Google said last week that it will launch a new tool, About This Image, for its search platform that can tell the difference between real and fake images on the internet. In a related blog post, the tech giant cited a study conducted by Poynter in 2022 that showed 62% of the users believed they come across misinformation daily or weekly.
This is not just happening to other people in other countries. About half of the Indian adults surveyed recently by McAfee experienced or knew someone who experienced some kind of AI voice scam. This figure is almost double the global average of 25%. The survey also noted that 83% of Indian victims lost money as a result of such voice scams, with 48% losing over `50,000. It is easy to understand why there is as much nervousness as there is investment in AI.
This is also an issue the dons of the Indian advertising industry are grappling with at the moment. While the industry—globally—has been among the early movers to incorporate AI tools into its product, back home, there is increasing soul searching about the most efficient use of such tools. There is an increased focus on finding ways to make sure they don’t end up exacerbating the already huge problem of scam ads.
There is no disagreement over how AI can help, though. No one will deny that it’s a great tool for small teams to save costs and enable faster ad creation. These tools can automate the most mundane tasks, they can be trained to optimise digital ad targeting, and can even help get over the copywriter’s block.
Now come to the challenges. The fundamental questions here are the same that every other industry is asking. If one can take off from the March letter by the Future of Life Institute, should we let machines flood our channels with duplicates and unoriginal work? Is there one good reason to AI-ise all the jobs, including the most fulfilling ones? Above everything else, do we really need non-human minds that might eventually outsmart and replace us?
Proponents of the AI-it-before-you-make-it line will tell you that ChatGPT-ing or DALL-E-ing it initially will get you started, and once you start on the road, your mind will automatically tune in to the new demands on the system.
But look at where that line of thinking has landed a vague medical devices company that was using Sachin Tendulkar’s name and likeness in fake advertisements endorsing its products. There is already an FIR against that firm for using the ace cricketer’s likeness without his permission. The FIR said the company was running its business online through the website sachinhealth.in, and also claimed Tendulkar had endorsed its product line.
This whole fiasco was reported by news outlets last week and the website has not been operational since but when it was, you could see how it had not only image-manipulated Tendulkar’s likeness in its logo, it was also using a fake Tendulkar voiceover to hawk its products. If there was no FIR, there was no telling Tendulkar had nothing to do with that image or the voice. In other words, while staying alert is perhaps harder than ever, we’ve never needed to do this more.
That said, faking—or “getting inspiration” from great ideas, as they sometimes put it—is not a new monkey. It is just that there is growing fear that the katzenjammer over easy fakes will undermine the whole business of trust on which the edifice of advertising is built. That apart, there are some studies already that don’t say much about AI’s capability to nudge people to actually buy products. Global delivery marketplace Borzo (previously WeFast) recently released a marketing case study that compared the effectiveness of human ads with AI-created ones. As part of the experiment, the service released two sets of advertising banners on Facebook, Instagram and Google.
The first set was created entirely by three AI tools—ChatGPT, which contributed the copy, and MidJourney and DALL-E, which developed the creatives. The pieces were compiled using Viewst. The second set was developed by the company’s marketing team comprising copywriters and designers. The two sets of banners were run simultaneously for over a month on a budget of $19,065. Google was allotted the bigger $12,775 chunk of the budget.
The paid specialists who optimised and evaluated the results were common for the two campaigns. They observed that AI-generated ads were more effective in getting attention but fell short when it came to generating clients. In comparison, the ads that were created by the human designers drew less attention but are more effective in generating customer leads—a staggering 9.5x more customers compared to the AI-generated banners—whilst burning less. The customer acquisition cost of AI banner ads was 2.5x higher than human-made banner ads.
So, if the whole purpose of advertising is to create customers at the end of the day, you are better off keeping your “human” teams happy and well remunerated and well rested.
At least for now.