The Argonne Leadership Computing Facility is home to the AI Testbed. Researchers can access platforms, including this Cerebras CS-2, to study how machine learning-based high performance computing applications will interface with today鈥檚 supercomputers to accelerate scientific discovery.
Newswise — Chat-GPT, DALL-E and other large language models (LLMs) are causing seismic shifts in the way people create and communicate. Users can interact with these 鈥�“chat bot” LLMs by asking or typing a question in plain language, rather than by learning a complicated computer code. And the results come nearly instantaneously. But we’ve only scratched the surface of what this technology can do.
Researchers from the U.S. Department of Energy’s (DOE) Argonne National Laboratory are at the forefront of harnessing artificial intelligence (AI) and LLMs to speed up the rate of scientific discovery and to change the way that people do science. They’re training LLMs to solve problems across different scientific disciplines. They’re also training human scientists to get the most out of AI.
Researchers have described LLMs as everything from an eager new assistant to a really smart baby.
During a recent workshop on artificial intelligence tools, Argonne computer scientists answered our most pressing questions about large language models for science. The answer to number 5 might shock you!
1. What is an LLM?
You’re using these AI tools. You kind of know what they are. But do you really know how they work? Can you explain it to someone in under a minute? Get ready to sound smarter at work and impress people at cocktail parties.
2. How are LLMs being used for science?
From a technical perspective, training LLMs means supplying the model with the data you need it to know. Depending on the results you’re trying to get, a computer scientist could potentially train a model on something like every song lyric ever written by Taylor Swift or everything humans have ever posted on the internet. But neither of those LLMs would be good at helping us cure cancer.
3. What is prompt engineering?
We’re filling LLMs with vast amounts of information, but how do we get quality results out? From a human perspective, scientists are researching how to ask LLMs the right questions in the right way. This process is called prompt engineering and it’s more nuanced than Googling something or yelling, 鈥�“Hey, Siri!”
4. Should you trust information that comes out of an LLM?
Researchers have described LLMs as everything from an eager new assistant to a really smart baby. They want to make you happy. They want to give you what you need quickly. That doesn’t mean that what they give you is correct or that it even makes sense. How do we stop worrying and love LLMs?
5. How are LLMs shaping the future of science?
Some technological advances are such game changers that you have to throw out the whole gameboard you were playing on. The automobile, the internet and the cell phone reshaped major aspects of our lives and our world. How will artificial intelligence and LLMs completely change the way we do research? Spoiler alert: we don’t know. But a lot of smart people are working to figure it out.
While we might not know everything we ever wanted to know yet, thousands of scientists, engineers and staff from DOE national labs, academia and technology companies have collaborated to a develop roadmaps to harness AI in scientific discovery. Learn more about the rapidly emerging opportunities and challenges for science, energy and security from Argonne’s AI Reports.
Argonne National Laboratory seeks solutions to pressing national problems in science and technology by conducting leading-edge basic and applied research in virtually every scientific discipline. Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.
The U.S. Department of Energy’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit 丑迟迟辫蝉://鈥媏苍别谤鈥媑测鈥�.驳辞惫/鈥媠鈥媍鈥媔别苍肠别.
RSS