Despite all its enormous potential, the field of artificial intelligence has been something of a backwater in the investment world. Some companies have ridden the AI wave in a big way: Google claims to have refined many of its services using AI, machine learning has boosted sales of Nvidia’s graphics processing units and the TikTok’s algorithm is said to be a big part of what keeps users coming back to its short videos.
But it’s hard to find a pure AI company that has grown through technology, or to identify a significant new market that has been created. This image may be about to change, and in a big way.
According to Pat Grady, partner at Sequoia Capital, something big has happened in AI in recent weeks. Generative systems – those that automatically produce text and images from simple text prompts – have reached a level where they could have many commercial uses. A partner at another top Silicon Valley venture capital firm, who describes recent AI history as a graveyard for start-up investors, also reports that the race is on to find game-changing applications. for this new technology.
Since the launch of OpenAI’s GPT3 text writing system two years ago, generative models like this have been all the rage in AI. The ethical issues they raise run deep, ranging from any bias they might imbibe from the data they are trained on, to the risk that they could be used to spread misinformation. But that hasn’t stopped the hunt for practical uses.
Three things have changed to turn these nifty party trick systems into potentially useful tools.
The first is that AI systems have gone beyond text. Last week, Meta unveiled the first system capable of outputting video from a text or image prompt. It was thought that this breakthrough would be in two years or more. Not to be outdone, Google responded with not one but two AI video systems of its own.
The biggest AI breakthrough this year has been in image generation, thanks to systems such as OpenAI’s Dall-E 2, Google’s Imagen and start-up Midjourney. Emad Mostaque, the London-based hedge fund manager behind Stable Diffusion, the latest image-generating system to take the AI world by storm, says images will be the “killer app” for this new form of AI. For the generation that grew up on TikTok and Snapchat, this new creative tool could be profound, he says. It also presents an obvious threat to anyone whose livelihood relies on creating images by other means.
The second big change comes from the rapidly falling cost of training giant AI models. Microsoft’s $1 billion backing of OpenAI three years ago highlighted the prohibitive expense of this for ever-larger designs. New techniques that achieve high-quality results by training neural networks with fewer layers of artificial neurons are changing the game. The computing resources used to form Stable Diffusion would have cost only $600,000 at market prices, according Mustache.
The third change was the availability of technology. Google and OpenAI have been reluctant to make their technology widely available, in part for fear of possible misuse. In contrast, Midjourney’s image system is available to all users through a freemium pricing model. Stable Diffusion went further, open source its software and publish details of how it trained its system. This allows other organizations to train an image model on their own datasets.
The risks that arise from these generative systems have received much attention. They produce new images or new text based on the millions of examples they have learned from, without any understanding of the underlying material. This can lead to absurd results, as well as deliberate misinformation.
But in a professional environment, at least some of these shortcomings could be controlled. The trick will be to find ways to integrate technology into existing work processes, creating tools that can suggest new ideas or speed up creative output, with human workers filtering the output. The idea is already being used to generate computer code.
The big question now, says one investor, is: will existing giants in industries such as marketing, media and entertainment be the first to use these powerful new creative tools? Or will they be disrupted by a new generation of newcomers with their roots in AI?
richard.waters@ft.com
#AIs #Sudden #Big #Leap #Utility