Since ChatGPT's launch in 2022, much of the discussion around AI has been in the domain of AI-generated content. The technology is improving efficiency and innovation in leaps and bounds across the software development, marketing, and entertainment industries.
But while generative AI gets a lot of face time, there are many new applications of other types of AI on a daily basis --- so many that it's almost impossible to keep up. And each advance in one of these fields helps create technology that other fields can use.
This ripple effect permeates everything, including science. The importance of AI in the development of products, services, and entertainment is important, but its usage in scientific fields has incredible potential for the good of humanity, the environment, and the world we offer future generations. Here are five ways that generative AI is advancing science and leading breakthroughs.
1: Studying the sun with AI.
The exterior atmosphere of our sun is hotter than the inner layer. For decades, solar scientists have tried to figure out why. Now, they're using AI to help solve the sun's mystery.
Scientists had data collected from the Atmospheric Imaging Assembly (AIA) instrument on NASA's Solar Dynamics Observatory, and images captured by NASA's Interface Region Imaging Spectrograph (IRIS) satellite. From there, scientists taught an AI machine learning algorithm to enhance the images, giving them a complete picture of the sun's outer and inner layers.
The AI-enhanced images allow scientists to see more detail to help them understand more about why the sun's atmosphere behaves the way it does. While the explanation of why the sun's exterior is hotter than its interior still remains unknown, Scientists anticipate that machine learning will help them get much closer to the answer.
2: Providing publicly accessible climate change data.
NASA, IBM, and Hugging Face have partnered to make climate data accessible to everyone. The association began when NASA combined its satellite data with IBM's watsonx.ai geospatial foundation model. Then NASA and IBM made the model available on open source Hugging Face, making it the first open-source AI foundation model built in collaboration with NASA.
These partnerships resulted from NASA's commitment to building a more accessible, inclusive, and collaborative scientific community. The agency chose to open source the model because they hoped greater availability would multiply the impact on the world.
3: Improving healthcare diagnostics, treatment, and personalized care.
Healthcare practitioners are using AI tools to improve patient care and services. One such tool is Corti.ai, which helps emergency care providers diagnose and treat cardiac arrest. Several hospitals and clinics also use AI to assist doctors in reviewing medical images and scans, for screenings, and to assist in diagnosis.
In the next five years, the American Hospital Association (AHA) predicts that AI tools will help support patient care by analyzing a patient's medical history data and making suggestions for care. With a projected shortfall of 18.2 million healthcare workers by 2030, AI will likely continue to expand to help fill in the gaps.
4: Transforming science by speeding research.
Scientific research takes time. The average scientific study takes from five to 17 years or longer to deliver results, depending on the subject. This timeline includes everything from the initial hypothesis to data collection, experimentation, analysis, and publication of findings. While scientists can't skip steps, they can rely on AI to speed up some processes, like data collection or analysis. Some ways AI helps scientific research include:
Summarizing research papers and transform study results into manuscripts.
Ingesting large amounts of unstructured data --- like satellite images and global weather data --- and developing causal models that make it easier for scientists to analyze.
Enhancing low-resolution electron microscopic images to become high-res images to improve quality without buying higher-cost equipment.
Solving complex physics problems just as accurately as scientists, but faster. Eventually, AI may be able to solve complex problems that humans can't.
Enabling researchers to communicate with colleagues from different backgrounds and languages, allowing scientists to collaborate around the world.
5: Developing new pharmaceutical drugs faster.
Clinical drug trials can last from ten to 15 years or longer. But with AI, researchers can safely accelerate parts of the process and bring new drugs to market faster. AI can predict toxicity profiles and interactions of trial candidates, help identify potential issues sooner, and process massive amounts of data. These steps can reduce the time and cost that goes into pharmaceutical research and the development of new drugs.
Our Future World with AI
Just like we've seen in other industries, AI has already proven to be a valuable tool to help scientists, doctors, and researchers. It's still early days with using AI in the science field, and we don't yet know everything it's capable of. However, with these successes, it's entirely possible that we'll see AI in more use cases, which could drive more discoveries, fuel interest in further AI development, and improve outcomes for people everywhere.