A file photo of French entrepreneurs Pierre Foutrell in front of the artwork “La Comtesse de Bellamy” (2018) created by Generative Adversarial Networks (GAN) called Obvious. , Photo Credit: Reuters
(This article is part of Today’s Cash, The Hindu’s newsletter on emerging topics at the intersection of technology, innovation and policy. Subscribe to get it delivered to your inbox. Here,
It seems like it’s the season for all things AI. ChatGPT has been in news since the beginning of this month. Over one million users have explored the large language model with a wide variety of question types. And the giant chatbot answered almost all questions with Wikipedia-like ease (though some of its answers were wrong).
OpenAI’s bot was an upgrade to its predecessor GPT-3, which launched two years ago, and is now licensed exclusively to Microsoft. It was the first language model to show that an AI can write like a human. This new bot based on GPT-3.5 has the ability to mix text and computer code. This means that it not only answers normal text questions and answers, but also fixes (in some cases incorrect) programming code.
GPT-3.5 and its predecessors are trained on sentences, words, parts of speech, loads of data from social media posts, Wikipedia entries, and news articles. Last year, Sam Altman’s OpenAI revealed another AI use case with the Dall-E app. The 12 billion parameter version of the machine learning model can convert text to images. ChatGPT and DAL-E is when Big Data meets super-fast computation.
While ChatGPT is purely a text-based interface, Dall-E uses text input to process image-based output. These bots are one end of the AI spectrum. They are built keeping the interaction between machines and humans at the center.
On the other end is the AI-based app Lensa AI. It uses a machine learning model, stable diffusion, to generate artistic portrait images based on an input of two dozen selfie photos. But, the ML model uses digital images scraped from the web to process the artistic output. It’s a well-equipped greenroom, but uses other creators’ tools to make the user look good digitally.
The app has been created by Prisma Labs. Six years ago, the company became popular when it allowed users to turn their selfies into portraits in the style of famous artists. Lensa AI is now built on an image-based deep learning model that uses a large amount of digital art from the web and the LAION-5B database to train its machine learning engine.
Artists whose works are scraped by this engine are unable to opt-in or -out of this algorithm. This has been the case even when Stable Diffusion was first introduced earlier this year. Some artists raised concerns that their images were being used by the algorithm without giving them credit or payment. Their complaint did not end there, artists also felt threatened that such ML models could threaten their livelihood as deep neural networks can create digital art at a much faster rate and scale than human artists.
Such complaints haven’t stopped Prisma from building a Lensa AI for general use. Within weeks of its launch, the app achieved the top position in Apple’s App Store. The annual paid version costs Rs 2,499 in India. But the company offers a seven-day free trial to play with the app. The app suggests that users submit 10-20 selfies for best results. Images should be close-up shots of the user’s face, and they should have a variety of backgrounds, facial expressions, and angles.
Once users submit images, small concentric circles zoom in and out against the black background of the app. After processing the images, the viral photo-editing app creates a “magical avatar” using the person’s facial features.
In 2018, Lensa launched its first photo editing tool. And now, this recent launch with the Magic Avatar feature helps transform digital self-portraits into artistic images across a variety of subjects. The app is a great tool for editing photos. And the free version has a three-image limit per day.
The photo editor has some privacy concerns, though its policy notes that it doesn’t use uploaded photos for anything other than applying filters and effects. However, it says it can use the information to train its algorithms to make the app more effective. It also mentions that it will collect data to personalize content.
And when a user signs the app’s terms of use, they grant Lensa AI a “perpetual, irrevocable, non-exclusive, royalty-free, worldwide, fully paid-up, transferable, sublicensable license to use make, reproduce, modify, distribute derivative works of your User Content, without any additional compensation to you and always subject to your additional express consent for such use where required by applicable law.
Now, if you’re a digital artist who plans to use some greenroom touch ups with the Lensa AI for your art work, those terms will return as you grant the app an irrevocable right to your content. will do.