How Fast Is AI Actually Progressing, and What Does It Mean for Glaucoma Patients and Researchers?

About this episode

This audio article is from VisualFieldTest.com.

Read the full article here: https://visualfieldtest.com/en/how-fast-is-ai-actually-progressing-and-what-does-it-mean-for-glaucoma-patients-and-researchers

Test your visual field online: https://visualfieldtest.com

Support the show so new episodes keep coming: https://www.buzzsprout.com/2563091/support

Excerpt:

How Fast Is AI Actually Progressing, and What Does It Mean for Glaucoma Patients and Researchers?Artificial intelligence (AI) has been advancing at breakneck speed in recent years. New AI models now perform tasks once thought years away, and these leaps are reflected in benchmarks, products, and research breakthroughs across many fields – including eye care. This article examines concrete measures of AI progress and translates them into what they mean for glaucoma care and research. We highlight real examples of AI tools already helping patients, summarize what new developments are on the horizon (from clinical trials to near-future innovations), and suggest questions patients and researchers can explore today to prepare for tomorrow’s advances. How is AI Progress Measured (and How Fast Is It Growing)?Researchers measure AI progress by performance on challenging tasks (benchmarks) and by tracking improvements in model design, data, and compute. In the last few years, all three of these factors have exploded. For example, one analysis found that the “frontier” of AI capabilities accelerated sharply around 2024 – roughly doubling its rate of improvement compared to prior years () (). In basic terms, AI systems can now solve problems almost twice as fast or as well as they could just a couple of years ago. Why is this happening? Since 2010, the computing power used to train leading AI models has roughly doubled every six months (), creating a 4–5× growth in compute per year. Training data sets (like text or images) have also been exploding – data sets roughly triple in size each year (). At the same time, model sizes (number of parameters) have been doubling annually. These three trends – massive compute, massive data, massive models – combine to create what some call a “trifecta” of rapid AI scaling (). The result is that capabilities often jump in swarms. State-of-the-art AI models that struggled with basic reasoning tasks even a couple of years ago are now solving mathematically complex problems, generating realistic images on demand, and even engaging in fluent medical knowledge conversations. For example, large language models (LLMs) like OpenAI’s GPT series have shown sudden leaps in abilities at specific size thresholds (). Each new generation (GPT-3 → GPT-4 → GPT-4.5, etc.) has outperformed the last on a wide range of benchmarks. Specialized systems for vision (image) tasks have also surged, with diffusion models and neural networks now producing realistic images or detecting subtle patterns with unprecedented accuracy. In short, the pace of improvement is not a slow linear climb – it’s accelerating in both raw metrics and real-world impact () (). Key takeaway: AI progress is concrete and measurable, and in the last 2–3 years performance on standard benchmarks and practical tasks has nearly doubled. This means new tools that were science fiction a decade ago are arriving faster than many expect.AI in Glaucoma Care TodayGlaucoma is a leading cause of irreversible vision loss worldwide, and it’s increasingly clear that AI can help us detect and manage it. Several AI-powered tools are already making their way into practice or close to it:AI-enhanced

Support the show