The realm of AI-generated art and the emergence of deep fakes have revolutionized the creative landscape. Envisioning everything from whimsical structures to surreal furniture and resurrecting ancestral portraits, the likes of Midjourney, DALL·E 2, and Stable Diffusion text-to-image algorithms have become indispensable tools for artists worldwide. Through the fusion of precise prompts and the boundless scope of human creativity, these algorithms have birthed astonishing visuals that have taken the internet by storm. Everypixel, the stock image search engine, recently published a comprehensive report that quantifies this burgeoning phenomenon: in a mere 12 months, text-to-image algorithms have generated an astonishing 150 billion images. According to Everypixel, the post-DALL·E 2 era has witnessed a staggering average of 34 million images being produced daily, a statistic that underscores the incredible potential — and simultaneous concerns — associated with machine learning's capabilities.
Everypixel's report provides a deep dive into the impact of various text-to-image software, spotlighting Midjourney, DALL·E 2, Stable Diffusion, and Adobe Firefly. DALL·E 2, introduced by OpenAI in April 2022 on an invite-only basis, transitioned to an open platform by September 2022, generating two million images daily. Though the exact calculation methodology remains elusive, Everypixel posits that approximately 916 million images have been created on this singular platform over 15 months. Alongside this, Adobe Firefly, a text-to-image feature for Photoshop, churned out over 100 million assets within six weeks of its March 2023 launch, and an astounding one billion visuals in a mere three months. Integration with Adobe Photoshop in May 2023 propelled image creation further, given the software's popularity among creatives globally. Collectively, Stable Diffusion, Adobe Firefly, Midjourney, and DALL·E 2 have generated over 15 billion AI-crafted images, surpassing Shutterstock's entire library and comprising a third of Instagram's image count.
However, amid the jubilant strides in AI-driven creation, a darker undercurrent emerges. Viral images depicting the Pope in unconventional attire or the fictional arrest of Donald Trump illustrate the formidable power of AI to fabricate reality. These creations have spread like wildfire, raising doubts about their authenticity. Notably, generative AI has evolved from photorealistic images to crafting lifelike videos and even entire movies from mere textual descriptions. This trajectory suggests that AI's momentum is far from waning. While experimenting with AI is undeniably captivating, the broader truth remains: its potential for misinformation poses a significant threat, necessitating a nuanced approach as this technology continues to evolve.
Comments