Science

IBM 5 in 5: Radically Accelerating the Process of Discovery will Enable Our Sustainable Future

Share this post:

Our mission at IBM is to help our clients change the way the world works. There’s no better example of that than IBM’s annual “5 in 5” technology predictions. Each year, we showcase five ways we believe technology will fundamentally reshape business and society in the next five years, informed by work occurring within IBM Research’s global labs and broader industry trends.

This year’s 5 in 5 predictions focus on accelerating the discovery of new materials to enable a more sustainable future. In line with the United Nation’s global call-to-action through its Sustainable Development Goals, IBM researchers are working to speed up the discovery of new materials that will address significant worldwide problems. Specifically, we are exploring how technology can be used to reinvent the materials design process to find solutions to such as challenges as fostering good health and clean energy as well as bolstering sustainability, climate action and responsible production.

We believe that, within the next half a decade, new materials or novel uses of existing ones will help address many of the global challenges we face: efficiently capturing carbon dioxide from our overburdened atmosphere, and storing it safely, mitigating climate change; finding more sustainable ways to grow crops to feed our surging population while reducing carbon emissions; rethinking batteries and energy storage before we have to rethink our world; developing more sustainable electronic devices and better antivirals.

Placeholder image

1: Capturing and transforming CO2 to mitigate climate change

Placeholder image

Placeholder image

2: Modeling Mother Nature to feed a growing citizenry while reducing carbon emissions

Placeholder image

Placeholder image

3: Rethinking batteries before we have to rethink our world

Placeholder image

Placeholder image

4: Sustainable materials, sustainable products, sustainable planet

Placeholder image

Placeholder image

5: Learning from our past for a healthier future

Placeholder image

We’re not there yet, but it’s not just wishful thinking. It’s possible to start getting there, and soon.

We need to turbocharge the way we do materials design, a traditionally hard and complex process — mainly because the chemical space of potential molecular combinations is incredibly vast. There are more possible combinations than there are atoms in the universe. Further, the final properties of materials do not depend only on what they’re made of, but also the processes used for their production and, ultimately, their final structure.

It typically takes roughly 10 years and upwards of $10-$100 million on average to discover one new material with specific properties. We want to cut both years and cost 90 percent, with the help of cutting-edge technologies. These technologies are artificial intelligence (AI), data augmentation with traditional, classical computing and emerging quantum computing, and so-called generative models and laboratory automation through the open, hybrid cloud.

The convergence of these technologies will allow us to modernize the human process of discovery in a fundamentally new way — moving away from serendipity, luck and chance to calculated confidence.

First, AI will consolidate all of humanity’s knowledge on a specific topic — say, a global challenge we want to address. Then supercomputers and, eventually, quantum simulations will cover our knowledge gaps. Using past data the AI obtained, we will create models to generate hypotheses about the new materials necessary to tackle that challenge. Finally, we will automate the making and testing of these materials, with the help of cloud technologies. All of IBM’s industry-leading work in quantum, AI, high-performance computing and hybrid cloud is therefore accelerating the process of discovery.

IBM designed an approach to accelerate material discovery where AI is a key component across the entire chain of the material discovery process. This includes its cloud-powered chemistry lab RoboRXN, which allows researchers to create new materials by predicting the outcome of chemical reactions. Materials can be synthesized 24 hours per day, 7 seven days a week, without disruption, and with limited interaction from humans. Scientists simply have to give the system a molecule they want to make, and the AI in the software will outline a step-by-step recipe along with a list of ingredients.

The free-to-use AI model behind RoboRXN was made available two years ago and has already predicted nearly 1 million reactions for students, professors and scientists. Since earlier this year IBM scientists around the world are using RoboRXN to synthesize materials for carbon capture, photoresists and antivirals. It will soon go to work generating materials for nitrogen fixation.

IBM RoboRXN synthesized a COVID treatment molecule (left) and molecule for carbon capture (right).

For example, researchers in Yorktown Heights, New York were able to run a synthesis reaction sent over the cloud to the RoboRXN lab in Zurich — 6,000 kilometers away — to try to generate a target molecule for a new material for CO2 capture. RoboRXN analyzed the resulting molecules, which were synthesized overnight and then shipped to the IBM Research Lab in Almaden, California for further analysis.

Right now, researchers are synthesizing candidate molecules to test RoboRXN’s capabilities. The long-term goal is that we will take candidate molecules from the AI-driven systems that are predicting new materials, synthesize them completely autonomously and test them.

Imagine not just one RoboRXN, but an entire bank of them receiving instructions via the cloud from chemists and material scientists around the world. It’s not just accelerating discovery, it’s turbocharging the production of ideas at scale.

This doesn’t mean that IBM is getting into the material manufacturing business. Our researchers are focused on developing the processes that will speed discovery and helping our clients and collaborators put them to use. One of IBM’s previous 5 in 5 predictions offers a good blueprint for how this can be accomplished.

In 2019, IBM predicted that, in the next five years, plastic recycling advancements like VolCat could be adopted around the globe to combat global plastic waste. VolCat uses a benign organic catalyst to selectively digest the most common household plastic – polyethylene terephthalate (PET) – back to its monomer constituents. After purification, the monomer can easily be re-polymerized to form new PET.

IBM Research plans to kick off the next phase of its plan to commercialize the VolCat plastic recycling process. We intend to team with an industry partner to design, build and operate a pilot plant to prove the scalability and economics of the VolCat process. If successful, that work would progress to manufacturing plants all over the world and enable manufacturers to make plastics, fibers or films out of the resulting monomers, without the need to create new plastics from petrochemicals.

IBM’s 5 in 5 demonstrates what’s possible when a scientific approach is applied to finding new methods and solutions to the world’s challenges. The world needs science more urgently than ever and, with it, we can face today’s uncertainties to realize tomorrow’s progress.

IBM Fellow, Vice President of IBM Europe & Africa and Director of the IBM Research Lab in Zurich, Switzerland

Kathryn Guarini

COO, IBM Research, Vice President, IBM Impact Science

More Science stories

IBM teams with industry partners to bring energy-efficient AI hardware to hybrid cloud environments

AI’s unprecedented demand for data, power and system resources poses the greatest challenge to realizing this optimistic vision of the future. To meet that demand, we’re developing a new class of inherently energy-efficient AI hardware accelerators that will increase compute power by orders of magnitude, in hybrid cloud environments, without the demand for increased energy.

Continue reading

Advancing the Potential of AI in Medical Imaging at MICCAI 2020

I believe one of the most promising areas for AI to make an impact is in the field of medical imaging. Through advancements in AI that allow for more intelligent and accurate analysis of video and still images, there is hope that clinicians will soon be able to widely augment the data and information they […]

Continue reading

State-of-the-Art Results in Conversational Telephony Speech Recognition with a Single-Headed Attention-Based Sequence-to-Sequence Model

Powerful neural networks have enabled the use of “end-to-end” speech recognition models that directly map a sequence of acoustic features to a sequence of words. It is generally believed that direct sequence-to-sequence speech recognition models are competitive with traditional hybrid models only when a large amount of training data is used. However, in our recent […]

Continue reading