, AI is still good at compiling, sorting, and classifying massive amounts of data. It’s just that now it’s increasingly able to assist in creating content in a way that it never had before.
All you need to do is provide specific inputs to the AI application. Render for me an image of a 1920s gangster taking a smartphone selfie. The technology didn’t exist 100 years ago, but the images do look convincing.
AI has evolved, but how do people use it now? Do you replace people and automate their work? Putting AI in charge? Or let artificial intelligence assist humans?
Need to be the mother of improv
The promise of using artificial intelligence to create visual and written content has been around for a long time. “A year ago, it couldn’t be done This, says Adam Binder, owner, founder and CEO of digital marketing and SEO company Creative Click Media.
Things are different now, Binder called in necessity. “When my I tried to set a deadline when the human writer was away. When it comes to AI, Binder made all the necessary inputs to get the app to do some copywriting.
The app he uses is GPT from OpenAI, a nonprofit AI research and deployment company -3. It uses existing copy and popular search choices from Google as raw material to generate its output. “The writing instrument creates a tone,” Binder said. “It’s scary how much it [with the author’s copy] is resemblance.
Still, “AI is a long way from replacing authors,” Binder said. “It can’t come up with papers.” He added that it couldn’t compare and contrast, or generate the same level of copy as a New York Times op-ed.
Google dropped the flag, declaring that AI-generated content violated its site Admin policy. However, Google may not be able to detect such copy. “AI writing is perfect,” said Chris Carr, president and CEO of digital agency Farotech. “Humans make mistakes.” ”
Get MarTech! Everyday. Free. In your inbox.
Let it sing
The input for GPT-3 comes from the first 20 pages of a Google search. Although it can Output a usable copy, but it’s not “final” or printable. “You have to hone it,” Carr said. “Humans have to humanize it. “
Carr demonstrated GPT-3 in interview. After typing “How do I use a hammer?” After the question, Carr was given a series of answers, divided into a series of options. By picking the feedback, Carr was able to piece together a copy of about 600 words in about five minutes. The average writer takes about an hour or two to produce the same output .
While AI can make copy, it cannot “make it sing”. AI-generated writing is bland, truthful, and aimed at selected points of view. AI cannot “connect the dots,” Carr said ” or “show personality.” Professional writers can write an argument or add a phrase to liven up the copy, something artificial intelligence can’t do.
If you include written content from well-known authors, then GPT -3 Can imitate their writing style. But that still leaves the copy to be edited and polished. “Great content comes from synergy. It’s about moving people’s minds with a great argument,” Carr said.
What AI shouldn’t do
Sadly AI is vulnerable to “GIGO” – garbage in, garbage out. There is a risk in asking an AI application like GPT-3 that it will pick up from what is best left as-is and unread Get information. Pieter Buteneers, director of machine language and AI engineering at communications PaaS company Sinch, warns that the internet, from which it draws its original content, is still full of misogynistic and racist content.
You don’t draw from it A place for inspiration,” Buteneers points out, as does Twitter. Instead, get information from news sites or Wikipedia. “Yes, there’s some crap in there, but it’s human-curated content.” The AI is all about picking patterns, but can’t tell what it’s picking up. “It doesn’t have common sense,” Buteneers said.
Another potential for abuse is fake content — spam or phishing — Carr noted. “It’s ridiculous that the power of doing bad things is so easy.” He added that Google couldn’t separate the wheat from the chaff. A good parallel example is the use of spam calls, now so common that people would have voicemails block calls instead of answering them.
In Sinch’s case, the company does offer “fraud filters,” where its AI can make judgments about what appears to be “phishing.” “It could be wrong, but it’s better to stop a false positive than let go of another 999,” Butnells said. While suspicious material can be flagged, humans must make a judgment as to whether to let it go or stop it.
AI is often tasked with finding bytes of insight from terabytes of data, but even then one has to worry about GIGO. Joyce Gordon, AI/ML Product Manager at Enterprise CDP Amperity, explains that AI must integrate data and resolve identities to ultimately enable a customer-centric approach where more data can be accessed and leveraged.
“This has to happen before brands start using AI to make judgments — otherwise, they’ll be making judgments based on partial or inaccurate data,” Gordon said. “The algorithm is only as good as the data used to power it.”
“All AI models contain human bias. Therefore, it is impossible to exorcise all the ghosts in the machine,” customer engagement said Paul Hebert, director of corporate strategy at platform Cheetah Digital. “Going forward, I think the trend will be to use ML to process tasks, to use AI to make suggested judgment calls, and humans to make final decisions.”
Hebert cites the HBR article linked earlier here. “As [Garry] Kasparov said, ‘Weak people + machines + better processes outperform powerful computers alone and, more significantly, stronger people + machines + inferior processes.'”
What digital marketers can do
The capabilities of AI will improve over time, although it has limitations. Digital marketers will have to learn on the fly and try to identify the best use cases.
Carr offers this advice to marketers: “Get out of [AI] or get run over by it. Like it or not, it’s coming.” Also, learn how Google works so you can Get the best results and the most comprehensive source content. Creative people are the glue that keeps content sticking.
Finally, Carr revealed his biggest concern: If the internet is corrupted by fake content and spam, “search results will become a wasteland of bots.”
” Start small,” Butnells said, but learn quickly. Learn quickly from mistakes. “Jump in — one toe at a time.”
“So, crawl, walk, run,” Herbert said. “[M] Marketers need to be slower than technology and as fast as ethics allow. Ethics are the stewards of AI engines.”
About the author
William Terdoslavich is a freelance writer with a wealth of information technical background. Before writing for MarTech, he was also responsible for digital marketing at DMN. A seasoned generalist, William has worked in the IT industry at Insights.Dice.com, Big Data at Information Week, and Software as a Service at SaaSintheEnterprise.com. He has also served as feature editor for mobile computing and communications, as well as feature editor for CRN, where he has to deal with 20 to 30 different technology topics over the course of a year. Ironically, it was the human element that drew William to start writing about technology. No matter how hard people try to organize and control information, it doesn’t work the way they want.