Are we harming the world in our pursuit to understand it better?
Around 7 pm on Mondays, Maya can be found in her natural habitat, carefully seeding a culture to be infected the next day or inoculating a batch of bacteria to be genetically transformed. This is where she is on most days, in a brightly-lit biology lab, surrounded by huge freezers, fume hoods, and biosafety hoods. In the background, you can sometimes hear the light buzz of the centrifuge spinning in a corner or spot the oven heating up agarose to help visualise some DNA. As an individual in this research ecosystem, her work has the potential to lead to findings that deepen humanity’s understanding of the world and perhaps better our lives.
But aside from this important work, another reality lingers – a quieter, more insidious cost hidden in plain sight. Each pipette tip Maya discards, each sample tube she pushes into the crammed freezers, and each plastic petri dish she tosses into the waste bin, represent a trail of an environmental impact that her work leaves behind.
“The proper use of science is not to conquer nature but to live in it”
– Barry Commoner, American cellular biologist, professor and ecologist
Maya is not alone. According to the Office of Laboratory Safety and Environmental Health (OLSEH), IISc, research activities at the Institute generate 70,000 kg of biowaste annually, which includes 50,000 kg of animal waste. In the last year alone, IISc has generated over 3,000 kg of waste categorised as toxic. A 2014 Nature article estimated that the total amount of lab plastic waste generated globally was 5.5 million tonnes.
There is more. According to the National Renewable Energy Laboratory in the US, labs consume at least five to ten times more energy than an office building of equivalent size, and this value can increase by 100 times if there are clean rooms and high-process operations. A major energy guzzler is fume hoods, enclosures that safely contain and ventilate hazardous fumes, vapours, gases and dust generated during chemical processes. Due to safety considerations, fume hoods usually run non-stop in most labs. When open, a single fume hood can consume as much energy as 3.5 homes in 24 hours, and very often, researchers are not aware of how energy-intensive the high-tech equipment they use daily is.
The energy consumption is also stark in fields like computer science and those that involve simulation-based experiments. While the Supercomputer Education and Research Centre (SERC) at IISc has enabled complex studies such as modelling for climate predictions, drug discovery, and aerospace engineering, its energy requirement is vast. Sathish Vadhiyar, Chair of SERC, shares: “The one [supercomputer] in IISc consumes about 500-550 kilowatts. It is at least, on average, 100-300 times slower than the fastest supercomputing machine in the world. If you scale the speed, the power consumption can run up to the rate of 25-30 megawatts. It is believed that this kind of power can provide electricity to an entire village.”
These systems run 24/7. Data centres employ transformers, diesel generators, and UPS devices. They also require elaborate cooling infrastructure as these machines generate a lot of heat. Most AI models work on devices called accelerators which can handle AI workloads with greater speed, efficiency, and cost-effectiveness compared to generic computing hardware. Sumit Mandal, Assistant Professor in the Department of Computation Science and Automation, IISc, reveals: “The daily carbon footprint of ChatGPT is 24 kg [of carbon], which is equivalent to cutting down 406 trees. At IISc, we have about 22,000 trees, and that means it will take less than two months to destroy them if we run ChatGPT. Of course, they’re not [actually] cutting down the trees, but those are the statistics.”
Big tech companies like Amazon, Microsoft, and Google are now looking toward small-scale nuclear power plants to meet the rising energy needs of their data centres and AI initiatives. Microsoft even plans to reopen the Three Mile Island power plant, despite it being the location of the worst commercial nuclear accident in US history in 1979. The potential for nuclear waste generation, or worse, disasters, is a major concern.
“The greatest threat to our planet is the belief that someone else will save it”
– Robert Swan, British explorer and global environmental activist
Researchers often get caught up in trying to find solutions to the world’s problems, including global warming and climate change. However not many are aware or are actively thinking of ways to make their research more sustainable. It begins by realising that this is an issue, and taking stock of the environmental impact.
Minimising waste generation is the first step. Dipshikha Chakravortty, Professor at the Department of Microbiology and Cell Biology, emphasises that every student must “account for each and every piece of material that they are using – every microlitre of enzymes and chemicals.”
One can also opt for reusable alternatives to single-use items and design experiments to obtain data robustly and reliably such that experiments aren’t repeated unnecessarily. Unavoidable waste produced should be handled in a way that reduces its harmful effects. At IISc, OLSEH collects and segregates waste from the labs into categories such as solid, solvent, biomedical, radioactive, oil, and so on. Waste materials are then handed over to contract-based vendors, who dispose of them based on Karnataka State Pollution Control Board regulations and the Atomic Energy Regulatory Board guidelines (for radioactive waste).
As individuals working in the lab, we can also personally take small steps like switching off appliances when not in use. During a recent “Shut the Sash” contest at the University of Virginia, the Office for Sustainability team reported saving USD 34,000 in energy costs in one month thanks to the efforts of 21 laboratories closing their fume hood sashes when not in use. They concluded that “maintaining these behaviours could result in energy savings of over USD 400,000 annually.” More researchers have begun to realise that as their labs get greener, it can relax the strain on their long-term budget significantly.
In data centres, innovative approaches such as using ambient cooling can substantially reduce energy requirements. Ambient cooling systems bring in cold air from outside and drive out hot air from the data centres using heat exchange equipment. This can significantly reduce energy consumption in data centres, when compared to using heavy-duty chillers that consume high energy.
There are other smart solutions. “In certain parts of a [computer] program, you don’t need to use very high precision numbers and operations, which will consume storage, generate heat, and consume a lot of power as well. One can be smart while developing such programs by switching to low precision,” Sathish suggests. Without compromising on accuracy, machine learning programs can also be made less energy-intensive through techniques such as pruning. Pruning, like in a tree, cuts down or avoids the computation of certain parts of an algorithm that do not affect the results. Sumit’s group explores the use of “in-memory computing” to make running programs efficient in speed as well as energy. “With this technology (in-memory computing), we don’t need a separate memory and separate computer system like in a GPU (Graphics Processing Unit). If we have a separate memory, then every time memory access happens, it takes a lot of time. It also consumes a lot of energy,” he explains.
“Scientists [should] ensure we are not among the last to jump on the sustainability bandwagon”
– Gaia Bistulfi, scientist and author
One of the root causes of increased environmental impact of research is the culture of “fast science” – the relentless push to publish quickly, often at the expense of rigorous methodology and reproducibility. Impact factors drive competition, encouraging more experiments, faster results, and a mountain of waste in the process. This is especially pertinent to well-funded labs at universities. “Because we have a lot of free resources here at IISc, we don’t think that this kind of thing may impact others, if not us, right?” Sumit points out. “Everybody should be aware of this kind of negative influence.”
A 2014 study estimated that the number of scientific publications doubles every nine years. What does this say about the growing negative impact of research on the environment?
As more and more scientists across the globe recognise this issue, there have been efforts at both university and national levels. In 2005, Harvard University became one of the first educational institutions to commit to what was then called the Green Labs Program. It aims to be fossil fuel-free by 2050 and fossil fuel neutral by 2026.
Climate@MaxPerutzLabs is an Austrian grassroots organisation where employees aim to make research at the institute more climate-friendly. They advocate the incorporation of sustainable practices as an integral feature of research, just like ethical and safety standards. The US-based My Green Lab is a non-profit that established the first-ever sustainability criteria for laboratory operations and products in 2013, which have been adopted by a thousand labs across the world. They survey labs and provide recommendations on equipment use, lab practices related to purchasing, chemicals and reagents, and waste disposal.
Back at IISc, Maya has switched to reusable autoclavable plates from the use-and-throw ones, to culture her bacterial colonies, and she operates the autoclave only at full capacity. Baby steps sure, but important nevertheless.
