David Colburn

You’re all probably channeling Ronald Regan’s classic debate line “There you go again” as you realize I’m writing once again about artificial intelligence, but I just can’t help myself. This stuff is moving so fast into all sorts of areas that affect our daily lives, and it remains as mysterious to most as the old DOS system that Microsoft’s Windows operating system was designed to shield PC users from having to experience.
One newsworthy item of late is the appearance on Amazon of AI-generated books pretending to be titles by human authors. Ironically, researcher Melanie Mitchell discovered that her 338-page book “Artificial Intelligence: A Guide for Thinking Humans” had been recreated as a 45-page e-book authored by a fictitious person called Shumaila Majid. An analysis of the book confirmed it had been written by generative AI.
A devious variation on this theme are the e-books that are “summaries” of legitimate books that are authored by AI. Some at least include a disclaimer that they aren’t written by the original author and have nothing to do with them, but the AI summaries are typically poor reflections of Cliff’s Notes versions of the books they are about, while the “authors” pick up some easy money by trading on the real author’s name and title.
And be prepared for more and more AI-generated books and articles in the months ahead, as new AI apps to create written work continue to pop up at a surprising pace. A quick Google search reveals 19 AI software programs for writing novels, for example. Many others focus on more technical writing. On the site Inkflow, all you need to give it is a title and the number of chapters you want, and it will generate a 20,000-word book based on that alone. The site operators do say that the book will need some human editing, and that it’s best for nonfiction works. So, if you’ve been mulling for years about writing “The Ultimate Guide to A Successful Deer Hunt,” have at it.
But beware – AI often gets things wrong. It probably wouldn’t tell you to use an M1 Abrams tank for a successful deer hunt but there’s no guarantee that everything it writes about deer hunting will be accurate.
A very handy type of AI app is one where you can take a picture of something with your phone and the AI will identify what it is. Sounds like a great thing to have when walking around in nature, but that app could turn deadly if you’re using it to identify poisonous vs. nonpoisonous mushrooms to pick and eat. It’s like facial recognition AI software for mushrooms, and as we’ve seen facial recognition programs lead to false identifications and arrests, we’re now seeing AI mushroom identification apps that mistake poisonous mushrooms for edible ones. Foraging for wild mushrooms is risky business, and there’s no substitute for investing the time, effort, and real-world experience to learn the unique characteristics of local species yourself. Oh, and if you’re thinking about buying an e-book mushroom guide for your phone, check it out well if it’s been published in the last couple of years – if it was generated by AI, it’s subject to the same kinds of errors.
A much darker side to the growth of the AI industry has been the use of AI tools in the creation of pornography, including child pornography. Fake images and videos created by AI abound online, with archives found on the so-called “dark web” containing thousands of images classified as child sexual abuse material. A particularly nefarious act is the creation of deepfake images and videos utilizing the faces of people mapped onto the faces of porn actors, making it seem like it’s those people participating in the film. The term “revenge porn” was created in response to such videos where a jilted spouse or lover takes the image of their former partner and uses it to create a porn video that they then spread online to harm them. With people routinely posting their own pictures to social media, material for perpetrators is easy to come by. Just last week, a 67-year-old third grade science teacher at a Christian school in Florida was arrested for creating AI child pornography images using yearbook photos. Legislators in Washington, D.C. and throughout the country are rushing to make AI-generated child sex photos and videos illegal. Last year, Minnesota legislators passed a law criminalizing the making and distribution of deepfake sexual imagery. A bill in the current session that would place the onus on cell phone, computer, and tablet vendors to install filters on those devices to automatically block access to websites known to display child pornography, revenge pornography, or obscene material harmful to minors has not advanced beyond an initial committee hearing.
Despite it all, AI is being used for enormous good in the world. For example, in the field of health care, AI is accelerating the development of new drugs, enhancing medical research, and improving diagnostic tools, while AI management tools are being used to improve efficiencies and cut down on costs. There are few areas of endeavor these days that haven’t been touched by AI in some way, and its potential is literally transformative. But as with most technological advancements in the digital age, there are nefarious actors who would use AI to propagate harm and evil in society. Social media is a prime example of technology conceived with positive intent that has been exploited for ill. It’s happening again with AI. Are we outsmarting ourselves with our technical achievements in AI, creating fantastic capabilities without a full understanding of their implications? The answer appears to be yes, and we’re once again playing catch up. But is AI racing too far ahead to catch up to it?