New Digital Horizons

Digital content creation, it seems, has reached an inflection point. Artificial intelligence has become so phenomenally powerful that it is now poised to reshape all of our creative industries. Generative AI, as the software in question is known, harnesses sophisticated deep-learning capabilities to produce novel images, text, music, video, and code — giving it a staggering array of potential applications and, according to Sequoia Capital, perhaps even the ability to generate trillions of dollars’ worth of economic value. As a result, the emergence of this extraordinary next-generation technology has sent ripples of excitement coursing through Silicon Valley.

Just last month, Stability AI and Jasper, two startups whose AI platforms auto-generate text and images, attained unicorn status — or billion-dollar valuations — after raising $101mn and $105mn in funding, respectively. They’re not the only movers and shakers to have burst onto the scene, either: other promising companies include Replikr, which uses generative AI to create ‘emotionally responsive’ customer service avatars; Musico Ltd., which enables clients to make original advertising jingles; and GoCharlie.AI, which auto-generates blog posts and other types of marketing content. And similar startups are springing up all the time.

But is this technology quite as revolutionary as its supporters claim? Well, to answer this, let’s first consider the text-to-image application. Here, the AI models allow users to input a text-prompt and the software will produce an original image that corresponds to the instructions. Now, experimenting hobbyists have already created a profusion of exotic images, but as the platforms become more refined, we can expect to see advertisers, for instance, crafting more varied content. Artists and photographers, too, will use AI to create and enhance images; and this will translate into unusual book covers, calendars, wall hangings, album covers, wallpaper, and so forth. It will even eventuate in new clothing designs, proffer striking new shapes for vehicles, and inspire architectural work — a process that is already under way.

This is quite remarkable. But more remarkable still, perhaps, is the potential AI displays when applied to text generation. Not only is it capable of producing website text and blog posts, some observers predict that it will soon be handling marketing and sales communications, and, eventually, most interactions between companies and their customers. Beyond this, generative language models will do much of journalists’ writing work; they will draw up legal contracts; be used to create educational materials and support student learning; they will aid clinicians by summarizing patients’ medical records and suggesting treatment regimens; and they will power academic research by mining vast amounts of literature to produce new insights and understanding.

It is clear, then, that this technology does possess immense transformative potential. But there are also real dangers. These include fraudulent activity and identity theft; disinformation campaigns; deep fakes; and violent and disturbing images and videos. There are also serious concerns about copyright infringement and structural unemployment. Nonetheless, these challenges will have to be navigated as the technology is here to stay — and, consequently, we can expect to see VC funding flooding into AI startups over the next few years.