A key topic of conversation lately is the significant consumption of energy used by AI systems and how it has become a significant emitter of CO2. Where a regular Google search uses around 0.3 watt-hours, a single ChatGPT query uses 2.9 watt-hours and training AI models uses about the same amount of power in a year as 130 homes. In short, AI uses a lot of power. And it’s only going to increase. In fact, according to Morgan Stanley, power demand from generative AI is expected to grow by an average of 70% annually by 2027, primarily driven by the expansion of data centres. This is as much power as Spain used in 2022!
Furthermore, Researchers at the University of Massachusetts Amherst recently looked at various AI models to estimate the energy cost required to train them and published their staggering findings. In short, the carbon footprint of training a single large natural language processing (NLP) model is over 272,000 kg of CO2 emissions, which is the equivalent of 125 round-trip flights between New York and Beijing.
But it’s not just our carbon footprint we need to be aware of when it comes to AI’s impact on the environment. According to the Center for Secure Water (C4SW), the rapid growth of AI is having a significant and serious impact on the world’s water supply. “The data centers required to run large AI models consume vast amounts of power. This significant energy consumption generates large amounts of heat, which in turn requires cooling systems to prevent server overheating. Most common cooling methods (e.g., cooling towers) depend on substantial quantities of clean, fresh water.” There are an estimated 11,000 data centres in the world, each using anywhere between 68,000 and 2.1 million litres of water per day. A large datacentre in the US uses about the same amount of water per day as 4,200 people. And to put this into context, of all the water on earth, only 3% is fresh and 2.5% is locked away in glaciers and polar ice caps, leaving only 0.5% of Earth’s water available to meet our needs.
AI integration into payments means a significant energy spike
Until recently, payments had not been a drain on energy resources. Traditional card-present terminals are simple to operate and primarily perform basic functions such as reading card data, verifying transactions, and printing receipts, all of which have very low energy demands.
But… the acceleration of online purchases, increasing levels of online payment fraud and the subsequent adoption of fraud-mitigating AI systems means payments are now adding to the drain on energy resources. This is largely due to the use of AI.
The AI-driven systems used by online merchants for fraud detection operate in a continuous analytical cadence, looking at transaction data using complex machine learning models. This high-frequency, high-volume data processing requires powerful servers and extensive computational resources. And data centres optimised for AI workloads consume vast amounts of electricity – a recent article reported that AI data centres could be consuming 25% of electricity in the US by 2030.
The proliferation of AI in payments
With the rate of payments fraud increasing year-on-year, the average online retailer now uses up to five AI systems to analyse a range of data to assess the probability of each transaction being fraudulent. This requires a high degree of accuracy, which means more sophisticated algorithms, extensive testing, and constant evolution and fine-tuning of AI models. And all this demands additional computational power – and therefore energy.
This additional strain on energy isn’t going away anytime soon, which is why there is study upon study that talk about ways organisations can reduce the amount of energy AI is consuming. Now, for most industries, this is extremely relevant, and important. Because AI has become integral to how they operate and is here to stay. Indeed, Google has just signed the world’s first corporate agreement to purchase nuclear energy from multiple small reactors in order to support AI technologies with clean energy sources whilst reliably meeting electricity demands with carbon-free energy, around the clock.
Now, for payments, there is actually a far simpler, much more energy efficient option – simply, don’t use AI. You’re probably thinking to yourself that this is crazy given the statement I just made about how integral and useful AI has become. But for payments fraud, there actually is technology that solves the problem at its root – it’s called CPoI.
Solving fraud
I have written a number of articles about how AI is costing merchants more than fraud, and the fact that AI is responsible for false positives, which is a massive issue that merchants may not actually be aware of. So, if you’ve already read these articles, feel free to skip ahead to the next paragraph. If you haven’t, here’s a quick summary:
While AI enhances fraud detection, it also increases the risk of false positives, which is when legitimate transactions are incorrectly flagged as fraudulent, and therefore blocked. The pressing matter here, is that there’s data from the global payments experts such as Mastercard that suggests the AI is actually blocking a significant amount of legitimate transactions by incorrectly identifying them as fraudulent. And it’s costing merchants $443 billion per year.
Burbank’s CPoI, or Card Present over Internet, is a groundbreaking technology that turns consumer devices into a one-time payment terminal. During the checkout process the customer simply taps their card against their own device and securely enters their PIN, thus creating a card present transaction over the internet. This offers a range of benefits including significantly less fraud and therefore liability shift (away from the merchant), lower processing fees, and a much faster, easier customer experience. But the best part about CPoI is that there are zero false positives.
So, if we remove the opportunity for fraud with online payments by changing the majority of them to card-present, then we reduce the need for fraud detecting AI systems, we reduce the impact online payments have on the environment, and we’re doing our part towards helping the world achieve Net Zero. A pretty big win all around if you ask me.
So, if you want to know more about CPoI, reach out to the Burbank team today for a chat and a demo.
댓글