Diffusion Models

Twitter

Facebook

Copy Link

Key Facts

Year2015

Key PeopleJascha Sohl-Dickstein (2015), Jonathan Ho (2020)

Top News

AI Video Generators Transform Text Into Video: How the Technology Works and What You Need to Know

AI video generators now let anyone create videos from text in minutes. These tools use diffusion models and transformers to turn written prompts into moving scenes. But success depends on writing effective prompts, managing consistency, and understanding content policies that govern what can be generated.

Technology3 sources

AI Video Generators Transform Text Into Video: How the Technology Works and What You Need to Know

Technology3 sources

Inception Mercury 3 Diffusion Model Claims 8X Speed Advantage Over GPT-6 Luna and Claude Haiku 5.5

Redwood City AI lab Inception released Mercury 3, a diffusion-based large language model that generates more than 1000 tokens per second on Nvidia chips. The model outperforms GPT-6 Luna and Claude Haiku 5.5 across benchmarks while running four to eight times faster, with a one million token context window and launch pricing of $0.10 per million input tokens.

Technology2 sources

Inception Mercury 3 Diffusion Model Claims 8X Speed Advantage Over GPT-6 Luna and Claude Haiku 5.5

Technology2 sources

Scientists Build AI Model That Reconstructs Images From Brain Scans in Record Time

Researchers at the Weizmann Institute of Science unveiled Brain-IT, an AI model that reconstructs images from brain activity with unprecedented speed and accuracy. The system needs only 1 hour of fMRI data to match results that previously required 40 hours, raising both medical possibilities and privacy concerns.

Science4 sources

Scientists Build AI Model That Reconstructs Images From Brain Scans in Record Time

Science4 sources

Black Forest Labs CEO Calls for Europe to Shift AI Development Mindset Amid Global Competition

Robin Rombach, CEO of $3.25 billion AI startup Black Forest Labs, urges Europe to adopt an optimistic approach to AI development instead of focusing on safety fears. The German company, known for its FLUX models and latent diffusion technology, argues that capital access and open-weight AI models are essential for Europe to compete with US and Chinese AI leaders.

Business2 sources

Black Forest Labs CEO Calls for Europe to Shift AI Development Mindset Amid Global Competition

Business2 sources

MIT Study: AI-Generated Images Often Untraceable to Original Training Data Sources

MIT researchers discovered that AI-generated images from large diffusion models like Midjourney and Stable Diffusion cannot be traced to specific training data due to attribution decay. The phenomenon means removing individual artworks or entire artist portfolios from training datasets produces identical outputs, raising questions about copyright lawsuits and intellectual property claims against AI companies.

Science5 sources

MIT Study: AI-Generated Images Often Untraceable to Original Training Data Sources

Science5 sources

Researchers find training humans to spot deepfakes boosts accuracy from 40% to 80% in just one hour

New research reveals that people can barely distinguish AI-generated faces from real ones, with accuracy rates hovering around coin-flip levels. But there's hope: researchers from the University of Aberdeen and Australian National University found that focused training on subtle perceptual cues can dramatically improve detection rates in just an hour, raising accuracy from 40% to 80%.

Science4 sources

Researchers find training humans to spot deepfakes boosts accuracy from 40% to 80% in just one hour

Science4 sources

Stanford's AI designs burgers that beat Big Mac while slashing environmental impact

Stanford researchers developed BurgerAI, a generative AI system that creates burger recipes optimized for taste, nutrition, and sustainability. In blind taste tests with over 100 diners, AI-designed burgers matched or exceeded a popular fast-food burger in flavor while reducing environmental impact by more than an order of magnitude. The breakthrough demonstrates AI's shift from prediction to design.

Science4 sources

Stanford's AI designs burgers that beat Big Mac while slashing environmental impact

Science4 sources

Google DeepMind releases DiffusionGemma AI model with 4x speed boost using image generation techniques

Google DeepMind unveiled DiffusionGemma, an experimental open-source AI model that generates text in parallel rather than sequentially. The model achieves over 1,000 tokens per second on NVIDIA H100 GPUs and runs on consumer hardware with just 18GB of memory. However, the speed gains come with a notable trade-off in output quality compared to traditional models.

Technology12 sources

Google DeepMind releases DiffusionGemma AI model with 4x speed boost using image generation techniques

Technology12 sources

Rogue One Director Gareth Edwards Embraces AI in Filmmaking, Calls It Better Than CGI

Gareth Edwards publicly endorsed generative AI in filmmaking at Amazon's AI on the Lot event, describing it as a tool that could rival the camera itself. The Rogue One and Jurassic World Rebirth director has been experimenting with AI for nine months and believes it excels at iteration and pre-production, though he cautions it lacks creative taste and requires constant oversight.

Entertainment2 sources

Rogue One Director Gareth Edwards Embraces AI in Filmmaking, Calls It Better Than CGI

Entertainment2 sources

Real Monet Painting Labeled AI Sparks Viral Social Experiment on Authenticity and Bias

A conceptual artist posted an authentic Claude Monet painting from his Water Lilies series on X, claiming it was AI-generated art. The social experiment drew 6.7 million views and hundreds of detailed critiques attacking the work's composition, color choices, and depth—until the reveal exposed how labels shape artistic perception more than the actual artwork itself.

Entertainment4 sources

Real Monet Painting Labeled AI Sparks Viral Social Experiment on Authenticity and Bias

Entertainment4 sources

Hong Kong scientists develop AI model that predicts severe thunderstorms four hours in advance

Researchers from The Hong Kong University of Science and Technology have created an AI-powered weather forecasting system that predicts dangerous thunderstorms and heavy rainfall up to four hours before they strike, with accuracy improved by over 15%. The breakthrough technology uses satellite data and deep diffusion techniques to transform early warning systems across Asia.

Science2 sources

Hong Kong scientists develop AI model that predicts severe thunderstorms four hours in advance

Science2 sources

Thermodynamic computing could slash AI image generation energy use by ten billion times

Scientists at Lawrence Berkeley National Laboratory have demonstrated that thermodynamic computing could generate AI images using one ten-billionth the energy of current tools like DALL-E and Midjourney. The breakthrough research shows promise for addressing the high energy consumption of generative AI, though significant hardware development challenges remain before the technology can rival existing models.

Science3 sources

Thermodynamic computing could slash AI image generation energy use by ten billion times

Science3 sources

Lemon Slice raises $10.5M from Y Combinator and Matrix to scale real-time AI avatars

Lemon Slice has raised $10.5 million in seed funding from Matrix Partners, Y Combinator, and notable tech leaders to build out its digital avatar technology. The startup's flagship model, Lemon Slice-2, transforms a single image into fully interactive, conversational video characters that can respond in real time, addressing what founders call the 'uncanny valley' problem plaguing existing avatar solutions.

Technology2 sources

Lemon Slice raises $10.5M from Y Combinator and Matrix to scale real-time AI avatars

Technology2 sources

AI image generators converge on just 12 visual clichés, study reveals creative limitations

A study published in Patterns journal reveals AI image generators repeatedly default to the same 12 generic visual motifs when generating images autonomously. Researchers paired AI models in a visual telephone experiment, finding that despite diverse starting prompts, systems like Stable Diffusion XL consistently produced Eurocentric themes like Gothic cathedrals and Parisian nightscapes—what they call 'visual elevator music.'

Science3 sources

AI image generators converge on just 12 visual clichés, study reveals creative limitations

Science3 sources

OpenAI's Sora Video Generator Raises Safety Concerns as AI-Generated Content Becomes Indistinguishable from Reality

OpenAI's Sora AI video generator transforms text prompts into realistic video clips, sparking concerns about misinformation, deepfakes, and child safety as the technology makes it increasingly difficult to distinguish between real and AI-generated content.

Technology3 sources

OpenAI's Sora Video Generator Raises Safety Concerns as AI-Generated Content Becomes Indistinguishable from Reality

Technology3 sources

Prev12 Next