AI companies are covered by a “fair use” doctrine for content generation—they reap without sowing
By Shubhranshu Singh
ChatGPT has caused huge excitement and great anxiety. It will change the way we engage with the internet. Searching for information and processing it have been disrupted. It will upend the model that runs the tech marketplace today and change its underlying economics. Many of today’s leaders will be pauperised whereas parvenu operators will grow into giants in record time.
It is well-appreciated that a new technology can deeply impact the world in permanent ways. In this case, the pace and trajectory will be like that of a crazy ball. The shifts will be even more significant on the second and third bounces. Cloud computing companies, social platforms, search engines, e-commerce giants, and many more will have to find ways to change and survive.
The now-commercialised AI systems could disrupt $100 billion in cloud spending, $500 billion in digital advertising, and $5.4 trillion in e-commerce sales, per IDC, a market research firm, and GroupM, a media agency.
Google makes almost $200 billion on display ads, the most on the planet. Now, this search audience may go away. What should do a website or publisher do if the relationship is not reciprocal? If chatbots lift information but send few visitors, how does that adverse equation get compensated? How can AI companies be prevented from taking content without hampering search rankings?
Microsoft, Amazon, and Google are rushing to be the engine behind the AI chatbots. Google introduced its chatbot, Bard, last month. Microsoft, through its acquisition of OpenAI lab, may pioneer sale of technology to others who look to have a play in the space.
The origin and authorship of content—words, sounds, images, concepts, ideas—is now suspicious. The very concept of ‘original’ is questionable. Where does the content you get from a chatbot come from? Is it ‘fair trade’ in the digital sense? Getty Images has taken Stability AI, the start-up behind the art generator tool Stable Diffusion, to court.
It has thrown open the question or copying and originality. AI companies have the cover of the “fair use” doctrine for content generation—they reap without sowing.
AI-generated images from Stable Diffusion are impossible to tell apart stylistically from the original artist. AI works out patterns, styles, and relationships by examining billions of images on the internet.
AI text-to-image generators such as Stable Diffusion, Midjourney, and DALL-E exploded onto the scene this year and, within months, have become widely-used. Stable Diffusion alone has more than 10 million daily users. These AI products are built on collections of images known as data sets, from which a detailed map of their contents, the model, is formed by finding the connections among images and between images and words.
Images and text are linked in the data set, so the model learns how to associate words with images. It can then make a new image based on the words you type in. Stability AI recently raised $101 million from investors, and is now valued at well over $1 billion.
The data set for Stable Diffusion is called LAION 5b and was built by collecting around 6 billion images from the internet in a practice called data scraping. LAION was able to scour what seems like the entire internet because it deems itself a non-profit organisation engaging in academic research. While it was funded at least in part by Stability AI, the company that created Stable Diffusion is technically a separate entity. Stability AI then used its non-profit research arm to create AI generators first via Stable Diffusion and then commercialised it in a new model called DreamStudio.
AI generators were built on the backs of copyrighted work, and through a legal loophole, they were able to produce copies with varying levels of sophistication.
Artists posting new art feed an ever-hungry engine that becomes more and more like them. Since origination is de novo, the copyright doesn’t quite come into play. Its not a copy, it’s a stylistic inspiration.
Emad Mostaque, the founder of the Stable Diffusion image generator, was a fund manager before he turned into a tech disruptor. He claims it is one of the ultimate tools for freedom of expression. The defenders say AI is an extension of the human usage of tools, only faster, better.
It isn’t a tool that makes things by itself, rather, it is the intention of the end user. Thus, there’s nothing that you can make with Stable Diffusion that you can’t make with Photoshop—it’s just become a bit easier.
An algorithm can make cannon fodder of a lifetime worth of work. These algorithms are learning like a human brain, only billions of times faster, more powerful and untiring. This is the future of content.
Will it be man with a multiplier or man vs monster?
The writer is vice president, marketing-domestic & international business, Tata Motors
Views are personal
Link: https://www.financialexpress.com/opinion/ai-shadow-on-ipr/3047562/?utm_source=Whatsapp