How artificial intelligence and advanced computing can pull us back from the brink of accelerating climate change

Barely a week goes by without another dramatic report on humanity and the planet reaching a climate change tipping point. The latest reports have been a breathtaking analysis by the World Meteorological Organization and criticisms by the Secretary-General of the United Nations. Both were shared in the last days of April.

Artificial Intelligence will determine whether we pass the tipping point or back off the brink.

Artificial intelligence is one of the significant tools left in the fight against climate change. Artificial intelligence has turned to risk prediction, the prevention of damaging weather events, such as wildfires, and carbon offsets. It has been described as critical to ensuring that companies meet their ESG goals.

However, it is also an accelerant. AI requires enormous computing power, which consumes energy when designing algorithms and training models. And just like software ate the world, artificial intelligence is bound to follow.

By 2030, artificial intelligence will contribute a whopping $15.7 trillion to the global economy, more than the GDP of Japan, Germany, India and the UK. There are many people using AI as ubiquitously as the internet, from using ChatGPT to build email and write code to using text-in-picture platforms to make art.

The power used by AI has been on the rise for years now. For example, the power required to train the largest AI models doubled roughly every 3.4 months, increasing 300,000-fold between 2012 and 2018.

This expansion offers opportunities to solve major real-world problems in everything from security and medicine to hunger and agriculture. It will also have a punishing impact on climate change.

The cost of high energy

Computing goes hand in hand with high energy costs and an increased carbon footprint, putting the accelerator on global climate change.

This is especially true for AI. The sheer number of GPUs running machine learning algorithms get hot and need to be cooled down; otherwise they melt. Training even one Large Language Model (LLM) requires an incredible amount of energy with a large carbon footprint.

For example:

As we enter the GPT4 era and the models get bigger, the energy required to train them increases. GPT-3 was 100 times larger than its predecessor GPT and GPT-4 was ten times larger than GPT-3. Meanwhile, larger models are released faster. GPT-4 arrived in March 2023, nearly four months after ChatGPT (based on GPT-3.5) was released in late November 2022.

To balance, we shouldn’t assume that as new models and companies emerge in the space, the carbon footprint of AI will continue to grow. Geeta Chauhan, AI engineer at Meta, uses open source software to reduce the operational carbon footprint of LLMs. Her latest work shows a 24-fold reduction in carbon emissions compared to GPT-3.

However, the popularity of AI and its exponential power undermine much of the climate action in effect today and question its potential to be part of the solution.

We need a solution that allows AI to thrive while reducing its carbon footprint. So what do we do?

Moderate the dependence on carbon

As always, technology will drag us out of this situation.

For the AI ​​explosion to be sustainable, advanced computing must come to the fore and do the heavy lifting for many tasks AI currently performs. The good news is that we already have advanced computing technologies that are ready to perform these tasks more efficiently and faster than AI, with the added benefit of using much, much less energy.

In short, advanced computing is the most effective tool we have for mitigating the carbon dependency of AI. With it, we can slow down the creep of climate change.

Several technologies are emerging in advanced computing that can solve some of the problems AI is currently facing.

For example, quantum computing is superior to artificial intelligence in drug discovery. As humans live longer, they encounter new, complex and incurable diseases in increasing numbers. This is called the Beatles’ best problem, in which new drugs have modest improvements on already successful therapies.

Until now, drug development has focused on rare events within a data set and making educated guesses to design the right drugs to target and bind to disease-causing proteins. LLMs can be used efficiently to help with this task.

LLMs are remarkably good at predicting which words in our vocabulary may best fit in a sentence to accurately convey meaning. Drug discovery is not very dissimilar in that the problem is identifying the best match, or configuration, of molecules in a compound to achieve a therapeutic result.

However, molecules are quantum elements, so quantum computing is much more effective at tackling this problem. Quantum computing has the ability to rapidly simulate large numbers of binding sites in drugs to create the right setup for treating currently incurable diseases.

Advanced Calculus: Quantum and Beyond

Quantum’s capabilities mean these can be solved much faster and with much less power consumption.

Another development with a real potential to improve artificial intelligence is photonics, or so-called optical computing, which uses laser light instead of electricity to send information.

Some companies are building computers using this technology, which is much more energy efficient than most other computer technologies and is increasingly being recognized as a path to achieving Net Zero.

Elsewhere, we have neuromorphic computers. This is a type of computer engineering in which the elements of the computer system are modeled on those of the human brain and nervous system. They perform calculations to replicate the analog nature of our neural system. Evidence of this technology includes designs from Mythic and Semron. Neuromorphic is another greener option that requires additional investment. Its hardware has the potential to run large deep learning networks that are more energy efficient than comparable classical computer systems.

For example, information processing through its one hundred billion neurons consumes just 20 watts, similar to an energy-saving light bulb in a house.

The development and application of these innovations are essential if we are to curb climate change.

Advanced IT leaders

There are many startups (and investors) around the world obsessed with advanced computing, but there are only a handful of companies that are focusing on so-called impact areas like healthcare, the environment, and climate change.

Within quantum computing, the most exciting companies developing use cases for energy and drug discovery are Pasqal (its co-founder was awarded the 2022 Nobel Prize in Physics), Qubit Pharmaceutical, and IBM. When it comes to photonics, we see leaders with global impact like Lightmatter and Luminous, while in neuromorphic computing, we’re tracking the progress of Groq, Semron, and Intel.

Advanced computing is vital to achieving the energy efficiency we need to fight climate change. It simply takes too much time and energy to run artificial neural networks on a GPU.

By adopting advanced computing methods as alternatives to AI, companies can greatly alleviate the impact AI has on the environment, while ensuring that its vast power can mitigate some of the impacts of climate change, such as anticipating wildfires or conditions extreme weather.

The existential endpoint is approaching for our environment. But the situation is not hopeless.

The implementation of advanced computing is a credible and powerful resource to counter the problem. We must invest in these technologies now to solve the greatest challenge facing humanity.

Francis Ricciuti is a VC at Runa Capital.