AI speeds scientific discovery: The scientist that never sleeps

Image credit

AI speeds scientific discovery: The scientist that never sleeps

AI speeds scientific discovery: The scientist that never sleeps

Subheading text
Artificial intelligence and machine learning (AI/ML) are being used to process data faster, leading to more scientific breakthroughs.
    • Author:
    • Author name
      Quantumrun Foresight
    • December 12, 2023

    Insight summary

    AI, especially platforms like ChatGPT, is significantly accelerating scientific discovery by automating data analysis and hypothesis generation. Its ability to process vast amounts of scientific data is crucial for advancing fields like chemistry and materials science. AI played a pivotal role in developing the COVID-19 vaccine, exemplifying its capacity for fast, collaborative research. Investments in "exascale" supercomputers, like the US Department of Energy's Frontier project, highlight AI's potential in driving scientific breakthroughs in healthcare and energy. This integration of AI into research promotes multidisciplinary collaboration and rapid hypothesis testing, although it also raises questions about the ethical and intellectual property implications of AI as a co-researcher.

    AI speeds scientific discovery context

    Science, in and of itself, is a creative process; researchers must constantly expand their minds and perspectives to create new medicines, chemical applications, and industry innovations at large. However, the human brain has its limits. After all, there are more conceivable molecular forms than there are atoms in the universe. No person may examine all of them. This need to explore and test the infinit diversity of possible scientific experiments has pushed scientists to continually adopt novel tools to expand their investigatory capabilities—the latest tool being artificial intelligence.
    The use of AI in scientific discovery is being driven (2023) by deep neural networks and generative AI frameworks capable of generating scientific knowledge in bulk from all published material on a specific topic. For instance, generative AI platforms like ChatGPT can analyze and synthesize vast amounts of scientific literature, assisting chemists in researching new synthetic fertilizers. AI systems can sift through extensive databases of patents, academic papers, and publications, formulating hypotheses and guiding research direction.

    Similarly, AI can use the data it analyzes to devise original hypotheses to broaden the search for new molecular designs, at a scale that an individual scientist would find impossible to match. Such AI tools when coupled with future quantum computers would be capable of rapidly simulating new molecules to address any specified need based on the most promising theory. The theory will then be analyzed using autonomous lab tests, where another algorithm would evaluate the results, identify gaps or defects, and extract new information. New questions would arise, and so the process would begin again in a virtuous cycle. In such a scenario, scientists would be overseeing complex scientific processes and initiatives instead of individual experiments.

    Disruptive impact

    One example of how AI has been used to speed up scientific discovery was the creation of the COVID-19 vaccine. A consortium of 87 organizations, ranging from academia to tech firms, has allowed global researchers to access supercomputers (devices with high-speed computing capabilities that can run ML algorithms) to use AI to sift through existing data and studies. The result is a free exchange of ideas and experiment results, full-access to advanced technology, and faster, more accurate collaboration. Further, federal agencies are realizing the potential of AI to rapidly develop new technologies. For example, the US Department of Energy (DOE) has asked Congress for a budget of up to USD $4 billion over 10 years to invest in AI technologies to boost scientific discoveries. These investments include “exascale” (capable of performing high volumes of calculations) supercomputers.

    In May 2022, DOE commissioned tech firm Hewlett Packard (HP) to create the fastest exascale supercomputer, Frontier. The supercomputer is anticipated to solve ML calculations up to 10x faster than today’s supercomputers and find solutions to problems that are 8x more complex. The agency wants to focus on discoveries in cancer and disease diagnosis, renewable energy, and sustainable materials. 

    DOE has been funding many scientific research projects, including atom smashers and genome sequencing, which has resulted in the agency managing massive databases. The agency hopes this data could one day result in breakthroughs that can advance energy production and healthcare, among others. From deducing new physical laws to novel chemical compounds, AI/ML is expected to do the brunt work that would take away ambiguities and increase the chances of success in scientific research.

    Implications of AI speeding scientific discovery

    Wider implications of AI speeding scientific discovery may include: 

    • Facilitating the rapid integration of knowledge across different scientific disciplines, fostering innovative solutions to complex problems. This benefit would encourage multidisciplinary collaboration, blending insights from fields like biology, physics, and computer science.
    • AI being used as an all-purpose laboratory assistant, analyzing vast datasets much faster than humans, leading to quicker hypothesis generation and validation. Automation of routine research tasks will free up scientists to focus on complex problems and analyzing tests and experiment results.
    • Researchers investing in giving AI creativity to develop their own questions and solutions to scientific inquiries in various fields of study.
    • Accelerating space exploration as AI will assist in processing astronomical data, identifying celestial objects, and planning missions.
    • Some scientists insisting that their AI colleague or co-researcher should be given intellectual copyrights and publication credits.
    • More federal agencies investing in supercomputers, enabling increasingly more advanced research opportunities for university, public agency, and private sector science labs.
    • Faster drug development and breakthroughs in materials science, chemistry, and physics, which can lead to an infinite variety of future innovations.

    Questions to comment on

    • If you’re a scientist or researcher, how is your organization using AI in research?
    • What are the potential risks of having AI as co-researchers?