AI News #86

Adobe Photoshop vs. Generative AI

Story: A neat article discusses the ongoing debate between traditional image editing tools like Adobe’s Photoshop and newer AI-based image editors. While Photoshop has long been the industry standard, AI-driven tools are gaining popularity for their ease of use and ability to automate complex tasks. The article highlights how AI tools are reshaping the landscape of digital creativity, making sophisticated editing accessible to a broader audience. You can find the most important arguments below:

Key Findings:

• "You can already manipulate images”: This argument acknowledges that AI image editors are simple to use, allowing users to perform complex edits with minimal effort. This democratizes image editing, making it accessible to non-professionals who may find Photoshop's detailed controls and steep learning curve daunting.

• "Photoshop was a huge, barrier-lowering tech, too - but we ended up being fine”: The argument suggests that, like Photoshop, AI tools may initially disrupt but will eventually integrate into the creative ecosystem without catastrophic effects. The key difference is the level of automation AI introduces, which could amplify both the positive and negative impacts.

• "I'm sure laws will be passed to protect us”: This argument reflects a cautious optimism that legislative frameworks will catch up to technological advances. It posits that as AI tools become more prevalent, governments will step in to regulate their use, especially in areas prone to abuse like deepfakes and misleading advertising

Pixit‘s Two Cents: We couldn't agree more with the arguments. The democratization of image editing tools is something we see clearly in our own work. While we still use Photoshop occasionally, most tasks are now easier and produce better results with AI-based tools. Another important point is scalability. AI enables the creation of fully automated applications, such as those used for product imagery—something that wouldn’t be feasible with Photoshop alone.

Perplexity’s Latest Update: More Flux and Fancy Charts

Story: Perplexity has rolled out updates aimed at enhancing the user experience for both casual and professional users. The company introduced two new image generators, FLUX.1 and Playground v3, for Pro users and improved the code interpreter, including the ability to generate charts.

Key Findings:

• Image Generation: Perplexity has incorporated FLUX.1 and Playground v3 for image generation. Both models are available to Pro users only. You can change the image generator by clicking on Settings (lower left) > Image Generation Model.

• Code Interpreter: Perplexity can now generate charts based on code inputs, making it a tool for data analysis and visualization. Additionally, you can now install libraries. 

Pixit‘s Two Cents: We We love to see more and more companies adopting FLUX.1 - simply because it is a game-changer for users who need more advanced image generation capabilities.

ALS. Stole His Voice. AI. Retrieved It.

Story: Four years ago, Casey Harrell’s life took a turn when ALS (Amyotrophic lateral sclerosis) stole his ability to speak, sing, and communicate effectively. Now, an AI has given him a new voice. Researchers at the University of California, Davis implanted electrodes in his brain, enabling a device to decode his intended speech directly from his brain activity. The implant has not only restored his ability to communicate but has done so with an accuracy that surpasses even the best smartphone transcriptions. This breakthrough marks a significant advancement in brain-computer interface technology, offering new hope for those with severe speech impairments.

Key Findings:

• Restoration of Communication: The implant allows Mr. Harrell to communicate with his family and friends in a voice remarkably similar to his pre-ALS voice, providing both emotional relief and a renewed sense of connection with his loved ones.

• Technological Milestone: The device demonstrates the rapid advancements in brain-computer interfaces, achieving an unprecedented level of accuracy and expanding the vocabulary it can interpret. This development showcases the potential of such technology to significantly improve the lives of people with speech impairments. 

Pixit‘s Two Cents: We are deeply moved by the story and the incredible potential that this technology represents. The ability to restore communication for those with severe impairments is a powerful reminder of the positive impact that technological advancements can have on individual lives.

Small Bites, Big Stories:

• Nvidia’s Llama-3.1-Minitron 4B is a small language model that punches above its weight: Nvidia released a compressed version of the LLama 3 model using pruning and distillation.

• Microsoft reveals Phi-3.5 — this new small AI model outperforms Gemini and GPT-4o: Microsoft has released the latest version of its small language model Phi-3.5.

• Fine-tuning now available for GPT-4o: You can now fine-tune GPT-4o on your own datasets.

• Introducing World’s First Automatic and AI-powered Deepfake Detector: McAfee released the Deepfake Detector on selected Lenovo AI PCs to detect AI-altered videos.

• Apple and Nvidia may invest in OpenAI: The tech giants might join OpenAI.

• Google Gemini will let you create AI-generated people again: Finally! You can create AI-generated people using Google Gemini (after Google stopped the generation of people after producing historically inaccurate images)

• Before Midjourney, there was NightCafe — and it’s still kicking: NightCafe is an AI-powered art-creation tools that many don’t know.

• The org behind the dataset used to train Stable Diffusion claims it has removed CSAM: LAIN, the German research org that create the dataset used to train Stable Diffusion, has released a new dataset that has been cleaned of known links to suspected child sexual abuse material (CSAM)