Here’s an example of how not to create a marketing site. PR agency Frac.tl made an AI that could be the stuff of nightmares – Grover, a PR bot that  churns out near-convincing lines of text, made headlines for being able to write convincing fake news stories, now has the ability to create fake marketing website content (i.e. blog entries).

Grover does this by creating a neural network of actual online stories which it then mimics. Frac.tl boldly showcased Grover’s capabilities in its fake website, thismarketingblogdoesnotexist.com. When you visit the site, the latest “blog entry” is a laughable piece about the best photo filters to use for marketing on Instagram.

The site may have some use as marketing instructional material for  what NOT to do when starting a marketing website, but read on to discover the site’s and Grover’s deeper purpose. You’ll also notice that the picture of the “author” is likewise fake, as the portrait is that of a “writer” that doesn’t even exist – “Barry Tyree” is a non-existent person that had his face generated by an AI known as StyleGAN.

Fake news items and content on fake marketing website  thismarketingblogdoesnotexist.com were created by an advanced AI named “Grover”. Hopefully it wasn’t named after the adorable, amiable character we know. Is nothing sacred? (En.wikipedia.org/wiki/Grover#/media/File:Grover.JPG).

Despite the clues, those who aren’t tech savvy could be easily fooled, if not for the clear warnings about the blog entry author’s lack of, well, existence. The website’s name could easily be misconstrued for being “tongue-in-cheek” and who knows, may be considered credible if enough gullible marketers were hoodwinked.

What then is the purpose of all this, apart from giving us that feeling of dread and making us wonder if anything we read online can be trusted? There is an actual method to the madness of creating this A.I., and it is a noble one: Grover was created by the Allen Institute for A.I. to detect fake news or fake content, and to do this, the A.I. must first be exceptionally good at generating it. This is because researchers discovered that a fake news A.I. would be familiar with the habits, quirks and traits of fake news items or content made by other A.I., and would use the same data (such as publicly available news items online) and thus have the highest success probability of detecting it.

To date, Grover is the only A.I. fake news generator made by “benevolent actors” that can tell human-written news items apart from machine-written ones with a 92% accuracy rate.

As of this writing, Grover isn’t fully functional yet nor has it been released to the public to detect fake news items or fake marketing websites, so we’ll all have to rely on our individual powers of discernment.

For more details on this story, click here: https://futurism.com/the-byte/ai-generates-horrible-marketing-sites

Leave a Reply